MADIA DISCUSSION PAPER 7 8297 MANAGING AGRICULTURAL DEVELOPMENT IN - AFRICA _ E~~M _ Fmo _~ FOREWORD The MADIA study and the papers comprising this MADIA Discussion Paper Series are important both for their content and the process of diagnosis and analysis that was used in the conduct of the study. The MADIA research project has been consultative, nonideological, and based on the collection and analysis of a substantial amount of concrete information on specific topics to draw policy lessons; it represents a unique blend of country-oriented analysis with a cross-country perspective. The conclusions of the studies emphasize the fundamental importance of a sound macroeconomic environment for ensuring the broad-based development of agriculture, and at the same time stress the need for achieving several difficult balances: among macroeconomic, sectoral, and location-specific factors that determine the growth of agricultural output; between the development of food and export crops; and between the immediate impact and long-run development of human and institutional capital. The papers also highlight the complementarity of and the need to maintain a -oalance between the private and public sectors; and further the need to recognize that both price and nonprice incentives are critical to achieving sustainable growth in output. The findings of the MADIA study presented in the papers were discussed at a symposium of senior African and donor policymakers and analysts funded by USAID in lune 1989 at Annapolis, Maryland. The participants recommended that donors and African governments should move expeditiously to implement many of the study's valuable lessons. The symposium also concluded that the process used in carrying out the MADIA study must continue if a stronger, more effective consensus among donors and governments is to be achieved on the ways to proceed in resuming broad-based growth in African agriculture. The World Bank is committed to assisting African countries in developing long-term strategies of agricultural development and in translating the MADIA findings into the Bank's operational programs. Stanley Fischer Edward V K. Jaycox Vice President Development Economics Vice President and Chief Economist Africa Regional Office MADIA DISCUSSION PAPER 7 COTTON IN AFRICA AN ANALYSIS OF DIFFERENCES IN PERFORMANCE UMA LELE NICOLAS VAN DE WALLE MATHURIN GBETIBOUO THE WORLD BANK _ WASHINGTON, D.C. _II Copyright © 1989 All rights reserved The International Bank for Reconstruction Manufactured in the United States of America and Development/THE WORLD BANK First printing November 1989 1818 H Street. N.W. Washington. D.C. 20433, U.S.A. MADIA Discussion Papers are circulated to encourage discussion and ment. at the address shown in the copyright notice above. The World Bank comment and to communicate the results of the Bank's work quickly to the encourages dissemination of its work and will normally give permission development community: citation and the use of these papers should take promptly and, when the reproduction is for noncommercial purposes. with- account of their provisional character. Because of the informality and to out asking a fee. Permission to photocopy portions for classroom use is not present the results of research with the least possible delay. the manuscript required. though notification of such use having been made will be has not been prepared in accordance with the procedures appropriate to appreciated. formal printed texts, and the World Bank accepts no responsibility for The complete backlist of publications from the World Bank is shown in errors. The findings, interpretations, and conclusions expressed in this the annual index of Publications, which contains an alphabetical title list and paper are entirely those of the authoris) and should not be attributed in any indexes of subjects, authors, and countries and regions. The latest edition is manner to the World Bank. to its affiliated organizations. or to members of available free of charge from the Publications Sales Unit. Department F. The its Board of Executive Directors or the countries they represent. World Bank, 18i8 H Street, N.W. Washington. D.C. 20433, U.S.A., or from The material in this publication is copyrighted. Requests for permission Publications. The World Bank, 66. avenue d'lena. 75116 Paris, France. to reproduce portions of it should be sent to Director, Publications Depart- Uma Lele is the manager of Agricultural Policy in the Africa Technical Department at the World Bank. Nicolas van de Walle is an assistant professor at Michigan State University. Mathurin Gbetibouo is an economist of Agricultural Policy in the Africa Technical Department at the World Bank. Library of Congress Cataloging-in-Publication Data Lele, Uma J. Cotton in Africa: an analysis of differences in performance. / Uma Lele, Nicolas van de Walle, and Mathurin Gbetibouo. (MADIA discussion paper; 7) Includes bibliographical references. 1. Cotton trade-Africa, Sub-Saharan. 2. Managing Agricultural Development in Africa (Organization) I. Van de Walle, Nicolas, 1957- . 11. Gbetibouo, Matiurin, 1949- . III. Title. IV. Series. HD9087.S78L44 1989 338.1'7351'0967-dc2O 89-22724 ISBN 0-8213-1323-1 Contents Summary and Policy Implications ..................................... 5 Introduction ........................................... 7 Cotton Performance in MADIA Countries .............................. 9 Cameroon ........................................... 11 Senegal ............. .............................. II Tanzania ........................................... 13 Kenya ........................................... 14 Other MADIA Countries ...................................... .... 14 Explaining Differences in Performance-Price Factors ................. 15 The Impact of Price Factors ........................................ 15 Farm-level analysis ........................................... 18 Comparative advantage ......................................... 20 Measuring Price and Nonprice Factors .............................. 20 Explaining Variations in Performance-Nonprice Factors .............. 22 Agro-Ecological Factors ........................................... 22 Cotton as an Instrument of Regional/Political Equity ...... ............ 23 Institutional Factors ........................................... 23 Input and output price supports ................................. 23 Research and extension ........................................ 27 Processing issues and backward coordination ...................... 28 Weighing the Evidence ........................................... 29 Recommended Institutional Reforms ............. ................... 30 Francophone Countries ........................................... 30 Anglophone Countries ........................................... 31 Annex I ........................................... 32 Annex 2 ........................................... 33 Notes ........................................... 34 Bibliography ........................................... 35 Acknowledgement An earlier draft of the paper received valuable comments from the following reviewers: lames W Adams, Manmohan Agarwal, Mohsin Alikhan, Charles Ameur, Francois Bocchino, Bela Belassa, Daniel Benor, Peter Brumby, Stephen Carr, Ajay Chhibber, Kevin Cleaver, Michael Collinson, Vincent Cusumano, Ronald C. Duncan, Carl Eicher, Peter H. Freeman, Alan Gelb, Arntraud Hartmann, Peter Hazell, William Jaeger, David Jones, Kathryn McPhail, John W Mellor, Minhchau Nguyen, F. Stephen O'Brien, Jose Oliveras, Michel J. Petit, Christian Polti, Thomas Reardon, loanne Salop, Rick Scobey, Anand Seth, Alain Seznec, John Staatz, David R. Steeds, Gert Stern, Roger Sullivan, Elton Thigpen, J.A. Nicholas Wallis, Harry Walters, and Michael T. Weber. Their contribution is greatly appreciated. While we have attempted to reflect their concerns, given the diversity of views on many of the issues addressed in this paper, it has not been possible to reconcile the different perceptions. The views expressed and any errors and omissions remain solely the responsibility of the authors. The authors are also grateful to Natasha Mukherjee for research assistance, and to Pat Blair and Suzanne Wood for editing. 3 Illustrations Acronyms Tables ADMARC Agricultural Development and Marketing I. Africa cotton: growth in production, domestic Corporation availability and trade volumes ..................... 7 BCGA British Cotton Growing Association 2. Seed cotton production in the MADIA countries, 1960/ CAP Cotton and Agricultural Processors Ltd. 61-1987/88 ...... 9 CCCE Caisse Centrale de Cooperation 3. Characteristics of cotton farming systems in the Economique MADIA countries ............................... 11 CCM Chama Cha Mapinduzi 4. CFDT presence in local cotton parastatals .......... 12 CFDT Compagnie Francaise pour le 5. Areas under cotton cultivation in the MADIA countries, Developpement des Fibres Textiles 1960/61-1986/87 ........... 13 CGIAR Consultative Group on International 6. Seed cotton yields in the MADIA countries, Agricultural Research 1960/61-1986/87 ........... 14 CICAM Cotoniere Industrielle du Cameroun 7. Ratio of producer prices for cotton lint at official and CLSMB Cotton Lint and Seed Marketing Board purchasing power parity exchange rates to world COTMAN Cotton Marketers' Association of Nigeria cotton lint prices, 1960/61-1986/87 ............... 15 CRC Cotton Research Corporation 8. Nigeria: Import parity price for cotton, 1986 ......... 16 IAR Institute of Agricultural Research 9. Producer prices for seed cotton, deflated by CPI ICAC International Cotton Advisory Committee 1980=100 ............ ............. 17 IRCT Institut de Recherches du Coton et des 10. Ratio of seed cotton to maize producer prices in the Textiles Exotiques MADIA countries, 1964/65-1987/88 ............... 18 IRA Institut de la Recherche Agronomique 11. Returns to labor from cotton and alternate agricultural KTDA Kenya Tea Development Association and nonagricultural employment in the MADIA LBA Licensed Buying Agents countries .... 19 LSMB Lint and Seed Marketing Board 12. Agro-ecological conditions in cotton cultivating regions MOA Ministry of Agriculture of the MADIA countries ............... 22 NCB National Cotton Board 13. Organization of the cotton sectors in the MADIA NCCCP National Consultative Committee on countries .... 24 Cotton Production 14. Numbers of African and expatriate cotton researchers SODECOTON Societe de Developpement du Coton du in anglophone and francophone Africa, 1984 ......... 28 Cameroun Annex Table I SODEFITEX Societe de Developpement des Fibres Seed Cotton Price Elasticities in the MADIA Textiles Countries ..... 33 SONACOS Societe de Developpement des Figures Oleagineux du Senegal I. Share of Sub-Saharan Africa in world agricultural SOTEXKA Socikta des Textiles de Kaolack exports, 1961-1987 ........... 7 TARO Tanzania Agricultural Research 2. Seed cotton production in francophone and Organzani anglophone Sub-Saharan Africa, (excluding Zimbabwe), TCA Organization 1961-1987 .... 8 TCMB Tanzania Cotton Marketing Board 3. Seed cotton yields in the MADIA countries, TEXCO Tanzania Textile Corporation 1970/71-1983/84 ............ .................. 10 TFC Tanzania Fertilizer Corporation 4. CFDT presence and seed cotton yields in selected francophone Sub-Saharan African countries ......... 10 5. Producer prices for seed cotton converted at purchasing power parity exchange rates in the MADIA countries, 1970/71-1986/87 ....... ............... 16 6. Production and real price indices for seed cotton in the MADIA countries, 1970-87 ........................ 21 4 Summary and Policy Implications The development of African economies depends to a great induced approach has come under heavy criticism for its extent on the performance of their agricultural sector, high costs, monopoly approach, and excessive focus on within which export crops play an important role. However, cotton. In anglophone countries, where population pressure Africa's share of world trade in traditional export crops has on the land has reduced fallow periods-leading to a declined steadily during the past two decades. Indeed, decline in soil fertility and yields while at the same time increased internal demand due to population growth is increasing the demand for food-a low input/low yield leading to higher imports of many commodities that were extensive approach to cotton production has made compet- once exported. An investigation of the factors affecting ing food crops increasingly attractive, and is threatening the various commodities across Sub-Saharan Africa can throw future of cotton cultivation-except as a risk minimization light on how best to improve internal production and foster strategy. What lessons can be drawn for the future of the both intra-African and international trade. cotton industry from these two sets of experiences? This paper focuses on cotton, a key export crop, which is The central conclusion of this paper is that, while grown in 30 out of the 44 African countries. The world differences in macroeconomic and sectoral pricing polities demand for cotton grew at 1.2 percent per annum from 1961 appear to have been critical, institutional factors have been to 1986, less rapidly than cocoa (2.1 percent) and tea (3.3 fundamental in explaining the sustained growth of cotton percent); but Africa's demand for cotton grew faster than for production in francophone countries. The CFDT approach most other commodities that it exports-with the exception has led to the development and extension of technology of palm oil. In the MADIA countries with which the paper is and assured the availability of inputs, marketing and concerned (Cameroon, Senegal, Nigeria, Kenya, Tanzania, processing facilities. In particular, it has ensured adequate and Malawi), output performance has been distinctly mixed. financing of the cotton sector, making possible, among By and large, cotton production in the francophone coun- other things, timely payments to farmers. Its high cost of tries has been superior to that of anglophone countries assistance could be brought down if the current diminution (except for Zimbabwe) since the early 1970s, even though in numbers of resident expatriate CFDT experts, as many of the latter had excelled earlier. This paper asks why. observed in Mali and C6te d'lvoire (see Table 4) continues A distinguishing feature of cotton production in the without a loss in cotton productivity. francophone countries is the structure of their domestic Although the impact of price incentives on cotton cotton industry. These countries exhibit a high degree of production is important for determining short-run shifts coordination between the upstream and downstream actors between food and export crops or for allocating labor in the cotton production apparatus. The presence of the between cotton and other crops (including off-farm wage Compagnie Fran(aise pour le Developpement des Fibres Textiles employment), this paper demonstrates, first, that a variety (CFDT) has provided them with more effective vertical of other incentives have been important, including CFDT's integration than is found in anglophone Africa, although own incentive to make profits, related directly to the both groups have a wide range of arrangements as to the volume of exports, and second, that prices must not be extent of integration/decentralization. Empirically, it is considered in isolation from a range of nonprice factors, difficult to separate the effects of the CFDT presence per se including the nature of the colonial legacy. The two sets of from those of integration, since CFDT has brought (in factors jointly explain the much greater relative success of addition to integration) professionalism, know-how and the cotton subsector in francophone Africa. Thus, the Societe detailed knowledge of both the country-specific circum- de Developpement des Fibres Textiles, SODEFITEX, in Senegal, stances and the international market and finance. and, the Socidte de Developpement du Coton du Cameroun, A closely associated difference between the two sets of SODECOTON, in Cameroon, have been relatively effective countries is the role played by institutional factors in agents of government policy. In the anglophone countries, alleviating physical constraints and ensuring effective price faulty mechanisms and procedures for paying producer incentives. Anglophone countries are characterized by a low prices and ensuring input supplies have seriously under- input/low yield technology whereas the francophone coun- mined their potential impact, and the cotton subsectors lag tries, induced by CFDT, feature a high input/high yield behind in terms of a wide variety of indicators. The technology. Despite its success in increasing production difference is particularly striking in Kenya, which has and yields per hectare in francophone countries, the CFDT otherwise made much more impressive strides in several 5 other crops, relative to most of the francophone producers. highlights why its technical assistance has been more This suggests the need to understand the commodity- effective than most, including other French assistance in specific challenges posed by each crop, the approaches francophone Africa. uniquely suited to the requirements of that crop, and the As for the recipient countries, politics has played a heavy particular constraints that may be difficult to alleviate. role in the support or demise of the cotton sectors in the Donors of MADIA countries need to be aware of the MADIA countries. Depending on the political strength of tremendous institutional weaknesses of many Sub-Saharan the cotton-producing populations, which has itself African countries, which makes it essential to consider depended on (but by no means assured) the economic carefully the choice of institutions and factors affecting the importance of the domestic cotton industry, governments long-term development of export crops such as cotton. have been willing or reluctant to let producer organizations They need to: (i) devote greater attention to the role of effectively represent the interests of the cotton producers agricultural research and its strong link to effective agricul- in the operation of the industry. Given the many complex- tural extension, input, credit, and marketing systems; (ii) pay ities involved in developing the cotton sector, governments greater attention to the financing of institutions; (iii) place need to (i) show a greater awareness of the professionalism greater emphasis on the quality, critical mass, and length of needed to develop crops, (ii) provide better incentives for time for which technical assistance is provided, with cotton-industry managers to respond to the needs of their particular emphasis on creating indigenous institutional producers, and (iii) address the technological, financial, and human capacity and an incentive structure conducive to international market, and ecological complexities involved good economic performance; and (iv) develop regional and in developing the cotton sector. The emphasis in the past international cotton research and marketing strategies for has been placed far too narrowly on producer price and within African countries (including, perhaps, establish- incentives. This paper argues that, in francophone Africa, ing a CGIAR-type presence), in view of the atrophying the chances of developing a viable indigenous cotton expertise in cotton of the traditional colonial donors and industry are likely to be greater if donors promote a the changing pattern of world market demand. CFDT has broader set of incentives and a long-term approach to been able to provide a regional approach to cotton agricultural research, training, and technical assistance, and research which transcends national boundaries-and which especially to the link between research and the demands covers many areas of the same ecological zones. The of the world market, rather than the short-term approaches declining role of CFDT (as a regional research organization) (based on parcelling out different parts of the industry) and its parcelling out to national research systems of currently being advocated. Some of the essential elements Ministries of Agriculture could impact adversely on cotton of the CFDT approach (which are also present in Zim- research; and unless a research organization of the CGIAR babwe's successful cotton sector) also need to be consid- type, but with strong linkages to producing households, ered for adaptation to cotton industries elsewhere. These supports cotton research, production and productivity of include (i) the need for credit to farmers (ii) the timely cotton in the small African countries could suffer the same payment of producer prices, and (iii) an excellent research fate as that of other export crops for which Africa has lost and input delivery system closely tied to the nature of the her share of the world market. Recognition of CFDT's internal or external market and based on a thorough accomplishments in addressing the above issues, and thus understanding of the location-specific circumstances affect- in successfully alleviating the technological and institutional ing cotton producers. constraints to cotton development in francophone Africa, 6 Introduction In general, the performance of the agricultural sector in Figure 1 sub-Saharan Africa since independence has been poor. Share of Sub-Saharan Africa in world agricultural exports, Export crops, in particular, have suffered. As can be seen in 1961-87 Figure I, the region's share of world agricultural exports fell Percent precipitously throughout the 1970s and 1980s. 21 The reasons for this poor performance are poorly under- stood and the subject of much debate. Some have empha- 20 \ sized the inadequacy of producer price incentives because of high taxation of the agricultural sector. (World Bank 1981, 10 1984; Eicher 1982) Adverse macroeconomic policies are said ,\ to have an even more significant effect on effective taxation rates in the agricultural sector than do sector-specific 17 - policies. (Krueger, Schiff and Valdes 1988) This school of , \ thought stresses the need to "get the prices right." Others, in contrast, have sought explanations in a host of non-price 15 \ factors, (Delgado and Mellor 1984; Lipton 1987; Ray 1988) 1, - including political (Bates 1981), technological, institutional, infrastructural, and human-capital constraints. (Lele 1988a; 13 -U Lele 1989). Some recent theoretical literature has argued 12 that favorable price incentives themselves promote capital formation, technological progress, and institutional innova- 11 tion. (Hayami and Ruttan 1985; Mundlak 1988). Relatively 19B1 1963 1965 1967 1690 1071 1973 1975 1977 1979 1901 1983 19S5 1S7 little empirical work exists, however, to test the relative importance of, or the nature of interaction between, price Source: World Bank, BESD Databank. incentives and nonprice factors. Table 1 This paper attempts to throw light on the debate by Africa cotton: Growth in production, domestic availability, reviewing the development of one commodity, cotton, and trade volumes (percent per annum) across selected African countries and by attempting to pinpoint the causes of relative success and failure of 196186 1970-86 1987-2000* different commodity development schemes. Cotton has Production 1.0 -0.1 0.9 been chosen in part because policy conclusions, ceteris Domestic paribus, could apply to a broad range of countries. At Availability 2.0 -0.9 0.9 present, Sub-Saharan Africa accounts for 4.5 percent of Gross Exports -0.6 -1.5 -0.2 world cotton production and about 10 percent of world Gross Imports 5.4 4.7 2.4 exports. The annual market value of African lint exports has Memo Items: averaged over US$1 billion in recent years. The crop is Cotton, World 1.2 1.2 0.7 grown in more than 30 African countries, under diverse Palm Oil, Africa 18.1 22.3 7.9 climatic conditions and in varying soil fertilities. Some 4 *Forecast million African households derive cash incomes from Source: The World Bank, "Price Prospects for Major Primary planting, handling, or processing it. Commodities, 1988-2000," volume 2, Washington, D.C., November With declining soil productivity and rising profitability of 1988. food crops, the future of cotton production in Africa has come into question. Its adverse environmental impact has often been criticized. Furthermore, international cotton prices fluctuate widely (the coefficient of variation for the factors that surround the cultivation of cotton, and (ii} the international price of cotton being 42 percent) and have institutional environment, including, in particular, the been badly depressed since 1986. Nevertheless, in many political support and technocratic content of institutions, parts of Africa, where long-term options are few, cotton their adequate funding and trained personnel, and the production may be an efficient means of increasing employ- existence of technological packages and systems for exten- ment and incomes. Domestic demand is currently growing sion. In demonstrating the fundamental importance of the faster than production, and gross cotton imports are institutional environment in determining performance, the soaring. Imports rose by more than 5 percent per annum paper shows that the adequate funding of institutions has between 1961 and 1986, and are projected to increase by 2.4 critically influenced their ability to implement a pricing percent a year (compared with only 0.7 percent for the policy, and that poor funding is. not necessarily explained world as a whole) until the year 2000 (see Table I). by currency overvaluation. Further, the effectiveness of This paper explains variations in the performance of price policies in ensuring a supply response has itself been national cotton subsectors by focusing on the key interac- heavily conditioned by the quality of the institutions that tions between price and nonprice factors. Price factors refer carry out cotton research, extension, input supply, and to output and input prices, and to exchange, interest, and commercialization-which in turn reflects the broad politi- wage rates. Nonprice factors include: (i) the agro-ecological cal environment that determines a national commitment to 7 succeed in cotton and to develop (and retain) human production currently accounts for almost half of Sub- capital. Saharan African output, compared to an II percent share in Two quite distinct camps can be seen among cotton 1961/62. producers over at least the past two decades. Broadly The next section describes the production record of the speaking, the two camps can be described as francophone six MADIA countries that are the focus of this paper. The and anglophone, although these terms are only convenient following two sections seek to explain divergences in shortcuts to refer to a great many organizational differences. performance in terms of the relative impact on cotton The cotton-producing francophone countries, for example, profitability of price and nonprice aspects of the production have emphasized technology-led intensification, while environment. Then the paper weighs the evidence in favor extensive cultivation has characterized the anglophone of well-coordinated, if not fully integrated, organization of systems. Figure 2 illustrates the growth of cotton production the cotton sector. in the fourteen francophone countries relative to that of The economic viability of the cotton sector in franco- their fifteen anglophone counterparts, several of which phone Africa has come under questioning, particularly since were very successful during the colonial period but have the collapse of world cotton prices in 1986. The last section since declined. As can be seen, the francophone group, discusses questions of economic, technological, and finan- which started producing cotton on a commercial scale only cial viability in the context of recent changes in inter- in the early 1960s, has overtaken the anglophone group, national trade and aid arrangements, and in the domestic which has shown slow growth at best. Output in the former macroeconomic, political, and institutional arrangements French colonies grew by 740 percent between 1960 and that have affected the cotton subsectors. The implications 1985, while the other African producers collectively of these changes for the reform measures being developed increased production by only about 60 percent in that time, by donors and governments are also discussed. Although primarily because seed cotton yields are substantially the examples are taken from the MADIA countries, the higher in francophone than in anglophone Africa as a result lessons could easily be generalized to other rainfed of the higher level of technology used. Francophone environments in Africa. Figure 2 Seed cotton production in francophone and anglophone Sub-Saharan Africa, 1961-87 (excluding Zimbabwe) Million metric tons 1.a - 1.7 1.8 1 5 Non-francophone 1-4 Francophone 1.3 1.2 1.1 1 0.9 0.8 0.7 0.8 0.5 0.4 0.3 0.2 0.1 0 1961 1963 1965 1967 1989 1971 1973 1975 1977 1979 1981 1983 1985 1987 Source: World Bank, BESD Databank. 8 Cotton Performance In MADIA Countries Cotton performance in the six MADIA countries-franco- was created in 1949 to promote cotton production and phone Cameroon and Senegal, and anglophone Nigeria, provide support for downstream activities in the then Kenya, Tanzania, and Malawi-parallels the larger history of French colonies. Until local cotton parastatals were created cotton performance in franco- and anglophone Africa. As in the 1970s, CFDT was virtually the only actor linking all Table 2 shows, production in both Cameroon and Senegal aspects of the francophone cotton sectors. is dynamic. Production in Tanzania, Nigeria, Kenya, and to Even after the local parastatals were created, CFDT a lesser extent Malawi, is stagnating. retained its historical ties with the francophone countries. Much of the difference in performance by francophone It owns substantial shares of the parastatals' equities, countries is attributable to the important role played by the averaging a 28 percent participation in 1987 in the eight Compagnie Francaise pour le Developpement des Fibres Textifes countries where CFDT is most actively involved. (Equity (CFDT), the parent company of the francophone African participation ranges from 40 percent in Mali's CMDT to 17 cotton parastatals. CFDT is a semi-private public enterprise percent in Chad's COTONCHAD.) In addition to equity in which the French state holds 64 percent of the equity. It participation, CFDT provides technical assistance in the Table 2 Seed cotton production in the MADIA countries, 1960/61-1987/88 (in metric tons) Year Cameroon Nigeria Senegal Kenya Malawi Tanzania 1960/61 29,200 203,013 1961/62 25,100 183,813 1962/63 41,400 167,511 1963/64 45,600 214,115 1964/65 44,000 141,353 54 252,612 1965/66 57,500 137,866 302 13,555 290,215 1966/67 55,800 158,994 1,186 13,632 390,410 1967/68 49,085 83,586 4,157 11,441 433,514 1968/69 68,013 172,219 9,905 13,127 11,156 295,875 1969/70 91,334 277,651 11,500 15,819 18,209 390,970 1970/71 38,394 119,651 11,483 17,230 21,389 209,900 1971/72 43,197 116,348 21,547 16,540 22,494 228,300 1972/73 45,296 146,832 23,451 17,220 22,099 197,300 1973/74 27,837 93,865 33,077 16,183 16,212 230,200 1974/75 40,042 155,046 42,375 14,560 21,407 195,000 1975/76 49,462 177,069 30,842 17,985 17,777 221,100 1976/77 47,767 243,677 45,208 19,806 17,956 126,000 1977/78 46,358 112,491 37,491 26,714 22,635 194,694 1978/79 59,497 110,131 33,546 35,442 24,218 16S,082 1979/80 80,335 86.475 26,868 29,213 22,411 177,755 1980/81 84,453 26,390 20,6C7 26,783 23,114 1 74,960 1981/82 79,819 79,971 41,007 24,258 21,739 133,038 1982/83 72,361 58,235 47,081 23,501 14,800 127,993 1983/84 94,580 37,115 30,461 16,271 13,368 140,393 1984/85 97,502 46,150 46,S13 39,281 32,122 152,267 1985/86 115,544 30,845 27,942 27,469 32,710 126,378 1986/87 122,772 75,377 26,871 30,938 21,757 138,060 1987/88 113,700 110,000 38,931 36,563 42,300 147,000 Growth rate (%) entire period: 3.9 -6.3 17.2 4.5 2.0 -2.6 1970-85: 8.1 -9.7 3.8 4.2 1.0 -3.6 Coefficient of variation (%) entire period: 43 54 59 37 25 38 1970-85: 39 56 31 31 24 21 Sources: Cameroon: SODECOTON, Rapport Annue! de la Campagne 1 985/86; Ministere de Cooperation, RPpublique Francaise, "Le Coton en Afrique de l'Ouest et du Centre," August 1987. Figures for 1986/87 and 1987/88 from personal correspondence with Francois Bocchino, General Manager, CFDT, Paric. Nigeria: de Matharel, "Study of Cotton Marketing, Seed Production and Quality Controi. Paris: CFDT, 1987, Table 1. Senegal: Ministere de la Cooperation, Republique Francaise, "Le Coton en Afrique de l'Ouest et du Centre," Paris, August 1987. Figures for 1986/87 and 1987/88 from personal correspondence with Francois Bocchino, General Manager, CFDT, Paris. Kenya: Ministry of Agriculture. Figures for 1986/87 and 1987/88 calculated from cotton lint production (in bales) assuming a 32% ginning outturn. Malawi: Malawi Statistical Yearbook 1975 and 1983; Malawi Economic Report 1987. Tanzania: 1960-1969: World Bank. 1970-1978: Marketing Development Board. 1979-1987: USDA, "Tanzania Cotton Annual Report," 1987. form of seconded personnel and ad hoc field missions to the Figure 3 African counterparts. As of December 1987, 133 CFDT Seed cotton yields in the MADIA countries, experts were resident in francophone Africa, excluding ad 1970/71 and 1983/84 hoc support missions. Apart from the revenue it earns from providing technical assistance and consultant services, 1000 kg/hectare CFDT earns a commission of 6 percent on the value of ' ' cotton exports from the former French West Africa. Thus, it ''3 has a substantial interest in increasing cotton exports, an '' -2 incentive that the erosion of trade and aid relationships ' with Britain has removed from anglophone cotton-produc- 1970/71 1983184 ing countries. o- CFDT research and technical assistance has promoted a o.l huge gap in yields between francophone and anglophone 07 countries, with the former being three to four times higher than the latter (Figure 3). As can be seen from Table 3, fertilizer and pesticide use is widespread in the franco- phone countries, reflecting use of a much higher level of technology generally. It is important to note that-although a French assistance may have overstayed in francophone Africa in general-the number of CFDT experts in the domestic African parastatals has been declining at least over the period 1985-87 (see Table 4). More importantly, Cameroon Senegal Nigeria Kenya Malawi Tanzania countries such as C6te d Ivoire and Mali, with the fewest Francophone Africa Anglophone Africa CFDT resident experts per cultivated area, have cotton yields that are as good or better than those in Cameroon Note: 1970/71 and 1983/84 are midyears in 5 year averages. and Senegal. It appears, therefore, that the effective transfer Source: Lele, van de Walle, and Gbetibouo. of technology to indigenous capacity is already taking place in francophone Africa and could be speeded up. Cameroon Figure 4 and Senegal, the francophone MADIA countries under CFDT presence and seed cotton yields in selected review in this study, are actually outliers in terms of the francophone Sub-Saharan African countries. density of CFDT expert population. As can be seen in Figure 4, the areas of higher yield are concentrated in the Seed cotton yields (kg/ha) northwest quadrant, where there are fewer expatriates; Thousands Cameroon clearly lies in the northeast quadrant, and 1 CS7 S Senegal somewhat in the middle. 1.4 ML8* In anglophone countries, use of chemical inputs remains 1.3 < 5N52 . W5 BFS7a more or less exceptional, representing a low input/low 1.2 T TGS5 ... ., B output extensive approach to growing cotton. Intercropping 11 TDAV8 5BAVl cotton with food crops has all but disappeared from the 5F 5 . SNSe francophone countries, but it remains a common practice in I l SNAVl much of anglophone Africa. 09 N TGS11 l In addition, French researchers have developed and 08 CHS7 MIGAV NGS6 GSS l CHAV distributed strains producing higher amounts of fiber per 07 CHOV SSa kilo of seed cotton. This ginning outturn, as it is called, DCS5 a averages 40 percent in Cameroon, for example, compared o 3 with 33 percent in Nigeria-or 21 percent more fiber from o.5 -] the same amount of cotton. Anglophone researchers, in CA83 l contrast, opted for developing higher-quality fiber. While V,97 l high-quality cotton fetches a premium price, the added 0'3 2 t I income does not compensate for the lower rates of CFDT experts per 10,000 hectares production found in anglophone countries. Note: BD = Burundi CM = Cameroon BF = Burkina Faso MG = Madagascar CA = Central African Republic NG = Niger Cl = C6te d'lvoire SN = Senegal CH = Chad TG - Togo 85, 86, 87, and AV refer to 1985, 1986, 1987, and their averages, respectively. Source: CFDT, Annual Reports. 10 Table 3 Characteristics of cotton farming systems in the MADIA countries Cameroon Nigeria Senegal Kenya Malaw' Tan2ania Average area planted per farm (ha) 0.6 0.8 0.35 0.45 0.63 1.25 Average land availability (ha/person) Far North: 1.98 1.14 3.22 Coast: 6.18 1.28 2.1 North: 11.16 Nyanza/Western: 0.46 Average labor use: (persondays/ha) Manuallabor 115 115 115 115 115 115 Animal traction 75 Mechanized traction 60 % of Acreage: intercropped with food crops 0% 70% 0% 70% 150% 00% Sprayed 95% 0% 99% 10-20% 90% 10% Fertilized (chemical) 95% 0% 96% 0-5% 13% Yields (kg/ha) 1,300 250 1,012 350 700 400 Ginning gatios 40% 33% 39% 32% 36.8%/6 33.5% Competing crops Groundnuts Maize Groundnuts Sugarcane Maize Maize Maize Sorghum/Millet Maize, Tobacco Sorghum/Millet Groundnuts Notes: These estimates concern only rainfed cotton cultivation. Cameroon figures concern 1986-87; Nigeria and Kenya figures 1 985-866 Malawi figures 1980-81; Senegal figures 1984/85; and Tanzania figures 1983-84. Sources: various unpublished documents. Cameroon Senegal Between 1960 and the 1970s, production growth in Camer- Cotton was introduced into Senegal only in 1963, as a cash oon largely reflected expansion of the area under extensive crop to diversify the economy away from heavy dependence cotton cultivation (see Table 5). Yields were generally on groundnuts. After an initially difficult start with irrigated stagnant, at 500-600 kilos per hectare (see Table 6). cotton in the Fleuve region, commercial-scale production However, the protracted drought of the early 1970s, when began around 1965 in rainfed areas of upper Casamance producers retreated into subsistence agriculture and pro- and Eastern Senegal, the two regions which now contribute duction declined from 91,000 tons to 38,000 tons within a up to 90 percent of all of Senegal's relatively small cotton year, resulted in major structural changes in the sources of output. Unsuitable soils and problems of salinity and of Cameroonian production growth. First, CFDT shifted its acceptable levels of rainfall, especially in upper Casamance, production strategy from an extensive to an intensive one. limit potential production to an area of roughly 40,000 To keep producers interested in cotton, CFDT emphasized hectares. the use of chemical fertilizers and pesticides, and plot Senegal's cotton sector has been dynamic despite its consolidation was encouraged to facilitate mechanization. harsh climatic environment. The late-arriving industry Second, the authorities encouraged a shift in production benefited from CFDT's experience in other francophone from the North, where continuous cultivation and declining countries. As early as 1966, seed cotton yields were above rainfall had resulted in declining yields, to relatively higher- 1000 kilos per hectare-low by the standards of franco- potential areas further South, where soils were less phone West Africa but much higher than the best yields in degraded and precipitation levels are higher and more anglophone Africa. When yields fell below 700 kilos per predictable. hectare during the mid- to late 1970s, mainly because of These policy changes yielded quick results. Production adverse weather conditions, new and better adapted, increased by 9 percent annually from 1975/76 to 1985/86, shorter-germination varieties with better fiber characteris- even though the area under cultivation declined by 0.3 tics were introduced to the farmers, with the result that percent a year. Almost two-thirds of the crop now comes both seed-cotton and fiber yields rebounded from below from the North Province, as compared to only one-third in 700 kilos per hectare and 270 kilos per hectare, respec- the 1960s, when the more northerly Far North Province tively, to more than 1000 kilos per hectare and 400 kilos per dominated production.' Yields have been consistently hectare in the 1980s. higher both because of the change in technology and the Cotton is second only to groundnuts among Senegal's shift to new production areas. They now surpass 1.2 tons agricultural exports. As in Cameroon, production prospects per hectare. In some regions of the Benoue, yields have depend on the availability of land of suitable quality and reached 2 tons per hectare, somewhat of a record for on the ability of SODEFITEX to keep Senegalese cotton rainfed cotton in Africa. cost-competitive on the international market. Cost-reducing Cotton is Cameroon's third largest agricultural export by technology, further intensification, and the adoption of such value (behind cocoa and coffee). It is generally believed high-performing varieties as IRMA 96-97, which was intro- that production will remain profitable at the farm level even duced in 1987, offer possibilities on the competitiveness with reduced real producer prices and abolition of subsi- front, although the high cost of CFDT's technical assistance dies on inputs. At the national level, however, the economic for training of indigenous capacity tends to undercut cost- viability of cotton production is an important issue, because reducing possibilities. of SODECOTON's high cost of operation, its monopolistic hold over cotton trade, and a belief that CFDT focuses excessively on cotton to the detriment of other crops. II Table 4 CFDT presence in local cotton parastatals Number of CFDT Area under cotton CFDT agents per Production of CFDT agents Average seed agents cultivation 10,000 hectares seed cotton per 10,000 cotton yields (in hectares) (in metric tons) metric tons (in kg/ha) Burkina Faso 1985 12 82,300 1.46 88,134 1.36 1,071 1986 11 94,625 1.16 115,491 0.95 1,221 1987 10 126,850 0.79 169,227 0.59 1,334 Average I1 101,258 1.14 124,284 0.97 1,209 Central African Republic 1985 23 79,563 2.89 45,516 5.05 572 1986 16 83,105 1.93 35,479 4.51 427 1987 14 65,677 2.13 24,904 5.62 379 Average 18 76,115 2.32 35,300 5.06 459 Togo 1985 10 43,562 2.30 54,756 1.83 1,257 1986 6 68,824 0.87 63,558 0.94 923 1987 3 61,408 0.49 79,067 0.38 1,288 Average 6 57,931 1.22 65,794 1.05 1,156 Madagascar 1985 9 23,595 3.81 33,813 2.66 1,433 1986 9 32,954 2.73 42,871 2.10 1,301 1987 7 42,850 1.63 40,886 1.71 954 Average 8 33,133 2.73 39,190 2.16 1,229 Niger 1985 3 4,627 6.48 3,884 7.72 839 1986 3 5,509 5.45 4,389 6.84 797 1987 2 9,421 2.12 8,138 2.46 864 Average 3 6,519 4.68 5,470 5.67 833 Burundi 1985 4 6,596 6.06 7,155 5.59 1,084 1986 3 6,664 4.50 7,895 3.80 1,185 1987 3 6,753 4.44 8,420 3.56 1,215 Average 3 6,671 5.00 7,823 4.32 1,161 Guin6e 1985 8 240 333.33 160 500.00 667 1986 10 1,161 86.13 568 176.06 473 1987 8 872 91.74 605 132.23 694 Average 9 758 170.40 444 269.43 611 Guinee Bissau 1985 5 2,428 20.59 1,215 41.15 500 1986 5 2,312 21.63 1,023 48.88 443 1987 4 1,889 21.18 1,329 30.10 704 Average 5 2,210 21.i3 1,189 40.04 549 Senegal 1985 9 38,842 2.32 27,942 3.22 719 1986 7 25,482 2.75 26,652 2.63 1,046 1987 6 29,000 2.07 38,700 1.55 1,334 Average 7 31,108 2.35 31,098 2.25 1,000 Cameroon 1985 48 89,232 5.38 115,544 4.15 1,295 1986 38 94,420 4.02 122,520 3.10 1,298 1987 27 94,555 2.86 113,900 2.37 1,205 Average 38 92,736 3.82 117,321 3.05 1,265 Cdte d'lvoire 1985 18 153,054 1.18 189,314 0.95 1,237 1986 18 159,296 1.13 213,532 0.84 1,340 1987 15 180,310 0.83 260,000 0.58 1,442 Average 17 164,220 1.02 220,949 0.76 1,345 Malia 1985 15 139,218 1.08 175,092 0.86 1,258 1986 16 145,747 1.10 201,653 0.79 1,384 1987 9 142,222 0.63 187,000 0.48 1,315 Average 13 142,396 0.88 187,915 0.67 1,320 Chad 1985 19 148,103 1.28 99,469 1.91 672 1986 20 124,075 1.61 89,469 2.24 721 1987 34 148,652 2.29 121,300 2.80 816 Average 24 140,277 1.63 103,413 2.26 737 Note: a Only in CMDT areas. Sources: CFDTpersonnel numbers: Compagnie Francaise pour le Developpement des Fibres Textiles, Rapport d'Activitb, issues from 1985,1986, and 12 1987. Production and area numbers: Ministere de la Cooperation, "Le Coton en Afrique de l'Ouest et du Centre," Paris, August 1987. Compagnie Francaise pour le Developpement des Fibres Textiles, Rapport d'Activite, issues from 1985,1986, and 1987. Table 5 Areas under cotton cultivation in the MADIA countries, 1960/61-1986/87 (in hectares) Cameroon Nigeria Senegal Kenya Malawi Tanzania 1960/61 54,846 270,000* 38,000 22,000 192,000 1961/62 63,612 300,000* 39,000 33,000 240,000 1962/63 67,868 300,000* 51,000 37,000 280,000 1963/64 72,227 300,000* 30 47,000 37,000 309,000* 1964/65 78,922 300,000' 102 58,000 38,000 348,000' 1965/66 91,755 300,OOOU 386 57,000 41,000 440,000 1966/67 97,820 300,OOOU 962 57,000 53,000 452,000 1967/68 98,081 300,OOOU 3,047 57,000 56,000 402,000* 1968/69 101,314 300,OOOU 6,447 60,000 45,000 364,000 1969/70 108,194 364,OOOU 9,805 65,000 37,000 436,000 1970/71 102,055 445,OOOU 13,618 81,000 48,000 440,000* 1971/72 99,046 405,OOOU 18,318 77,OOOU 53,000 400,000* 1972/73 87,679 405,OOOU 20,359 53,OOOU 50,000 400,000* 1973/74 61,176 445,OOOU 28,630 53,OOOU 53,000 380,000' 1974/75 64,520 486,OOOU 38,588 53,000U 54,000 395,000* 1975/76 73,178 500,OOOU 37,483 80,000 38,000 231,OOOU 1976/77 59,930 526,000 43,845 79,000 40,000 364,000U 1977/78 48,436 704,000 47,109 80,000 40,OOOU 389,OOOU 1978/79 47,130 635,000 48,299 126,OOOU 36,OOOU 455,OOOU 1979/80 56,594 526,000 30,908 90,000 34,OOOU 455,OOOU 1980/81 65,227 476,000 29,913 134,000U 32,OOOU 440,OOOU 1981/82 63,343 445,000 31,977 106,000 35,000 395,000U 1982/83 54,629 429,000 42,018 112,000 36,OOOU 360,OOOU 1983/84 71,092 405,OOOU 33,353 146,000 36,OOOU 344,000U 1984/85 73,316 405,OOOU 46,336 180,000 51,OOOU 340,OOOU 1985/86 89,232 405,000* 38,842 180,000- 50,000* 335,000- 1986/87 94,420 410,000^ 25,490 180,000 53,000 335,000 1987/88 450,000* 180,000' 53,000 338,000 Growth rate (%) entire period: -0.6 2.2 14.8 5.2 0.4 1.0 1970-85: -1.5 -0.4 4.8 7.2 -1.7 -0.5 Coefficient of variation (%) entire period: 24 26 68 48 20 19 1970-85: 23 18 30 39 18 14 Notes: U Unofficial FAO figure * FAO Estimate Sources: Cameroon: Minist&re de la Coop6ration, R6publique Francaise, "Le Coton en Afrique de l'Ouest et du Centre," Paris, August 1987. Senegal: Minist6re de la Coop6ration, R6publique Francaise, 'Le Coton en Afrique de l'Ouest et du Centre," Paris, August 1987. Figure for 1986/87 from personal correspondence with F. Bocchino, General Manager, CFDT, Paris, 1988. All other countries: FAO, 1948-1985 World Crop and Livestock Statistics, Rome, 1987. TUnzania Tanzanian cotton production, based largely in Sukumaland during the period 1984-88 to 86,000 tons in 1988/89, despite in the western region, grew more than tenfold between the an erosion of cotton producer prices vis-5-vis those of late 1940s and the late 1960s, peaking in 1966/67. This maize from a peak of more than 4 to I in 1955 to about 2 growth was largely the result of an expansion of acreage to I in 1987/88. (An even further deterioration, to a I to I under cotton cultivation, though average yields also ratio vis-A-vis food crops, was reported to the authors in a increased somewhat as improved seeds, fertilizers, and personal communication from Carr for the current 1989 pesticides began to make their way to smallholders period.) Nevertheless, cotton production is still expected to (Collinson 1974). Intensification of food crop (particularly hover around 80,000 tons, according to the International maize) production and marketing released land and labor Cotton Advisory Committee, ICAC. for increased cotton production, resulting in increases in Cotton remains Tanzania's second largest agricultural both average plot size and in the number of households export by value (behind coffee). It is still popular because, growing cotton. By the late 1960s, the average area devoted being more drought resistant, it offers greater income to cotton per household exceeded 1.25 hectares, compared security. Nevertheless, production is likely to remain with 0.9 hectares in the late 1950s. depressed. Food crops, apart from being essential for After the early 1970s, yields decreased by 3.2 percent survival, require less labor and can be marketed more annually and production declined steadily as input use flexibly through barter and other nonofficial channels. regressed. Since producer price reforms were introduced in Furthermore, movement of households into more concen- 1984 as part of the structural adjustment process, produc- trated, contiguous village areas through Ujamaa has led to tion has picked up from an average 42,000 tons of fiber the creation of artificial land pressure. 13 TabLe 6 S-,s, cotion yields in the MADIA countries, 1960/61-1986 (in kilograms per hectare) Year Cameroon Nigeria Senegal Kenya Malawi Tanzania 19606/1 532 1,057 1961 /62 395 766 1362/63 610 598 1963/84 631 693 1964685 558 471 529 726 19656S6 627 460 782 238 660 t 966; / S 570 530 1,233 239 864 1967/68 500 279 1,364 201 1,078 1968/69 671 574 1,536 219 248 813 9 69.7 0 844 763 1,173 243 492 897 1970/71 376 269 843 213 446 477 1971/72 436 287 1,176 215 424 571 1972/73 517 363 1,152 325 442 493 1973/74 455 211 1,155 305 306 606 197/4 '76 621 319 1,098 275 396 494 1 973/76 676 354 823 225 468 957 1976/77 797 463 1,031 251 449 346 1977/78 957 160 796 334 566 500 1978/79 1,262 173 695 281 673 369 1979/80 1,419 164 869 325 659 391 1980/81 1,295 55 689 200 722 398 1981/82 1,260 180 1,282 229 621 337 1982/83 1,325 136 1,120 210 411 356 1983/84 1,330 92 913 111 371 408 1984/85 1,330 114 1,012 218 630 448 1985/86 1,295 76 719 153 654 351 1986/87 1,298 184 1,054 172 410 412 1987/88 409 203 798 439 Growth rate (%) entire period: 4.5 -8.1 -0.4 -1.3 3.1 -3.8 19 70-1985: 9.3 -9.8 -1.1 -3.1 2.7 -3.2 Coftficient of variation (%) entire per od: 43 62 24 23 27 38 1970-85: 40 53 1 9 25 24 32 Source: Derived from Tables 2 and 3. Kenya In Kenya, cotton has been grown in the West and on the ments have generally been extremely uneconomical (Lele coas, since the turn of the century; it was introduced into and Meyers 1986). the Rift 'Valley province in 1974/75. Nevertheless, cotton remains an unimportant crop relative to tea and coffee, with Other MADIA Countries only 1 percent of value added in agriculture. Production of The performance of cotton in Nigeria has been dismal. seed cotton more than doubled between 1965/66 and Nigeria was once a major exporter, but production has 1978/79, peaking at 35,400 tons. Again, however, area declined by an apparent 8 percent a year since 1970 and expansion (averaging 7.2 percent a year) has been the most Nigeria must now import to cover much of its internal need s,gnificant source of growth, while average yields have for cotton. (Estimates of the contribution of domestic declined (by 3.1 percent). As a result, Kenya's output is production to the requirements of the domestic textile srmall in relation to the more than 90,000 tons of fiber that industry range from 20 percent to 65 percent.) Yields its textile industry processes each year, and substantial appear to have declined (by 9.8 percent a year between imports are required. 1970 and 1985) faster than area (0.4 percent), and stood at Several features of the recent evolution of Kenyan output around 100 kilos per hectare in 1985. Most of the crop is deserve notice. First, production has shifted away from its grown in the North, and, as in Cameroon, Nigerian experts traditional center in the Western province. What growth attribute some of the decline to diminishing rainfall (Yayock there is has come almost entirely from the Central and and Kumar 1988). Eastern province, where production resumed in the 1960s Malawi is one of the better performing countries in terms after a long setback due to repeated pest attacks.2 Second, of yields among the anglophone set. Production stagnated irrigated production, from roughly 3,200 hectares on the until 1983/84, picked up briefly during the course of Tana River's Hola and Bura irrigation schemes, accounted structural adjustment (when relative prices shifted in favor for about 25 percent of the total by 1986. Irrigated cotton of cotton), and then dropped back to 1970s levels in 1986/ yields average 2000-3000 kilos per hectare, compared with 87. Due to population pressure, there is a growing tendency 200-3,00 ;.n- rainfed cultivation, although irrigation invest- to intercrop cotton and food crops. 14 Explaining Differences in Performance-Price Factors Differences in the relative performance of the cotton by highly adverse macroeconomic policies. Lele has industry in the MADIA countries are largely related to the described elsewhere the negative effects of public expen- profitability of growing cotton instead of an alternative crop. ditures, exchange rates, and trade policies on the agricul- This relative profitability can be affected by price and non- tural sectors of the two countries (Lele 1988b; Lele 1984). In price factors. Indeed, in some countries such as Tanzania the case of cotton, particularly relevant factors were: (i) the and Nigeria, macro policy reforms since 1987 have to some rising prices of food crops relative to cotton in both extent altered the relative profitability of growing cotton by countries, (ii) the adverse incentive effects of extreme improving its price vis-5-vis competing crops such as maize. shortages of consumer goods (in Tanzania), and (iii) the However, price correction, while frequently essential, tends increased cost of wage labor (in Nigeria). Table 7 shows the to be a once-and-for-all phenomenon, whereas increased divergence in incentives for cotton production in Tanzania productivity, which is affected by nonprice factors, can affect and Nigeria by comparing producer-price equivalents for relative profitability on a more continuous basis. cotton lint at official and purchasing-power-parity exchange rates. At official exchange rates, the producer-price equiva- The Impact of Price Factors lents have been substantially above the world price in Absolute price differences across countries are essentially Nigeria since 1977/78 and in Tanzania since 1985. In terms due to macroeconomic policies. In this respect, Nigeria and of purchasing power, however, the producer price was only Tanzania are prime examples of the disincentives created 78 percent of international cotton lint prices in Nigeria in Table 7 Ratio of producer prices for cotton lint (at both official and purchasing power parity exchange rates) to world cotton lint prices, 1960/61-1987/88 (in percent) Year World Cameroon Nigeria Senegal Kenya Malawi Tanzania lint Purchasing Purchasing Purchasing Purchasing Purchasing Purchasing price Official power Official power Official power Official power Official power Official power ($US/kg) parity parity parity parity parity parity 1960/61 0.60 52.35 78.17 1961/62 0.64 50.60 64.14 1962/63 0.63 49.80 56.47 1963/64 0.60 52.17 61.81 1964/65 0.60 52.23 65.60 57.61 1965/66 0.59 52.44 67.48 58.00 1966/67 0.56 54.21 63.18 60.78 1967/68 0.57 53.73 64.12 60.24 72.15 1968/69 0.62 49.29 66.84 55.11 57.36 1969/70 0.56 55.25 83.04 57.97 63.48 1970/71 0.60 48.47 49.29 71.64 75.87 42.80 42.84 71.71 67.12 69.81 68.90 77.49 73.23 1971/72 0.72 42.51 44.29 60.31 59.95 36.03 36.72 63.89 62.87 58.81 57.57 63.93 62.14 1972/73 0.76 43.78 43.91 74.74 74.74 40.19 40.27 66.95 66.95 62.27 62.27 60.65 60.65 1973/74 1.24 36.77 35.26 48.91 52.83 27.76 26.46 43.82 46.75 39.81 43.99 38.65 39.46 1974/75 1.30 36.42 34.75 57.79 58.59 25.38 23.99 52.12 53.84 44.45 48.77 36.34 35.07 1975/76 1.16 45.12 41.02 131.29 112.78 35.22 27.61 72.81 73.83 51.47 56.80 52.62 45.89 1976/77 1.62 37.29 34.20 91.95 65.56 27.82 23.76 48.18 50.55 36.55 39.97 43.99 41.40 1977/78 1.45 48.27 43.01 106.99 77.35 33.83 29.03 74.99 76.72 45.42 52.73 49.67 47.09 1978/79 1.39 52.90 45.76 113.38 78.00 38.45 35.03 91.69 87.75 53.04 61.55 64.09 60.12 1979/80 1.50 57.32 49.01 110.98 72.47 39.51 36.19 91.72 90.62 50.90 60.39 58.32 58.55 1980/81 1.87 52.95 45.85 118.48 72.98 35.66 35.41 74.47 73.80 40.89 47.50 58.36 52.41 1981/82 1.75 49.30 46.05 131.56 73.67 32.45 34.98 67.50 69.39 49.30 54.47 66.08 45.53 1982/83 1.45 55.83 53.76 158.43 88.09 36.62 37.46 69.50 68.53 67.15 78.41 82.11 44.18 1983/84 1.59 48.58 45.42 147.26 69.36 32.95 33.51 54.38 56.83 60.66 68.52 79.04 41.18 1984/85 1.66 51.12 46.47 167.62 56.82 27.64 31.90 58.65 56.54 53.12 55.59 70.73 35.08 1985/86 1.41 61.59 57.83 204.24 78.22 31.55 32.78 64.70 63.27 55.71 58.61 101.71 43.63 1986/87 1.06 82.08 90.57 194.34 91.51 64.68 60.60 90.85 111.02 68.87 78.30 79.93 177.08 1987/88 1.65 79.39 62.20 54.88 268.95 54.14 58.47 63.13 82.14 48.22 72.57 47.58 241.11 Notes: Ginning conversions were used to convert producer prices for seed cotton to producer prices for cotton lint. Sources: Producer prices: see previous table. Ginning ratios: Cameroon: time series from SODECOTON, as cited in Freud, Baris, and Zaslavski, "La Politique Agricole du Cameroun de l'lndependence a Nos Jours," MADIA Study, 1987. Senegal: 39% from Ministere de la Cooperation "Le Coton en Afrique de l'Ouest et du Centre," Paris, August 1987. Nigeria: 33% from de Matharel, "Study of Cotton Marketing, Seed Production and Quality Control," annex ix, Paris: CFDT, 1987. Kenya: 32% from MOA/MCD/CLSMB, "Policies for the Development of the Cotton Sub-Sector," Nairobi: Ministry of Agriculture, March 1987. Malawi: 37% (an average of ginning ratios of 3 varieties) from MOA, "National Agricultural Research Plan," Lilongwe, 1987. Tanzania: 33.5% from FAO/ World Bank Cooperative Programme, "Tanzania, Agricultural Sector Review Mission, Working Paper 3: Export Crops," Rome: FAO, July 1987, p. 48. 15 1985/86, and as low as 44 percent in Tanzania. After the Table 8 naira was devalued in late 1986, Nigeria's farm gate price Nigeria: Import parity price for cotton, 1986 was only 48 percent of the c.i.f. price of imported cotton lint Naira Percent of cost (Table 83. per ton at source Figure 5 charts the evolution of official producer prices for seed cotton in the six MADIA countries, converted to Foreign price 1 986a reflect purchasing powver parity. Producer prices in Kenya, (assuming $1 = N 41 1,882 75.3 Malawi., and Nigeria were higher than those in Cameroon Freight + insurance 400 6.3 and Senegal for much of the period, sometimes by as much C.I.F. Lagos 2,282 91.3 as 50 percent. (in terms of purchasing power parity, Portfees 49 1.9 .anzania's producer prices were similar to those inCamer- oon.) In the last few years, the Cameroonian government Landed border price 2,331 93.2 has reversed course; it now tends to pursue the most Transport from Lagos to generous cotton producer price policy in Sub-Saharan Africa Kaduna (0.2 N/km for 850 km) 170 6.8 (Table 9). Cost at source 2,501 100.0 By and large, however, a comparison of producer prices across countries does not adequately explain the relative Producer price performance of cotton in the MADIA countries. Such (post-1986 devaluation)" 1,200 measurements mask differences in price levels among Producer price as % countries resulting from differences in transportation costs of import parity price 48 and exporting and importing status. In order to measure the true comparative incentives to cotton producers, Note 8 Seedcotton equivalent. therefore, it might be more appropriate to measur_ the Source: Ura Lele and Vishva Bindlish, "How Important are the Relative price of cotton relative to competing crops such as maize or Effects of Economy Wide and Sector-Specific Policies in Explaining groundnuts. the Past Performance of Nigerian Agriculture?" MADIA Study, 1988. Figure 5 Producer prices for seed cotton converted at purchasing power parity exchange rates in the MADIA countries, 1970/71-1986/87 0 3 0.2 0- 1 U Cameroon + Nigeria o Senegal A Kenya X Malawi VTanzania 1970/71 1972/73 1974/75 197A/77 1i78/79 1980/i1 1982/83 1984/85 198s/87 Source: Table 6 !6 Table 9 Producer prices for seed cotton, deflated by consumer price index (CPI 1980=100) Year Cameroon Nigeria Senegal Kenya Malawi Tanzania (CFA/kg) (N/kg) (CFA/kg) (KSh/kg) (Kw/kg) (TSh/kg) 1960 0.70 1961 0.57 1962 98.41 0.47 1963 89.12 0.51 1964 84.25 0.53 1965 82.45 0.52 1966 80.44 0.42 1967 77.61 0.45 92.33 1968 80.13 0.52 92.24 0.30 1969 86.87 0.53 88.70 0.30 1970 82.08 0.43 73.22 3.13 0.32 4.14 1971 81.53 0.37 70.45 3.32 0.29 3.88 1972 75.43 0.43 71.14 3.60 0.31 3.61 1973 83.76 0.44 63.93 3.65 0.31 3.35 1974 80.84 0.46 56.65 4.37 0.32 2.81 1975 71.20 0.68 47.19 4.95 0.30 2.95 1976 82.84 0.56 57.66 4.58 0.30 3.68 1977 85.35 0.49 57.96 5.30 0.32 3.30 1978 75.89 0.41 56.06 5.20 0.30 3.41 1979 76.69 0.36 53.28 4.71 0.27 3.13 1980 80.00 0.40 55.00 4.07 0.23 3.00 1981 81.28 0.38 56.66 3.88 0.26 2.55 1982 83.73 0.39 54.71 3.52 0.31 2.28 1983 78.63 0.35 56.22 3.30 0.30 2.28 1984 89.01 0.31 50.30 3.33 0.27 2.15 1985 93.96 0.36 44.51 3.20 0.27 2.25 1986 91.06 0.48 53.63 3.02 0.24 2.63 1987 83.96 0.45 73.03 2.94 0.25 2.63 Coefficient of variation (%) entire period: 7 20 22 19 9 20 1970-85: 6 21 14 18 8 20 Sources: Cameroon: SODECOTON/MOA from Carole Gagne-Gervais, "Cameroon: The Cash Crop Sector: Its Performance and Future Development Possibilities," USAID/Yaounde, November 1984. Data after 1983: World Bank, "Cameroon: Country Economic Memorandum," 1987. Nigeria: Uma Lele et al. "Nigeria's Economic Development, Agriculture's Role, and World Bank Assistance: Lessons for the Future," MADIA Study, 1989. Senegal: Minist6re du Developpement Rural, Direction Statistique. Kenya: Uma Lele and L Richard Meyers, "Agricultural Development and Foreign Assistance: A Review of the World Bank's Experience in Kenya, 1963 to 1986," MADIA Study, December 17,1986, Table Vl11.2. Malawi: MOA, "Ministry of Agriculture Annual Statistical Report," 1986. Tanzania: Marketing Development Bureau. CPIs from World Bank BESD Database. As Table 10 shows, until the mid-1980s, producer prices important than producer price differences in explaining the moved against cotton (and in favor of maize or, in Senegal, relative profitability, and hence the relative performance, of groundnuts) in all the MADIA countries except Malawi. As a cotton in the MADIA countries. Relative profitability at the result, returns to labor use were generally much higher for farm level changes, for example, if the value and timeliness food crops than for cotton. Indeed, in Kenya, where nonfarm of different input subsidies are factored into the analysis employment and wages are attractive, it has frequently and net returns to cultivation are calculated. Prior to been argued that returns to labor in cotton are not structural adjustment reforms, all six MADIA governments competitive with off-farm opportunities. In Tanzania, where provided free seeds and subsidized sprayers, insecticides, unofficial maize prices have tended to be between three fertilizers, and credit. However, as shown in Table 3, and four times the official price, relative returns to cotton modern inputs were widely used only in Cameroon and had eroded substantially by the early 1980s (Table 10); the Senegal, where delivery systems operated effectively. In the ratio of cotton to maize prices, which stood at 4:1 in the other countries, family labor tends to be the only significant 1950s had moved to nearly 2:1 in 1987/88. The opposite input in cotton production. The "effective" input subsidy results in Malawi are due both to relatively favorable prices element is thus greater in Cameroon and Senegal than in of cotton vis-a-vis maize and to relatively high cotton yields, the other four countries. reflecting Malawi's superior record in cotton research Two important issues need to be investigated in this among anglophone countries (Anthony 1986b). Also, official context-(i) the profitability of cotton at the farm level producer prices for maize have generally been low in under alternative technologies and the relative returns to Malawi, which until recently was a persistent exporter of labor in cotton vis-a-vis other agricultural or nonagricultural maize in an environment of high transportation costs (Lele pursuits, and (ii) the comparative advantage of cotton 1989a; Lele 1989b). production in Africa under alternative technological Productivity differences across countries may be more scenarios. 17 Table 10 Ratio of seed cotton to competing crop (maize or groundnut) prices in the MADIA countries, 1964/65-1987T88 Year Cameroon Nigeria Senegal Kenya Malawi Tanzania Producer Consumer Producer Producer Producer Producer Official Open market price for price for price for price for price for price for price for price for maize maize groundnut maize maize maize maize maize 1964/65 1.50 1965/66 1.50 1966/67 1.50 1967/68 1.50 2.73 1968/69 1.90 3.23 1969/70 1.88 1.61 3.38 1970/71 1.76 1.61 3.54 3.28 1971/72 1.77 1.62 3.18 3.37 4.23 1972/73 1.72 1.30 2.97 2.87 4.58 1973/74 1.90 1.34 3.13 3.43 4.35 1974/75 1.79 1.15 3.37 4.34 3.42 1975/76 1.23 0.78 1.01 2.74 3.77 2.73 1976/77 1.72 0.72 1.13 2.04 2.71 2.25 2.50 1977/78 1.44 0.61 1.13 1.41 3.24 3.52 2.50 1978/79 1.30 0.86 1.18 1.13 3.54 3.94 2.71 1979/80 1.17 1.06 1.33 1.49 4.26 4.19 2.82 1980/81 1.33 0.82 1.32 1.57 3.48 3.25 3.00 1981/82 1.38 0.69 1.36 1.43 3.41 3.24 3.20 1982/83 1.50 0.83 1.11 1.87 3.26 2.45 2.47 1983/84 0.72 1.11 1.87 2.40 3.39 2.69 0.58 1984/85 1.03 1.11 0.96 2.56 3.31 2.73 0.70 1985/86 0.91 1.25 1.16 2.67 4.10 2.10 0.74 1986/87 0.93 1.11 2.05 2.63 4.10 2.68 2.05 1987/88 1.36 2.63 3.90 2.37 Sources: Cameroon: Cotton prices are from SODECOTON data. Maize prices are from FAO, 'Statistics on Prices Received by Farmers," Rome: FAO, 1984. Maize consumer prices from Direction de ia Statistique et de la Comptabilit6 Nationale. Senegal: Sidi Jammeh and Chandra Ranade, "Agricultural Development and Foreign Assistance to Senegal: A Review of the World Bank's Experience in Senegal, 1969-1986," MADIA Study, March 1987. Nigeria: Uma Lele et al., "Nigeria's Economic Development, Agriculture's Role and World Bank Assistance: Lessons for the Future," MADIA Study, 1989. Kenya: Lele and Meyers, "Agricultural Development and Foreign Assistance: A Review of the World Bank's Experience in Kenya, 1963 to 1986." Data for 1986/87 from Ministry of Agriculture. Malawi: ADMARC Reports and Ministry of Agriculture. Tanzania: Marketing Development Bureau, and United Republic of Tanzania, Ministry of Agriculture and Livestock Development, "Annual Review of Agricultural Marketing, 1987," Dar es Salaam, 1988. Farm-level analysis. For the farm-level analysis, the hand, the use of traction is often economically unjustifiable important issues are the extent to which cotton production without the increase in yields made possible by improved is economically viable either for internal consumption or for varieties. On the other hand, without labor-saving technol- exports, and whether under traditional low input/low yield ogy at the critical stages of planting, weeding, and harvest- extensive production technology or modern high input/high ing, labor shortages can make the application of modern yield intensive technology-all in the context of rapidly inputs economically infeasible. This is the case in the changing exchange rates, leading to very substantial anglophone countries, where labor use in cotton competes changes in the prices of tradeables (inputs and outputs) vis- with that in food crops. Of course, successful development 3-vis nontradeables (e.g., labor). To make such an analysis, of cotton can also increase food security by increasing the the following building blocks are required: li) the factors level and stability of incomes (Weber et al.). influencing returns to labor in cotton vis-a-vis other alterna- However, there is some question whether, at current tives under traditional and modern technology; (ii) the producer and input prices, the application of more inten- economic value of these factors, and (iii) the costs of sive modern technology provides high returns to factors of developing and maintaining a cotton industry in the context production, especially labor (which is still the most critical of growing macroeconomic imbalances. Data limitations input in African agriculture), in the absence of subsidies. For allow only very preliminary judgments on these important example, Carr has documented that, with the devaluation of questions. the Tanzanian shilling, the kilograms of cotton required to The example of Cameroon indicates that relative returns purchase insecticides increased five-fold (from 79 kilos in to cotton may be better if production involves the use of 1970/71 to 414 in 1988/89). Even after inefficiencies in modern technology. In Cameroon, under high input/high cotton and input marketing and prices are corrected, the yield technology, especially with the use of animal or majority of farmers will not obtain a marginal return of 2 to motorized traction, returns to cotton production are sub- I (considered the minimum incentive needed) from using stantially above both the minimum wage and returns to recommended pest-control practices. Carr concludes that food crop production even though Cameroonian wage rates most farmers will continue their present tactic of using and food prices have been high. The need for complemen- insecticides only when a localized pest attack is particularly tarity in the introduction of mechanical and biological severe. Thus, cotton production in Tanzania will continue to technology is important to stress in this context. On the one depend largely on the area allocated to cotton, which in 18 turn is a function of the extent to which demand for food risen substantially faster than cotton prices over time. This makes cotton production attractive (Carr 1989). is in part because the use of anirmals or, particularly, For the present analysis, we have estimated returns to motorized traction reduces labor requirements from 115-120 person-days of labor on the basis of farm-management data person-days per hectare to about 60 days. Mechanical on production costs for cotton for all MADIA countries. cultivation does not have the same dramatic effect on These estimates were then compared to returns to labor returns in other MADiA countries because of their low from alternative forms of employment, with the results yields, however. Thus, in Nigeria, despite relatively favor- shown in Table II. The calculations are complicated by able prices, very low yields reduced cotton's profitability methodological issues as well as by the poverty of data on relative to maize until 1987, when seed cotton prices were actual farm-level practices and yields. Nevertheless, they raised. In Senegal, on the other hand, cotton's profitability help to explain the shift of labor away from cotton in such has eroded despite relatively high yields (albeit lower than countries as Kenya and Nigeria, where farmers use low Cameroon's) and subsidies on inputs. owing to a producer input/low yield technology. In both countries, nonagricul- price that is the lowest among the MAI)IA countries. tural employment opportunities have grown, and returns to The analysis so far has assumed that official producer cotton production are low compared to any other form of prices of cotton are in fact the relevant measure for labor earnings. In Kenya's case, this is not due to acute calculating returns. In Nigeria, Kenya, and Tanzania, how- macroeconomic policy distortions of the sort found in ever, cotton producers are rarely paid on time (with delays Nigeria, although the agronomic, disease and pest prob- of up to 9 months in Kenya and up to 2 years in Tanzania), lems encountered in Kenyan cotton production make the nor do producers receive the full official price. An instruc- problem of increasing productivity particularly vexing. tive comparator in this regard is Senegal, where the Caisse In West Africa, Camefoon's high cotton yields make the Centrale de Cooperation Economique ICCCEI provides working crop profitable. indeed, when animal or motorized traction capital to the Senegalese cotton parastatal, SODEFITEX, is used, cotton has a 50 to 100 percent advantage over thus ensuring timely payments to cotton producers. maize even though, as has been noted, maize prices have While it is tempting to attribute the payment problem in Table 1 1 Returns to labor from cotton and alternate agricultural and nonagricultural employment in the MADIA countries (in US$ per day, using official exchange rates) Cameroon Nigeria Senegal Kenya Malawi Tanzania Cotton: 1.05 1.85 0.38 2.45 Manual 2.85 0.91 Animal Traction 5.45 1.10 Motorized 7.39 1.44 Alternate Crops: Sorghum/Miiiet 3.78 1.87 Maize 4.52 2.84 3.00 1.58 1.22 204a Groundnuts 3.43 2.08 6_05b Rice 4.33 6.50 Sugarcane 6.70 Coffee 3.83 Tea 3.82 Alternate Employment: Farm Labor 5.59 Hired Labor 4.77 0.70 Sugarcane Estate 0.54 3.20 Family Labor 0.90 Government Farm Worker 0.81 Municipal Govt. Employee 1.72 Min. wage, private sector 2.00 4.25 Min. wage, public sector 1.73 a Returns to labor for maize at official prices. b Returns to labor for maize at open market prices. Sources: Cameroon: All data calculated from price (1 986/87), labor use and yield information in Price Waterhouse Assoc. (Africa), "SODECOTON - Mission de Diagnostic," Annex Table 14, January 1987. Cotton calculations based on 115 days/ha (manual labor); 75 days/ha (animal traction); 60 days/ha (mechanized traction); maize: 65-70; sorghum/millet: 62; groundnuts: 78-90. Nigeria: Uma Lele et al. 'Nigeria's Economic Development, Agriculture's Role and World Bank Assistance: Lessons for the Future," Ch.5, MADIA Study, 1989. Senegal: data from: Republique du Sbn6gal."SODEFITEX, "Programme de Sauvegarde et d'Ajustement Cotoniere," Dakar, 1987, p. 55. Assumptions made: a/ cotton: use of 200 kg NPK and 3 pesticide treatments: b/ groundnuts: extensive cultivation; c' maize: use of 16 kg of pre-chosen seed. Kenya: numbers calculated from labor use data from Kenya Development Planning Division, Working paper (1 987) provided by Michael Westlake, Ministry of Agriculture, Nairobi, Kenya. Calculations based on 115 days of labor per ha for cotton; sugarcane: 280 days/ha; maize: 148; coffee (smallholder): 342; tea (smallholder): 486; price and yield information from Michael Westlake and MOA. Malawi: numbers calculated from price and yield information from ADMARC figures and MADIA database. Labor use assumptions: cotton: 115 days of labor per ha: maize: 78. Data from Cox and Spurling, 'Unit Farms in Malawi," Bvumbwe Research Station. Tanzania: numbers from Ministry of Agriculture and Livestock Development, "Annual Review of Cotton," Dar es Salaam, 1987 and 'Price Policy Recommendations for 1985 Agricultural Price Review," Dar es Salaam, 1985. Assumptions made: a/ sugarcane number assumes all family labor: bl maize numbers assume price, yields, and labor use for "high potential" variety. Exchange rates from IMF, International Financial Statistics. 19 Kenya and Nigeria to the erosion of cotton profitability cially when drawing inferences of comparative advantage in resulting from currency overvaluation, this argument does the production of a specific crop on a cross-country basis not necessarily apply. In Kenya, the exchange rate has not and over time. Variations in assumed shadow prices of been greatly overvalued, and in both Kenya and Nigeria domestic factors and exchange rates, for example, will cotton producers cater mainly to an internal market. Rather, influence the DRC estimates and, thus, the conclusions as one must concluide that when cotton payments per se to which country has the comparative advantage or the become uncertain, high official producer prices and input extent to which it has gained, maintained, or lost the subsidies have little positive impact on producer growing comparative advantage over time. A review of various DRC decisions. estimates made by the Bank's operational staff suggests, for In this context, the annual nature of cotton cultivation is instance, that the DRCs for cotton production in Nigeria worth stressing. Unlike perennial crops such as coffee or (included in the Agricultural Sector Review of September -cocoa, cotton does not require many sunk costs, so the 1987) assumed a seed cotton yield of 1000 kilos per hectare. producer needs to be persuaded anew to plant cotton, and On this basis, cotton production in Nigeria seems "more his/her reaction to uncertainty is swifter. In general, the advantageous" than in Cameroon or Kenya. But Nigeria's elasticities of cultivated land to price changes tend to be reported average cotton yields for the past three decades much larger in areas devoted to annual crops than in those never rose above 350 kilos per hectare. A more realistic set of perennial crops, which occupy land quasi-permanently. of assumptions should show a higher DRC for Nigeria; and The short-run elasticity for perennial crops is often esti- that cotton in francophone African countries is at least as mated to be almost nil. Short-run elasticities for cotton efficient a producer of foreign exchange as in Nigeria even acreage, in contrast, range from 0.13 to 1.84 for MADIA though the francophone countries have been heavy users of countries.3 As to price elasticities of output, the largest imported technology, with all DRC's being less than one. short-run (current to two-year lagged periods) elasticity of This happens partly because of the yield difference. If a perennial crop is reported in a comprehensive study by francophone DRC's are adjusted for the overhead cost of Askari and Cummings to be 0.87. However, as Askari and expatriate assistance, the figures should be slightly higher., Cummings report, short-run output elasticity can be as high Nevertheless with the declining number of expatriates in as 1.95 in the case of an annual crop such as cotton, and we the francophone countries' cotton companies, and given the have found elasticities as high as 2.02 for Nigeria and 2.06 fact these latter companies now have a mandate far beyond for Kenya (see Annex Table 1). cotton development, it is necessary to revise downwards In conditions of uncertainty, producers tend to attach the costs pertaining specifically to cotton production. more importance to food security (even though cotton, being less susceptible to drought than maize, is attractive Measuring Price and Nonprice Factors both for spreading the risk of crop failure and as a source Two important issues for the future of cotton production in of cash income). Thus, the characteristic prudence of Africa are the extent to which nonprice factors can increase producers in the face of uncertain returns to cotton seems productivity of cotton and whether subsidies are needed to to explain the wide swings in seed cotton acreage and maintain and develop cotton production as an export crop production in Tanzania, in addition to their overall stagna- or as a cheaper alternative to importing cotton to meet tion. It also helps to explain the persistence of intercrop- internal demand. The answers to these questions require, ping of food crops with cotton in the anglophone countries. first of all, an understanding of the relative importance of The uncertainty of payment and lack of an effective price nonprice factors in cotton supply response. support also clearly decrease the incentive for cotton We have developed a set of models featuring price and producers to invest in such technology as sprayers and proxies for nonprice variables to see if their relative pesticides, which would allow intensification and higher importance can be measured. (See Annex I for a full yields. The fact that the least-performing cotton-producing discussion of methodology.) Whereas price variables are countries are those where the variability of prices is also included in the models explicitly, the effects of nonprice the largest (see Table 9) attests to the importance of factors are assumed to be the residual effects on cotton output effective price support. The consequences of the lack of after price effects have been taken into account. There are effective producer prices are visible in the large variability several reasons for this. First, data on nonprice factors such of acreage devoted to cotton in the countries where as investments in research and development (R&D), exten- producers face high price variability for both cotton and sion, infrastructure, and quality of management are hard to competing crops. come by. In addition, the inclusion of nonprice factors into Comparative advantage. The second, related issue for a single-crop supply model raises the conceptual problems assessing variations in performance is the comparative described in Annex 1. Also, since output prices are not advantage of cotton production in Africa under alternative always paid nor inputs delivered in timely fashion, the technological scenarios, given rapidly declining soil fertility, extent to which a price effect or a nonprice effect is at work growing population pressure, and the increasing attractive- is difficult to discern. We have considered the ability or ness of food crop production (and nonagricultural employ- inability of institutions to deliver inputs or make timely rnent), on the one hand, and stagnant or slowly increasing payments as a nonprice factor, although it has price effects international cotton prices and the rapidly increasing cost by influencing the profitability of cotton production. By of imported inputs, on the other. using only the relative output prices of cotton vis-a-vis Although the concept of domestic resource cost (DRC) maize as the price variable, we would risk understating the does not reflect the costs of developing and maintaining importance of the price response by not capturing the sraallholder production, DRC estimates provide a practical effects of nonprice actions on prices themselves. Where and simple way to rank crops in terms of their comparative inputs delivered on time are subsidized but captured in costs or foreign exchange earnings or savings. These institutional nonprice variables, price elasticity coefficients estimates need to be used with caution, however, espe- clearly understate the role of prices. Furthermore, where 20 subsidies are not included, price effects are overstated. producer price (see Figure 6). In countries where the Few estimates of price elasticities, including those reported institutional support to cotton production (including input in the seminal study of Askari and Cummings, have subsidies) has been more effective, the supply of cotton attempted to capture these complex interactions, in part has effectively shifted outward, and nonprice factors, as because of the difficulty of obtaining relevant data. defined here, explain most of the variations in production In an attempt to provide a quantitative measure of the levels. Nonprice factors, through various schemes (research, importance of nonprice factors, we have run two sets of credit, and subsidized inputs), have allowed procurement regression models to explain production and yield varia- prices to remain more or less constant in real terms lin tions in the MADIA countries as a function of price, weather, Cameroon) or have compensated for the adverse effects of and institutional variables. The models show that nonprice declining real prices (in Senegal). By contrast, in countries factors, as proxied by a time trend following the traditional where there has been no significant technological change way of accounting for supply shifts-almost invariably have via mechanized or even ox-plough cultivation lor via a more significant impact on production and yield than fertilizer and pesticide use) and no significant institutional prices alone. That is, the coefficient of the time trend was improvement-i.e., where labor is the only major factor of significantly different from zero more often than was the cotton production-the relative price of cotton vis-3-vis coefficient of the price variable. competing crops has been the most important factor of The significance of nonprice factors in the MADIA cotton supply response. This explains why price elasticities countries is even better illustrated by simply plotting the are higher in Nigeria and Kenya than in Cameroon and production levels of seed cotton together with its real Senegal. Figure 6 Production and real price indices for seed cotton in the MADIA countries, 1970-87 (1970 = 100) Cameroon Kenya Malawi 450 '0 450 _S0 . .... 4 00 <004 *00 350 350 350 100 300 300 250 250 250 200 / 200 200 150 /150 100 500 100 00 so 5 0 0 _ _ _ _ _ _ _ _ _ _ _ _ 1970 1972 1974 1976 1908 1980 1902 s984 19S6 19 072 1974 1976 1978 1980 1982 :984 1986 1970 1972 197- 1976 1978 1980 0D82 9S 19.6 Nigeria Senegal Tanzania 450 450 450. 400 400 4X00 350 350 350 300. 300. 300. 250 250 / 250 200 200 / 200 150 150 150 0*0 0 1970 1972 1974 1976 1978 1980 1982 1984 1986 1970 1972 1974 1978 1978 1980 1982 1984 t986 1970 1972 1974 1976 1978 1080 1982 1984 1986 o Production index + Real price Index Note: Producer prices deflated by Consumer Price Index (1980- 100). Source: Tables 2 and 6. 21 Explaining Variations in Performance-Nonprice Factors Given declining soil productivity and the rising profitability Tanzania due to continuous cultivation, and cotton's lack of of food crop production, the question arises whether cotton responsiveness to fertilizers during research trials near production can be made attractive relative to food crops for Ukiruguru, suggests that environmental problems-though internal consumption, and perhaps even exports, without they are not unique to cotton or to Sub-Saharan Africa-are improvements in technology and input delivery and, if more complex than previously recognized and have not necessary, subsidies on inputs. These and other nonprice received the attention they deserve. factors have a major impact on cotton performance in the Controversy arises because the long-term autonomous MADIA countries, as the following discussion will show. effects of population growth are difficult to separate from those resulting from cotton production price incentives. Agro-ecological Factors Traditional cultivation systems relied on an extremely long Cotton can be produced under a wide range of tempera- fallow period to allow soils to rejuvenate and regain organic tures and rainfall conditions (between the extremes of 12- matter. As long as population densities were low, this fallow 40 degrees Celsius, and at a minimum rainfall of 300 mm system amply compensated for the fragility and low quality per annum), provided that temperatures and rainfall are of soils in most cotton areas. But population has been sufficient at appropriate phases of its growth cycle. It can increasing. The crop's success has halted the rural exodus also be grown on a wide spectrum of soil types and in francophone cotton-growing areas and, in the case of tolerates a wider range of soil acidities than most other Southeast Benoue in Cameroon, has attracted large crops. Table 12 summarizes the diversity of agro-ecological numbers of immigrants, who tend to overexploit the soils. conditions in which cotton cultivation takes place in MADIA As a result, areas under cultivation have increased and countries. In general, the lower the altitude, the better the fallow periods have shortened, if not disappeared entirely. growing conditions. Good drainage is an important factor, The intensification package used for cotton, particularly in since shallow or clayey soils that are prone to excessive the francophone countries-fertilizers, pesticides, animal waterlogging are unsuitable for the plant. In addition, strong traction-quickly reveals the general process of declining winds, as are found in pockets of northern Nigeria, can have fertility, which is less conspicuous in traditional cropping a deleterious effect. Pests and diseases are a particularly conditions. These problems are exacerbated by land tenure serious problem, especially under the low input/low yield systems where titling and land rights are uncertain. approach to cotton production that is prevalent in Kenya. Technical as well as socioeconomic solutions to these By far the most serious and controversial problem relates problems urgently need to be formulated. These solutions to the environmental impact of cotton production. Major need not imply the abandonment of cotton. But because crises may be brewing in a number of countries, especially cotton is often grown in ecologically poor regions highly in francophone West Africa (World Bank 1988), where the susceptible to environmental degradation, cotton producers impressive growth of areas under cotton cultivation has may need to be even more conservation-minded than their been associated with deforestation, erosion, and soil counterparts in forest ares. Cameroon's SODECOTON, in exhaustion. Certainly, evidence of declining soil fertility in particular, needs to become more responsive to conserva- Table 12 Agro-ecological conditions in cotton cultivating regions of the MADIA countries Cameroon Nigeria Senegal Kenya Malawi Tanzania Soil Types Lithosolsa (Benoue Regosals; Lithosols; Ferralitic; Cambisols; (Coast) Ferruginous wI Loamy soils; Valley); Vertisols; Laterites; Ferruginous Vertisols; (Western) Lithosols; Ferralitic Granitic soils Vertisols; Lithosols; w/ Laterite; (Geita); Alluvial Calciumorphic; black cotton soils Alluvial; Soil Chemistry variable pH's variable pH's variable pH's variable pH's variable pH's variable pH's Mean annual 18-39 19-40 15-40 16-35 20-35 15-28 temperatures (in Celsius) Rainfall average 500-1,300 300-750b 600-1,200C 800-1,450 800-1,000 600-1,000 (in millimeters) pockets of 1,000- 1,450 Altitude (in meters) 200-800 300-600 50-350 900-1,220 500-1,000 1,000-1,500 (in Coastal province) Notes: a Poor drainage at times. b to 8 months with less than 1 00 mm rainfall. c For Haute Casamance region where approximately 60% of total cotton in Senegal is produced. Sources: All information found in various agricultural atlases. Information on soils, rainfall averages, and altitudes in Cameroon from personal correspondence with F. Bocchino, General Manager, CFDT, Paris, 1988. Information on soils in Tanzania: Michael Collinson, "Cotton Development in Tanzania: A Review of the Cotton Program in Sukumaland," Africa Rural Development Study Background paper, World Bank, September 1974. 22 tionist concerns. Being a paternalistic organization, it may Given that cotton is often grown in marginal areas where be better able to address these concerns than parastatals the farmer's concern for food security and risk abatement is in other, especially anglophone, countries. high, a particularly consistent set of policies is needed for cotton cultivation-not simply toward the cotton subsector, Cotton as an Instrument of Regional but toward the production environment in general. Govern- or Political Equity ment policy toward improvement of the transport infrastruc- Since cotton is often grown in poor regions,5 the cotton ture, together with good price and supply stabilization, industry has become an important instrument of govern- could have helped to avert the extreme production swings ments' regional development and income distribution that were seen during the Sahel drought of the early 1970s, goals. In the Sahelian francophone countries, where produc- for example. Lele has pointed out elsewhere how unstable tion possibilities are very limited, for example, cotton supplies of maize in government outlets during the 1970s development is often the cornerstone of a whole rural prompted producers in rural Tanzania to minimize their development strategy, a tendency that is reinforced by the dependence on problematical and expensive market sup- fact that many of the cotton parastatals have now evolved plies and to grow more maize for domestic consumption into regional development agencies. In Kenya, the promo- (Lele 1984). During the 1960s, in contrast, Collinson tion of cotton has been perceived as a way of improving observed increased purchases of food by producers and a regional income distribution and countering the long- parallel shift in acreage to cotton cultivation (Collinson standing tendency, dating back to the colonial period, for 1974). males to migrate out of the western cotton-growing areas in Institutional Factors search of off-farm employment (Kenya, Ministry of Agricul- ture 1962). Regional development programs have played a Some key institutional factors have played a role in particularly positive role in the expansion of cotton in the ensuring the effectiveness of incentives to cultivate cotton francophone countries of West Africa, where CFDT's pres- in the MADIA countries. These factors can be subdivided ence has allowed a more technocratic approach to cotton under three headings: (i) those that bear directly on the development to prevail. profitability of cotton through their effect on price factors, Anglophone countries-for many country-specific ethnic liiJ those that have enhanced technological know-how and ideological, and political reasons that are beyond the scope support to the cotton industry, and (iii) those that help of this paper-have shown a greater tendency to allow improve marketing options. Under the first heading, one political and ethnic factors to dominate their relatively more can cite: prompt payment and stability of prices received by decentralized cotton sectors. Thus, Kenya has promoted the producers; timely delivery of inputs and their sale at development of private and cooperative enterprises at predictable price levels; and the ability to secure credit to least partly as a counterweight to Asian domination, dating purchase inputs. The second set of technology-related from the colonial period, of the cotton marketing and institutional factors includes: research and extension; processing industries. Nevertheless, relative to tea and infrastructural complexity, especially the density of roads coffee, cotton is an unimportant crop in Kenya both and buying posts in cotton-growing areas; and, finally, the economically and politically. This lack of political clout may quality of upstream activities to ensure reliability of supply help to explain why the marketing and processing indus- to ginneries and speedy processing of the raw seed cotton. tries are not yet fully owned and operated by Africans. The Diverse combinations of private/public and vertically extent to which this is a result of the complex problems of integrated/decentralized support systems as well as weak/ the cotton sector, which discourage African entrepreneurs strong market mechanisms are possible. from investing in marketing and processing (given other, Input and output price supports. There are great more lucrative options)-as distinct from the unwillingness differences between francophone and anglophone coun- of the government to allow Kenyan private entrepreneurs to tries in the effectiveness of output and input price supports proceed-remains a highly debated issue, especially in the which have directly affected the profitability of growing context of liberalization of the cotton industry, proposed by cotton. In Cameroon and Senegal, cotton farmers are paid the World Bank as part of the structural adjustment process on time, have inputs delivered to them on time, and (see below). benefit from credit and active research and extension, as In Tanzania, cooperative ginneries expanded rapidly in well as from input subsidies. In those countries, the use of the 1960s in the traditional Western Growing Area of modern inputs is growing steadily. By contrast, before Sukumaland, Mwanza, and Shinyanga, in part reflecting the recent structural reforms in Kenya, Tanzania, and Nigeria, clout of the cotton producers, given the importance of the not only were payments to farmers delayed by six to nine crop in the national economy. However, since the early months or more; improved seeds were distributed hapha- 1970s, the role of the export crop producers and producer- zardly, with frequent complaints of late and unpredictable dominated cooperatives has been undermined. The ruling deliveries, insufficient quantities, and uneven or degraded political party (the Tanganyika African National Union, quality. Inputs such as sprayers and insecticides were TANU, and later, the Chama Cha Mapinduzi, CCM) came to frequently unavailable to farmers in Kenya and Tanzania perceive them as a competitive political force, and the because cooperatives lacked enough working capital to buy producers have become increasingly alienated from the and store appropriate stocks. These factors, along with the political process (Hanak 1988). During the 1970s, Tanzania effect of continuous cultivation without a proper fallow also took a political decision to decentralize governmental period, and aggravated by suboptimal use of fertilizers, power to regional administrations, which eroded the role of have led to soil exhaustion and to falling yields. the functional ministries generally. For cotton, the result has Even after the 1986 structural reforms, however, and apart been a more loosely organized cotton industry with less from a number of logistical problems, questions remain as effective technical services from the Ministry of Agriculture to the technological and financial viability of input use in and a major setback for agricultural research and extension East Africa. Years of research trials near Ukiruguru, for (World Bank 1983c). example, have shown an erratic response of continuously 23 cultivated cotton to fertilizer use. This high degree of risk, northern provinces of Cameroon and the provider of all together with the increased cost of imported inputs, now agricultural inputs. In addition, it owns and operates the make input use unattractive to producers without a subsidy. commercial ginneries and oil mills and has an important To isolate the sources of problems and to undertake equity in Cotoniere Industrielle du Cameroun, CICAM, the effective remedies, it is important to know to what extent country's only textile factory. Perhaps most important, the organizational environment produces differences in the SODECOTON has taken an increasingly active role in effectiveness of service delivery. Organizational arrange- promoting food crops. Its extension agents have advised ments in both anglophone and francophone Africa have farmers on maize, rice, and groundnut cultivation and have come under fire during the structural adjustment process, engaged in some marketing. albeit for different reasons. Table 13 illustrates how the All these activities make SODECOTON the dominant cotton subsector is organized in the different MADIA rural development institution in North Cameroon. In 1986/ countries. The diversity of arrangements of the support 87, its activities generated CFAF 15 billion (about US$ 43.3 systems is striking, in particular the different roles played million) in income, including some CFAF 2 billion in the by public institutions. Also of considerable interest is the food crop sector, for the region's rural population. Reflect- extent of integration of the cotton sector and its effect on ing its importance, SODECOTON has had a role in the management. Cameroon and Senegal are at one extreme, implementation of all the major rural development projects with public sector institutions dominating the entire sub- financed by international donors. In Senegal, much the same sector from farm gate to processing and export. can be said of the dominant role of SODEFITEX in the In Cameroon, SODECOTON, which was established in 1974, cotton region, although the presence of groundnuts as a is expected to act as an agent of government policies in the competing cash crop has reduced its regional importance, northern provinces, rather than as a profit-oriented enter- especially given the relatively more attractive producer prise limited to cotton production and marketing. Thus, it prices for groundnuts since the early 1980s. undertakes some applied cotton research on behalf of the CFDT's involvement in francophone Africa has raised the Institute for Agronomic Research (IRA), the public research question of whether integration of services or CFDT's institution for the agricultural sector, and supports an presence per se has been the source of SODECOTON's extension force of about 650, reaching about 140,000 success. This is difficult to establish empirically; one can, farmers. It is also the major source of rural credit in the however, stress that countries such as Zimbabwe, which Table 13 Organization of the cotton sectors in the MADIA countries Cameroon Nigeria Senegal Kenya Malawi Tanzania Research: IRA Research institute ISRA MOA MOA MOA/Parastatal linked to university. TARO Extension: SODECOTON Ministry of SODEFITEX MOA MOA MOA Agriculture Cotton Seed: SODECOTON Ministry of SODEFITEX CLSMB MOA TCMB Agriculture Fertilizer: SODECOTON Licensed private SODEFITEX CLSMB (from MOAJ MOA TFC/Parastatal merchants Pesticide: SODECOTON Licensed private SODEFITEX CLSMB (from MOA) MOA TCMB merchants Credit: SODECOTON National banking SODEFITEX CLSMB (from MOAI MOA Coop Rural system Development Bank/ National Bank of Commerce/ Tanzania Investment Bank Farm gate SODECOTON Private/semi-public SODEFITEX CLSMB/Coop ADMARC Cooperative purchasing: companies Unions/Private Societies Firms Ginning: SODECOTON Private/State SODEFITEX CLSMB/Coop Private Cooperatives/ Unions/Private Private Firms Oil Milling: SODECOTON Private/State SONACOS CLSMB/Coop Private Cooperatives/ Unions/Private Private/Parastatal Firms Transport: SODECOTON + Private SODEFITEX/private CLSMB/Private ADMARC/Private Cooperatives/ private fleets TCMB Export Marketing: SODECOTON Private SODEFITEX CLSMB ADMARC TCMB Textile: CICAM Private ICOTAF/STS/CCV Private Private TEXCO & SOTEXKA Sources: Cameroon: Ministere de la Cooperation, "Le coton en Afrique de l'Ouest et du Centre," Paris, August 1987, p. 18. Senegal: Ministere de la Cooperation, "Le coton en Afrique de l'Ouest et du Centre," Paris, August 1987, p. 1 8 and from personal correspondence with Francois Bocchino, General Manager, CFDT, Paris, 1988. All other countries from various internal World Bank documents. 24 have brought to bear a similarly professional approach, status has been evident not only in anglophone Africa, but have also succeeded in cotton. CFDT's managerial role, even in Cameroon. While SODECOTON has been well while diminished over time, remains important in Camer- funded, its weakened financial status since 1986 has oon, as elsewhere in francophone Africa. However, as reduced its ability to provide inputs to producers, which in pointed out earlier, the cotton sector of Mali and C6te turn has led to a decline of 100 kilos per hectare in seed d'lvoire, where cotton production is even more successful, cotton yields.) are close to fully indigenized. At the end of 1986, 38 Tanzania's cotton sector is even less integrated than that expatriate CFDT agents were still subcontracted to SODE- of Kenya, with different institutions in charge of extension, COTON in an advisory management role, as well as to credit, marketing, ginning, and exporting. In addition, support extension, ginning, and transport services. CFDT institutional instability has plagued the cotton sector since also procures some imported inputs and spares through its the early 1970s, in much the same way as it has the rest of own purchasing division in Paris. And it has been influential Tanzanian agriculture. During the 1960s, the Lint and Seed as a link between the donors and SODECOTON. serving as Marketing Board (LSMB) which had been created in 1953, a project consultant for the World Bank, for example. played a key role in coordinating cotton sector activities Indeed, CFDT helped prepare the US$25 million Northern and marketing seed and lint, while cooperatives performed Province Rural Development Loan (in addition to the SDR ginning and oil refining (Collinson 1974; Lele 1975). In 1973 9.5 million IDA loan) that SODECOTON executed on behalf LSMB was replaced by the Tanzania Cotton Authority (TCA), of the government. In addition, CFDT still holds 3Q percent which centralized all cotton activities, taking on the cooper- of SODECOTON's capital, as well as equity positions in atives' functions when cooperative unions were dissolved in CICAM and La Compagnie Cotoniere, the agent in charge of 1976 and replaced by Ujamaa village cooperatives. In 1984, most cotton exports, which acts as sales agent for SODE- TCA handed over some ginneries and oil mills in the COTON. Despite its obvious importance, however, the Western Growing Area to the regional farmers' corporation, presence of CFDT itself need not be crucial. Rather, our whose shareholders were the Ujamaa villagers. Finally, in analysis indicates that CFDT's professional approach is the 1985. TCA was dissolved and the Tanzania Cotton Marketing important factor. Board (TCMB) was created to take its place. At that time, The example of Kenya contrasts strongly with that of the cooperatives were reconstituted and put in charge of all Cameroon. The Cotton Lint and Seed Marketing Board ginneries and primary marketing functions. In each phase, (CLSMB) was created in 1955 to serve as the Kenyan the institutions in charge of marketing encountered finan- government's main instrument for intervention in the cotton cial difficulties, leading to delays in payments to producers, sector, but it is much less vertically integrated than while the separation between credit, input distribution, and SODECOTON. The Ministry of Agriculture (MOA) undertakes farm gate purchasing resulted in extemely low repayment most activities upstream of cotton production and supplies rates, which only exacerbated financial difficulties. CLSMB with seeds, insecticides, sprayers, and land prepa- More recently, in 1986/87, excellent weather and struc- ration services, which are then distributed to producers tural adjustment policy changes, including an improved either free (seeds) or on credit plus a fee of 12 percent of producer price (especially relative to maize), led to a purchase cost (all other inputs). CLSMB is not directly bumper crop, showing that producers are capable of involved in food crop production or in any other aspect of delivering a strong supply response. However, institutional rural development outside of cotton. Because CLSMB is factors constrained the management of the resulting sup- formally responsible for the purchase, transport, and plies. TCMB purchased less than two-thirds of the crop due processing of seed cotton, it appears to many reviewers to , to lack of funds, transportation problems, and weaknesses be an integrated operation. In practice, however, CLSMB of the cooperatives. Even then, the purchases far out- has delegated many of these functions to cooperative weighed the ginning capacity of the aging mills, causing unions and private firms. Nine cooperatives or privately delays of several months in the payments for cotton owned ginneries coexist with CLSMB's own five ginneries. In purchased. Tanzania's example shows not only that pro- recent years, CLSMB has used private transporters to ducer prices alone are inadequate to ensure sustained supplement its own fleet. It is not involved in cotton oil growth in production, but also that an exclusively price- refining, which is entirely in the hands of the private sector. based policy is counterproductive in the absence of CLSMB helps farmers to purchase inputs by providing complementary actions with regard to marketing, process- them with interest-free credit through MOA and the ing, and transport. Without the latter, producers lose faith cooperative societies (though the latter are notoriously in the system and become less willing to respond to corrupt and farmers complain of late and nonpayment for relative price changes. their cotton). It also extends interest-free credit to cooper- Nigeria is an interesting case because it has recently ative unions and societies to provide them with working changed institutional arrangements. Until 1986, the National capital for their ginning operations. Loan recovery has been Cotton Board (NCB) had a strategic position in the sector, a serious problem. As early as July 1980, CLSMB had the not unlike that of CLSMB in Kenya. Although the Ministry of equivalent of US$5 million in outstanding loans, evenly Agriculture was put in charge of extension services to cotton distributed between farmers and ginneries. Debts have producers, the NCB was to distribute seeds and other increased because of the need to finance below-cost sales inputs, fix producer prices after consultation with the of cotton lint to the textile industry, where domestic lint industry, and license agents for its purchases. These competes against cheaper imports. While this is necessary Licensed Buying Agents (LBAs) were supposed to buy the to ensure a steady flow of lint for the gins, it is a function cotton at fixed prices, and then ensure proper bagging, for which CLSMB is not equipped financially. The situation storage, and transport to the gin. In December of 1986, worsened in the early 1980s, before the world cotton market however, NCB was disbanded along with all the other crop collapsed in early 1986. (The domino effect on cotton marketing boards, amidst widespread allegations of corrup- productivity of the marketing parastatals' poor financial tion, inefficiency and collusion with the LBAs on the prices 25 farmers actually received. One report described weak- francophone Africa, many governments will object strongly nesses in the seed distribution system, with late distribu- and will opt instead for joint European public sector tion of seeds of increasingly poor quality. It cited numerous partnerships. Donors, on the other hand, have argued that allegations of farmers receiving less-than-official prices, late privatization, albeit involving minorities, is likely to break payments or no payments at all, of collusion between up the monopoly of CFDT, bring in more appropriate buying agents, false weighing, and illegal payments for processing technology, and will perhaps result in greater quality grading (de Matharel 1987). Farmers responded by indigenization over time through joint Asian-African owner- turning to alternative crops (maize) or to late plantings with ship of business. low input intensity. The basis of criticism of the cotton subsector in franco- Nigeria's cotton subsector is now more difficult to phone countries, and what is sometimes called the "CFDT summarize because of the complex web of institutions system," is thus important to understand. It includes (i) involved. Marketing has been taken over by a variety of CFDT's high financial costs associated with monopoly agents, including the old LBAs, state government-owned operations and unnecessarily advanced technology in companies, private merchants, and the textile milis them- cotton processing; (ii) an alleged bias against foodcrops, selves. A National Consultative Committee on Cotton and (iii) the high environmental costs that were discussed Production (NCCCP), created in 1986, is supposed to earlier. The following discussion will focus on the first two coordinate policy for the subsector. Prices are now set by criticisms. the Cotton Marketers' Association of Nigeria (COTMAN), A recent French government study on African cotton founded in 1987 and comprising about 100 members from concluded that average costs per kilo of cotton lint the textile industry and from companies and institutions produced were no higher in francophone countries than in involved in cotton marketing. However, COTMAN's authority other African and non-African countries.6 While this is to fix producer prices has been a matter of dispute within correct for production costs at the farm level, the cost the government. It is certainly not clear that COTMAN can estimate does not include the cost of expatriates in in fact guarantee the prices that farmers actually receive; management positions. Their number and the emphasis on these now appear to be determined by supply and extension are bound to be expensive. The fact that CFDT demand, with the farm gate price determined by the has not yet produced a cotton management system that is degree of monopsony power held by the buying agents. A completely and successfully Africanized may demonstrate report by Alikhan describes the emergence of a greater the importance of professionalism in the cotton industry, collaboration among the participants of the cotton industry combined with the political stake that producers develop in since these reforms. These are apparently yielding early a successful operation (Lele and Meyers 1986). Neverthe- positive results (see Annex 2). less, it is clear that the cost of CFDT's assistance to the local Nevertheless, the francophone countries appear to be cotton companies should be reduced, especially in coun- much more strongly committed to their cotton subsector tries such as Cameroon, where the CFDT experts are still than the anglophone countries are. They have made greater numerous. efforts to ensure that coordination among institutions is A true assessment of CFDT's costs would be extremely satisfactory and that the sector is adequately funded. difficult to make. Not only are managerial costs often The reasons for this apparent commitment, however, are ambiguous, but the value of positive externalities created more difficult to fathom. One clue may lie in the consistent by efficient institutions is almost impossible to estimate. presence of CFDT in one form or another in every franco- Benefits will vary among countries, but they may be phone country since independence. As noted earlier, the substantial. relationship of CFDT's revenues to the growth of cotton Cost comparisons are further complicated by the differ- exports from Africa has been a key to its own commitment ent environments in which cotton parastatals function. In to African cotton development. CFDT has supplied the Kenya, the relatively efficient ministerial structures allow cotton parastatals with technical assistance, a viable techni- the CLSMB to focus on cotton marketing alone, while in cal package, access to capital, and influence among donors Cameroon and Senegal, the parastatals have had to com- and decisionmakers. There is now much disagreement as to pensate for the shortcomings of central government admin- whether CFDT's operations should be parcelled out to istration by taking on costly rural development activities various other agencies-e.g., research and extension to the assigned to them by government mandate. ministries and provincial departments of agriculture, input These regional development activities make it difficult to supply and credit to cooperatives, and cotton processing to assess the cost effectiveness of CFDT's assistance to the private enterprises-and whether the cotton parastatals local parastatals in francophone Africa. Where a widening of should be maintained as integrated companies. It may be, the scope of operations has happened, as with CIDT in however, that CFDT's presence has brought about the C6te d'lvoire, SOCOTON in Togo, and SODEFITEX in stability and coherence needed for the development of a Senegal, the new range of operations covers activity from viable sector strategy, as well as the political influence to rural infrastructure to health and education. In the case of promote government commitment to the sector. SODEFITEX, for example, the broader regional mandate Another clue to the greater commitment of francophone encompasses: development and technical assistance for Africa to cotton may lie in the absence of a minority cereals (millet, sorghum, maize, and rice); construction of business community, such as the Asians of Kenya and hydro-agricultural facilities and technical assistance to Tanzania, antagonism to whom led to premature indigeni- irrigated perimeters; marketing and processing of paddy zation of the sector. Whether the proposed schemes to rice; marketing of maize; development of snack groundnuts; break up the parastatals would mean Africanization of the development of livestock (including veterinary and breed- cotton sector or its transfer from European to Asian hands ing); promotion of animal traction and providing "support" remains one of the most intriguing-though least openly to local blacksmiths; promotion of and assistance to village discussed-issues. To the extent that privatization means a associations (in literacy and managerial training); construc- larger role for Syrian and Lebanese businessmen in tion and maintenance of feeder roads and village water 26 supply; primary health care; and applied research and location-specificity of the complex problems it faces. monitoring-audit. A strict application of cost/benefit anal- In Great Britain, the Cotton Research Corporation (CRC), ysis to CFDT's assistance would need to take account of was in charge of promoting research.8 Despite limited some or all of these functions as well as of the non- funds, the CRC served effectively as the hub of a research quantifiable and indirect externalities generated by the network. It carried out excellent research, disseminating the cotton parastatals. results, financing and coordinating research projects, and The same issue of the cotton agency's relation to the assuring the long-term presence of research scientists in ministry is posed when evaluating the criticism that CFDT's Africa who gained experience over time. Indeed, much of approach has undermined food crop production. Although the growth of cotton in eastern and southern Africa in the SODECOTON is criticized for devoting most of its energy to 1950s and 1960s is directly attributable to CRC activity. In cotton, it should be remembered that it was established the early 1970s, CRC fielded some 31 scientists in anglo- precisely to address the complex problems of developing phone Africa. CRC's resources were generated by a cess on an industrial crop. Where other rural development institu- consumers of textiles in the United Kingdom. As the British tions are efficient, as in Kenya, institutions such as the Tea textile industry waned, however, so did resources for CRC, Development Authority are rarely criticized for focusing which was eventually disbanded in 1976. Although the narrowly on their own mandate, even though they might British government supported cotton research in East Africa usefully serve a broader set of producer needs (Lele, until 1982 on an ad hoc basis, so as to fill the void left by Christiansen, and Kadiresan 1989c). In any event, SODE- CRC, many of CRC's functions were not taken on by other COTON did develop an interest in promoting food crops, in institutions. Table 14 is eloquent on this point. By the early part because other institutions were ineffectual and in part 1980s, ten francophone countries were benefiting from the because SODECOTON realized that producers' interest in services of 40 expatriate scientists for cotton research, while growing cotton was weak as long as food security concerns the nine anglophone countries had only nine. were not met, the 1973-74 drought being the most dramatic In neither anglophone nor francophone countries has lesson to this effect. Neverthless, development of food much priority been given to building indigenous research crops has tended to pose problems for SODECOTON due capacity or developing top-quality African scientists to work to the lack of a market, and SODECOTON ended up losing on cotton. This has created particular problems in anglo- revenues when it supported sorghum prices. phone countries, now that CRC has withdrawn. In Nigeria, The cotton parastatals do seem to benefit from the best for example, the Institute of Agricultural Research (IAR) has resources at the expense of other crops. They tend to formulated an ambitious and expensive plan for cotton attract the most market-oriented and productive producers research, but the program has suffered from inadequate with the best land, along with the lion's share of agricultural human and capital resources.9 Improved varieties have inputs like credit and fertilizers. Consequently, the best- been developed, but they continue to exhibit ginning trained manpower want to work for them. Indeed, one of percentages of no more than 35, some 15 percent lower the important roles of the cotton parastatals has been to than that in neighboring francophone countries. In any develop an excellent cadre of trained professional Africans event, multiplication and distribution of improved seeds in much the same way that earlier tobacco schemes had has been haphazard, negating much of the research effort. done in Tanzania (Lele 1975). Over time, the absence of For example, the dissemination of improved seed has been services and well-trained manpower in other agencies can so disorganized that there are no records of the number lead to regional imbalances in research, extension, credit, and location of recipient farmers or of the varieties being and fertilizer access, with all the drawbacks that such disseminated, and few attempts have been made to imbalances entail. This is a cost of success, however, and maintain quality over several seasons. The story is no the solution is to pay greater attention to other elements of different in East Africa. rural development, not to undermine the cotton subsector The contrast with francophone Africa is evident in the by shifting responsibility for its development onto ineffec- area of single cropping. While intercropping with food crops tual ministries of agriculture. is a rational producer response to relative input scarcities Research and extension. A notable characteristic of and to an environment of risk and uncertainty, it also keeps African cotton is the fairly successful research systems cotton yields low because of the competition for moisture established by both Great Britain and France after World and (if the other crop is tall) for sunshine. In the franco- War 11. Some critics argue that francophone countries now phone countries, research and extension services have need to put all research and extension under the Ministry provided farmers with the seeds, fertilizers, mechanization, of Agriculture, as in the anglophone model.' After impres- and farming techniques to undertake single cropping. The sive performance in many countries in the 1950s and 1960s, more widespread use of farm mechanization, higher input however, the anglophone system of research and extension use, and consolidated plots has helped diminish the labor by and large failed to improve productivity during the 1970s scarcity problem, encouraging farmers to grow cotton alone or afterward. In France, the lnstitut de Recherches du Coton et or side by side with food crops, rather than intercropping it. des Textiles Exotiques (IRCT) was created in 1946 to coordinate In Kenya, however, the interface between research and French colonial research on cotton. Seed multiplication and extension has been so weak that intercropping is probably dissemination of results were greatly helped by close increasing, particularly in western Kenya. The majority of collaboration between IRCT and CFDT, which has continued farmers now intercrop with potatoes, maize, and/or beans. since independence, as well as by collaboration between Fertilizer is not used on rainfed cotton, and insecticide is the two French agencies and national cotton parastatals. used only by a minority of farmers. Land preparation is Varietal research has continued, leading, for example, to the predominantly by ox-plough, with a minority using rented development of the higher ginning outturns referred to tractors. This explains a tendency toward late plantings, as earlier. The IRCT, CFDT, and the related francophone farmers wait for the first rains to soften up the soil. national systems have successfully handled the require- Improved seeds have been developed by the Ministry of ments of the cotton crop, together with the enormous Agriculture, but they are simply dropped off at market 27 Table 14 Numbers of African and expatriate cotton researchers in anglophone and francophone Africa, 1984 Agronomy- Genetics- Crop Fiber Socio- Grand Social Science Plant Breeding Protection Technology economics Total Anglophone Scientist Africa African Expatriate African Expatriate African Expatriate African Expatriate African Expatriate African Expatriate Kenya 6 0 5 1 7 0 0 0 0 0 1 8 1 Malawi 1 0 2 0 3 0 0 0 0 0 6 0 Nigeria 4 0 3 0 4 0 2 0 1 0 14 0 Sudan 3 0 3 0 5 2 2 0 1 0 1 4 2 Swaziland 1 0 0 0 0 1 0 0 0 0 1 1 Tanzania 2 0 1 1 2 0 0 0 0 1 5 2 Uganda 3 0 4 0 4 0 0 0 0 0 11 0 Zambia 0 0 1 1 0 1 0 0 0 0 1 2 Zimbabwe 1 0 2 1 3 0 1 0 0 0 7 1 Total 21 0 21 4 28 4 5 0 2 1 77 9 Francophone Scientist Africa African Expatriate African Expatriate African Expatriate African Expatriate African Expatriate African Expatriate Benin 0 1 2 0 2 0 0 0 0 0 4 1 Burkina-Faso 2 1 1 1 1 0 0 0 1 1 5 3 Cameroon 1 1 0 2 1 2 0 0 0 1 2 6 CAR 1 1 1 1 2 2 0 0 0 1 4 5 Cote d'lvoire 1 2 1 2 1 3 1 1 0 1 4 9 Madagascar 2 1 0 0 1 0 0 0 0 0 3 1 Mali 3 2 2 1 3 1 0 0 0 0 8 4 Senegal 1 1 1 1 0 0 0 0 0 0 2 2 Chad 1 1 1 2 1 2 0 0 0 0 3 5 Togo 2 1 0 1 1 1 0 0 1 1 4 4 Total 14 12 9 11 13 11 1 1 2 5 39 40 Source: Ken Anthony. "Sub-Saharan Africa Agricultural Research Review, Cotton Research," unpublished paper, 1986. centers and do not reach many farmers. Input distribution payoffs associated with such an investment. ard credit have always suffered from uncertain and late Processing issues and the need for backward coordina- delivery, even before the system more or less collapsed in tion. The importance of local processing is also fairly 1986 (see below). Thus, cultivation techniques remain unique to cotton as a cash crop. Ginning, oil refining, and largely rudimentary, and yields extremely low. As noted of course, textile milling, all provide cotton with several earlier, cotton may simply not be profitable at such low specific constraints. First, each step in the processing chain levels of technology. requires financing. With uncertain or low levels of funding, Another area where the results of research have affected any link in that chain can be broken, leading to a vicious francophone and anglophone countries differently is circle of late payments and debt. Second, variation in related to the focus of basic research. While the British- output levels can have serious repercussions for the costs inspired cotton research systems focused mainly on spin- of ginning, refining, and milling, and operating above or ning properties (to obtain longer and higher quality fibers, below capacity can lead to the same vicious circles. These as prescribed by the Shirley Institute for industrial research, difficulties are not unique to cotton among cash crops, but in Manchester), the French insisted on obtaining higher their potential impact is serious, because of (i) the wide ginning yields, which would enable CFDT to earn higher output variation discussed earlier, and (ii) the smaller profit profits without having to reduce procurement prices'of margin with the subsector, owing to cotton's relatively low seed cotton. There may be a limit to this line of research, value. however. The focus on increasing the ginning ratio has led Each country in this sample has at one time or another breeders to introduce ever smaller seeds. At some point, witnessed important losses because of output variation at these may spoil the quality of the fiber because the small some stage in the processing chain. This has been a major seeds pass through with the lint during ginning. problem in a country such as Kenya. As a member of the In order to alleviate some of the upstream obstacles that East African Economic Community, Kenya exported its are causing relatively low production yields and ginning better quality cotton lint and imported lower quality lint ratios in anglophone Africa, it is possible that the Consul- duty free from Uganda and Tanzania for its textile mills. tative Group on International Agricultural Research (CGIAR), With the closing of the border in 1977 and the introduction which has been successfully supporting food crops, could of a 30 percent import tax, domestic mills began to now build on CFDT's experience and serve as an appropri- purchase the Kenyan cotton, driving lint exports down to ate vehicle for dissemination of research results on cotton. insignificant amounts. The lint is appreciated for its high, if Through CGIAR, instead of a smaller and more localized variable, quality and is generally preferred to foreign lints entity such as CFDT, benefits of research would reach a of the same type, although mill managers complain that wider audience. Private research funding in anglophone Kenyan output is dirty, necessitating a costly cleaning Africa is hardly foreseeable, given the size of the necessary process that results in a weight loss of 5 to 20 percent. investment, the high risks, and the long maturation of 28 Weighing the Evidence Although the impact of price incentives on cotton produc- interventions-ranging from upstream research to down- tion is important for determining short-run shifts between stream marketing assistance-has helped alleviate most food and export crops, or labor allocations between cotton constraints and risks to the cotton sectors of its recipient and other crops (including wage employment), this paper countries. But a monopsony-monopoly of the CFDT type has demonstrated that prices must not be considered in need not be the only structure of the cotton industry. The isolation from a range of nonprice factors that affect evolution of CFDT itself from a completely sealed research- production. Instead, the two sets of factors must be extension-production-marketing organization into more addressed jointly in explaining the much greater relative open national organizations that allow entry to other actors success of the cotton subsector in francophone Africa. is certainly a change in the right direction. In Nigeria, where SODEFITEX and SODECOTON have been relatively effec- the cotton industry was at first completely decentralized, a tive agents of government policy in Senegal and Cameroon, new mixture of backward, albeit limited, integration and of respectively. Producers have not only consistently received an outgrower program also seems to be a promising the full official price for their output, but they have also development.'0 benefited from extension and modern inputs that helped The francophone experience conveys another important make cotton production attractive. In the other (anglo- lesson-that efficient credit is an important explanation for phone) countries of the sample, price and input policies the relatively high level of adoption of modern technology have been implemented in ways that seriously undermined in cotton husbandry. Because cotton is bound to have a their potential impact. As a consequence, the cotton sealed marketing chain-inasmuch as there are no leakages subsectors in these countries lag behind in terms of a wide in the form of home-consumption or undue on-farm variety of indicators. This is particularly striking in Kenya, storage-the cotton companies were able to provide which has had a much more impressive record of agricul- farmers with credit (in cash or kind) and recover it by tural modernization in areas other than cotton than most of deduction at source from the proceeds of the farmers' the francophone producers, suggesting that cotton poses sales. The monopsony buyer's position of CFDT also unique challenges among cash crops. The question is helped, although that position is being lost where the whether a low input/low output aproach, as followed in cotton companies are carrying a broader regional develop- anglophone Africa-while more efficient in the sense of ment mandate, including the servicing of such crops as rice lower DRCs than one which relies on expensive imported and millet, which can be consumed and whose marketing is inputs-is likely to result in increased cotton production in undertaken by other agencies. circumstances of growing internal demand for food crops. Will the CFDT-assisted cotton companies continue to be The key to the relative success of cotton development in as successful without CFDT's (high-cost) support? As noted francophone Africa has been the ability of the industry to earlier, countries such as C6te d'lvoire and Mali, with the maintain effective coordination among the different layers fewest CFDT experts, are precisely those where cotton of participants. As the previous discussion has demon- yields are the highest. It appears, therefore, that the strated, a vertically integrated research-production-market- transfer of technology to indigenous capacity, at least ing apparatus may be necessary in order to promote judging from these countries, can be done effectively and production of a commodity in a high risk/low return that CFDT can usefully step up the pace of indigenization. environment. The vertical (backward) integration which has Long-term commitment to the process and the establish- characterized the CFDT-supported cotton companies has ment of it as the explicit goal is, however, critical. Past helped to circumvent the supply risk, especially in semi- donor assistance (e.g., the World Bank's) to cotton did not arid low-income regions where the primary (legitimate) have such an explicit goal, even though it supported CFDT- concern of the producer is for food security. related institutions in Africa. By and large, the French CFDT, through its various 29 Recommended Ins'itutdonal Reforms The African state's involvement in agricultural development in recent years. The parastatal's losses include outlays of has been called into question by a number of donors, CFAF 6 billion for rural development activities undertaken including the World Bank. This has led to calls for reform of on behalf of the government, as well as an estimated CFAF the public institutions involved in the agricultural sector. In 2.6 billion spent on input subsidies.'2 These numbers imply the cotton subsector, these calls have become more that a significant part of SODECOTON's recent deficit insistent since the downturn in world prices for cotton in results from the particular status it has acquired as 1986, which created financial difficulties for a number of northern Cameroon's most (if not only) efficient public cotton parastatals. development institution. SODECOTON was clearly a viable Comprehensive reforms of cotton institutions are taking enterprise when cotton markets were healthy, despite lax place, or being considered, in all MADIA countries except management, generous producer prices, and a variety of Malawi. In Nigeria, the NCB was abolished in 1986, while in expensive developmental responsibilities, which cotton Kenya a program has been formulated to divest the boards in anglophone countries do not bear. ginneries from the CLSMB.'I In Cameroon and Senegal, it SODECOTON's overall efforts have provided valuable-if has been suggested that the role of SODECOTON and difficult to quantify-externalities to rural development in SODEFITEX be limited to seed cotton marketing, that the region as a whole. These developmental functions upstream functions (research and extension) progressively should not be undermined, at least not until the ministerial revert to the Ministry of Agriculture, while downstream structures become strong enough to take them over. functions (input supply and credit) are given to the private Meanwhile, their costs must clearly be taken into account in sector. This section evaluates the problems of the cotton any cost/benefit evaluation of SODECOTON's performance. subsector and the desirability of institutional reforms in the One solution being considered by donors-the transfer of context of the conclusions of the preceding sections. Given functions to ministries of agriculture under a more general- the importance of institutional effectiveness in explaining ized approach to research and extension (in contrast to a performance, what advice can be given to policymakers commodity-based approach)-presents serious problems about the best solutions for cotton's present difficulties? in francophone Africa because agriculture ministries and Francophone Countries their decentralized provincial services are especially weak and will take time to develop. In Cameroon, SODECOTON's financial problems did not There is, however, a combination of reforms that could emerge until 1986 and the collapse of world cotton prices. quickly bring SODECOTON's deficits down to a reasonable Since then, SODECOTON has been buying farmers' cotton size, allowing time for the cotton markets to rebound. Such at a loss, not least because of the strength of the CFA franc. reforms include: (i) reducing transportation and market In the middle of 1986, for example, when world cotton costs, lii) freezing or cutting producer prices, and (iii) markets bottomed out, the average equivalent producer reducing the number of CFDT experts and delegating price (CFAF 352 per kilo) was greater than the c.i.f. price (Le responsibility for services increasingly to producer groups. Havre), which fell to CFAF 266. As a result, SODECOTON ran Much the same could probably be said of Senegal and an operational deficit of roughly CFAF 20 billion (approxi- SODEFITEX, where improved management and, especially, mately US$40 million) in 1986, CFAF 13 billion in 1987, and minor improvements in the distribution and use of inputs CFAF 8 billion in 1988. For the ten previous years, the and prices would probably restore profitability. More annual deficit had averaged only about CFAF 150 million, grassroot-oriented cooperatives such as those beginning to and there had even been small profits in 1983/84 and be fostered in Mali may take over the commercial functions 1984/85. This record of relative success undoubtedly led if they are allowed enough time to develop. SODECOTON to relax its management style, and encour- More fundamental long-term measures might include a aged the government to continue raising producer prices higher degree of domestic industrial transformation, with (by 1985, the highest in Africa). Both SODECOTON and the the development of a national textile industry and greater Cameroon government were unprepared for the downturn training of Cameroonians to assume the functions of CFDT in 1986. experts. Nevertheless, the current highly integrated To put these numbers in some perspective, however, the approach has been effective in disseminating a viable cotton subsector's role in the northern economy should be technological package for cotton. While costs can and borne in mind. In 1985/86, cotton provided the government should be cut in the short run, it seems very unlikely that with some CFAF 2.1 billion in public revenue. Various taxes the most realistic international price scenarios will be able on SODECOTON itself (excluding income taxes on its to cover the overhead costs involved in making African employees) have amounted to an additional CFAF I billion producers viable. Proposals to take away SODECOTON's 30 research and extension functions and transfer them to the chaos is a necessary cost of liberalization, but that market Ministry of Agriculture-without benefit of CFDT's regional forces will eventually organize the subsector and provide research network and without simultaneously establishing a the necessary incentives for cotton farmers. In addition, a broader regional or international system for cotton costly and ineffectual, if not counterproductive, public research-may well result in the collapse of the cotton sector intervention will have been eliminated. Even so, industry in francophone Africa. Although every effort needs farmers' incentives to grow cotton will continue to be to be made to avoid duplication of functions with line constrained by the very low levels of productivity currently ministries, it might be better to give CFDT ten years to attainable without a more sophisticated technological indigenize its operations, including the development of package. The dissemination of this package, which is a pure African expertise and incentives related directly to the public good, will continue to imply the need for effective volume of exports. Similarly, privatization of input distribu- public institutions for agricultural research and extension tion in order to cut costs should be considered only with (Yayock and Kumar 1988). the greatest caution, due to the need to link distribution More generally, one of the lessons from francophone with credit and output marketing. Africa is that cotton booms followed the development of a viable technological package for intensive cultivation. As Anglophone Countries long as farmers could not expect more than 250-400 kilos of In contrast to Cameroon and Senegal, cotton's problems in seed cotton per hectare, their loyalty to cotton was weak, Nigeria, Tanzania, and Kenya preceded the downturn in and completely determined by food security concerns. world prices and are more clearly linked to the sector's Intensification alleviated these concerns not only by greatly institutional deficiencies. Thus, CLSMB's financial problems increasing the farmer's cash income, but also (and perhaps in Kenya have their origin in several factors. First, CLSMB more crucially) by freeing up labor for food crop cultivation. was seriously underfunded and financed most of the This lesson was confirmed in Tanzania: the rapid growth of subsector's expansion during the 1970s by borrowing. The production under extensive cultivation from 1950-70 is cost of servicing the resulting debt amounted to around proving unsustainable, and the cotton industry has faltered Kenyan Shs. 118 million (US$7.4 million) between 1979 and as relative prices have become less favorable. Most of all, 1985. Second, CLSMB has had a very poor loan recovery without intensification the growing population pressure will rate on its own lending to both farmers and cooperatives. continue to make greater demands on resources for food The need to compensate for unrecoverable loans cost crop production, reducing exports and increasing cotton CLSMB some Shs. 94 million in the same period. By the imports. end of 1985, CLSMB was no longer able to secure loans for Political problems may continue to undermine effective its operations, and producers were left unpaid for the reform, as is evident in Tanzania and Kenya, but donors previous year's crop. have devoted little attention to their implications. For CLSMB thus appears to have been the victim of internal instance, in Tanzania it is not clear that the CCM Party will mismanagement. Furthermore, neither CLSMB nor the be willing to relinquish control of the cotton industry to the Kenyan cooperatives have been responsive to producer cotton producers or to the producer cooperatives through interests. Indeed, the way Tanzania has been able to genuine political decentralization. In Kenya, where the undermine effective collective action by its cooperatives ginneries were traditionally owned by Asians before they suggests that farmer-oriented grassroots cooperatives may were nationalized and given to cooperatives, privatization of not develop in other countries as early as donors have the gins is now unlikely. Turning gins operated by the tended to assume. At the same time, CLSMB's ability to CLSMB over to the cooperatives (which had lost them to the promote the cotton subsector has largely depended on a Board because they mismanaged them) hardly seems a number of other institutional actors, such as the Ministry of desirable option unless more professional cooperatives are Agriculture and cooperative societies, over which it has little developed that are also more directly accountable to the control. And its developmental functions are less highly cotton producers. This will, however, take a consistent evolved because of the less integrated nature of the cotton approach by donors and governments over a long period of subsector that characterizes anglophone Africa. It is difficult, time-and the experience of the last two decades does not therefore, to blame CLSMB's problems entirely on its own provide much basis for optimism about the outcome, financial woes, let alone to blame it for the poor record of unless there is simultaneously a greater decentralization of cotton production, even if it has operated in a country with the cotton industry to the producers, combined with more a relatively good macro and institutional environment. technical inputs for its long-term development. In Nigeria, the optimistic scenario is that the current .. Annex 1. Theoretical Derivation of the Estimated Cotton Supply Models pe purpose of this annex is to provide the theoretical such investment outlays. One would need a hedonic ustification for the statistical models of cotton supply weighting scheme to account for the quality and relevance .iscussed in the text. of these investment outlays on production increases. Third, The models are derived from the conventional supply the long time lag before research results become effectively :e.havor based on profit maximization applied to cotton translated into innovative practices varies between 8 and 12 producers in the MADIA countries. The theoretical deriva- years, at least by those African farmers surveyed in Eicher "ion need not be repeated, since any standard microeco- (Eicher 1985). Thus, even assuming that the size of the niomic textbook gives a thorough treatment. However, one expenditures on R&D and extension could correctly relate must recognize that the theoretical derivation of the supply the degree of technological advance, the problem of timing Junction may not be a straightforward textbook example the effects in the interpretation of the regression results because the francophone cotton companies are multiple remains. Using twenty-year observations-which would be output/multiple input firms. The outputs are heterogenous, a blessing given the status of the data on these countries- ranging from tangible commodities such as cotton and rice for a regression with an 8 or 12-year lag and at least two to intangible public goods such as health and literacy. The explanatory variables, substantially reduces the number of f-lowing discussion will, nevertheless, focus on the stan- degrees of freedom. Furthermore, while the stated direct dard single output, which is assumed to be a composite measures are less problematic in aggregate supply estima- good. tions, the fact that the expenditures are often not crop- The model of cotton supply comprises two behavioral specific poses an additional problem in single crop supply equations (area and yield) and one identity (production). estimations. The farmer's output reaction is decomposed structurally Our knowledge of the cotton industries of these countries :'to: suggests that the shift of the supply schedules during the (i) an area response, which depends upon relative prices period under investigation was the result of nonprice and land availability; and factors. Since prices do not shift the supply schedule (at (ii) a yield response, which is determined by input use least directly), and very little (price-) induced technological (which in turn is determined by input price), the quality of change in cotton husbandry in the selected countries is services provided to farmers, and weather conditions. known to us, it would be tempting to attribute the supply Output is then determined as an identity expressing shift largely to the effects of nonprice factors. A time trend quantity as the product of yield per unit of land and area is a conventional and simple measure of technological cultivated. change, which cuts through the intricacies of the problems The structural system above can also be collapsed into a mentioned above. The failure to include a time trend in the single reduced form equation featuring output and input estimation of supply-response equations, when such tech- prices and some exogenous shifters as explanatory varia- nological changes have occurred, has led many researchers bles. While explicit measures of the price variables are not to find negative supply elasticities and to recourse to the difficult to find, those of the nonprice shifters pose target income hypothesis to justify their findings. While the conceptual and statistical problems. target income hypothesis may be true, its application to The relevance of using a time trend as a proxy for the supply-response analysis, which leads to a backward effects of nonprice factors on the development of the bending supply curve, should be attempted only as a last cotton industries in the countries that were studied needs resort. to be explained. The choice of the trend variable as a proxy In summary, the choice of the time-trend variable is rather than a (more conventional) direct measure of the motivated by the fact that technological progress and the nonprice effects-such as expenditures on R&D, extension, deterioration (or improvement) of institutional support did and infrastructure-in the specific case of the MADIA occur in the cotton industries of the selected African countries was dictated by the following considerations. countries. Given that this technological and institutional First, it is clear from the graphs in Figure 6 that cotton support did not stay static, as the paper shows, it seemed production increases in Senegal and Cameroon were the most appropriate to capture the effect via the time trend. results of shifts of, rather than price-induced moves >-long, The time trend is not the nonprice factors. Rather, it is a the supply schedule. In these two countries, the relative proxy for their effects (technological and institutional) on prices have been constant or even declining over 1961-85, cotton production in the selected countries. wvhereas production has increased at least sevenfold. The estimation results of regression models based on the Second, using such explanatory variables as expenditures above discussion follows: on R&D and extension says nothing about the quality of 32 Annex Table 1 Seed cotton price elasticities in the MADIA countries Price Elasticities:a Nonprice Effect: Corrected Durbar Timetrend R2 Watsor? Cotton Maize* Coeff T-statistic Area Cameroon 1.84 -0.84 0.27 1.41 Senegal NS NS 0.92 3.11 0.16 1.46 Nigeria 0.23 NS 0.81 1.91 Kenya 0.67 NS 0.91 2.05 Malawi 0.54 NS NS NS 0.43 1.58 Tanzania 0.13 0.13 0.72 2.29 Production: Cameroon NS NS 1.25 2.34 0.76 1.71 Senegal NS NS 0.68 2.23 0.12 1.94 Nigeria 2.02 NS -2.11 -2.59 0.83 1.24 Kenya 2.06 -1.13 0.53 6.04 0.70 1.29 Malawi NS NS NS NS 0.18 1.96 Tanzania 0.60 -0.35 0.66 1.83 Yield: Cameroon NS NS 68.45 5.21 0.93 1.33 Senegal 1.21 -1.33 0.52 2.10 Nigeria 2.30 NS -2.88 -3.24 0.71 1.67 Kenya 1.60 NS 0.60 2.10 Malawi -2.07 NS 16.97 4.68 0.43 1.12 Tanzania NS -0.4 -15.26 -2.57 0.53 2. 07 Note: a The t-ratios for the reported price elasticities are all larger than two. NS Not Significant *in the case of Senegal groundnut data was used. Source: Lele, Gbetibouo, and Mukherjee Forthcoming. Annex 2.' Nigerian Cotton Industry: Latest Developments Some recent developments in Nigeria which may be worth ing to get involved in large-scale farming. There seems tc highlighting in this paper are: the internal nature of the have been some reversal of this thrust, however, a2c cotton subsector, the government policy initiative of back- industrial concerns now seem to be moving toward t1he ward integration and self-reliance under the Structural creation of nucleus farms which provide the necessary basc Adjustment Program, the private sector (textile industry) from which to support a large-scale outgrower program. response and its early positive experiences emerging in terms of increased yields, a changing pattern of cotton Increased Yields by Supplyig Inputs production, accessibility of the farmers to rural credit (and One of the largest textile manufacturers in the country na hence inputs), and mechanization. These initiatives, established a nucleus farm of about 2,000 hectares. anr. together with the T&V Extension System, are making a dent indicates that it has achieved yields of 2.0 tons per hectare in closing the large gap between demand and supply. This, however, is a small part of the program, since thc manufacturer relies heavily on a large number of sinali Policy Initiative and Response farmers for the bulk of its requirements. Under the out- The observations that the report makes on Nigeria's cotton grower scheme, provision is made for: land to selectec subsector are correct. However, the cotton subsector scene farmers, land-preparation assistance in the form of tractor is changing fast in Nigeria. The government's policy of hire services (at cost to the farmer), good planting material/ backward integration by industries for local sourcing of raw seed, fertilizer, chemicals, cash loans, extension services materials is paying off. The government is also relying more and marketing linkages of buying back the crop at C on private sector initiatives. In the cotton subsector, the minimum guaranteed price or the market price, whicheve, private sector (textile manufacturers) response has been is higher. Land allocated to the farmer is I to 2 hectare, anc encouraging, and it has initiated programs which could form the outgrower farmer cultivates it predominantly by usins, an alternative or complement to measures taken under the his/her family labor. World Bank-supported Agricultural Development Projects Positive Experiences (ADPs) in making possible the revival of the subsector in Nigeria. While these experiences are still in the infancy Some key interesting developments have been reported bc stage, they are worth noticing. the textile manufacturer since the inception of the scheme The textile industry intially responded to the govern- (i) initially, loans (80 percent of it given to the outgrowe ment's policy directives on backward integration by begin- farmer in kind, the remaining 20 percent in cash) were provided by the manufacturer; now, the nucleus farn This annex was graciously provided by Mohsin Alikhan. project entity refers/certifies outgrower farmers to tNh 3,: Nigerian Agricultural and Cooperative Bank, a public sector Notes undertaking which extends loans. Its repayment record, according to the manufacturer is about 96 percent; (ii) the 1. See Marcel Roupsard, Nord Cameroun: Ouverture et Developpement, manufacturer indicates that outgrower farm yields have (Paris, 1987) pp. 391-418 for an exhaustive discussion. Ambiguous increased from about 250 kilos per hectare to 800 kilos per evidence suggests a secular decline in precipitation levels all over hectare in just over 3 years; and (iii) last, but more northern Cameroon, which may have speeded up this evolution of significantly, the farmr r s hs on pproduction. For a discussion, see Alain Beauvilain, "Remarques significantly, the farmer retains his own piece of land, on sur la situation pluviometrique actuelle au Nord du Cameroun," which he grows whatever he wishes (mostly mixed farming), Revue de Geographie du Cameroun, 5 (no. 1):47-62. thus not affecting his traditional security interest. On the 2. The resumption of production in these provinces was partly land provided by the manufacturer, the outgrower only the result of an innovative practice of extending the cultivation of crops cotton. Given the high price of cotton (since it is cotton over two seasons. The crop is sown in the short rainy highly protected through tariffs), the outgrower farmer season (October/November) and is let to mature during the enjoys a very favorable return for his efforts, currently longer rainy season (March/May). earning a minimum price of N 2,000 per metric ton for 3. Full discussion of the magnitude of these elasticities is Grade I cotton and N 1,800 per ton for Grade II. Indepen- provided in Lele, Gbetibouo, and Mukherjee, forthcoming MADIA dent reports indicate that farmers, contrary to previous study. reports of complaints, are now eager and willing to join the 4. The exact cost of CFDT experts could not be found by the outgrower scheme. The manufacturer indicates that farmer authors. Salary estimates from secondary sources and verbal demand to join outgrower schemes is higher than it can accounts vary from $US50,000 per expert to $US150,000 including meet. all benefits. The range is so wide that we prefer not to attempt a quantitative estimate of the DRCs. Demand-Supply Gap 5. In the West African Sahel, for example, cotton is typically the The manufacturer plans to increase areas cultivated from main source of agricultural income in the interior of the country, 1,000 hectares in 1988/89 to 25,000 hectares by 1992/93, where per capita incomes are a third to a quarter of the levels in involving some 12,500 participating farmers. He has already coastal areas. opened a ginning mill and is in the process of installing an 6. See Ministere de la Cooperation, RLpublique Francaise, "Le oil mill. Based on the success of this textile manufacturer Coton en Afrique de l'Ouest et du Centre: situation et perspec- another large textile manufacturer has started work on a tives" (Paris, 1987), pp.74-86. The study estimates, for example, similar scheme in the same state. that the average cost per kilo of fibre varies from CFAF 819 in These two schemes are under implementation in Gon- Cameroon to CFAF 506 in Mali, CFAF 528 in Burkina Faso and These t sh ea ud ipCFAF 629 in the C6te d'lvoire. Cameroon's cost structure is among gola State, where cotton production has increased from a the highest. mere 750 tons of seed cotton in 1985/86 to 8,500 tons in 7g See for example, Claude Freud, Quelle Cooperation? (Paris: 1987/88, making it the fifth-largest cotton producer in the Editions Karthala, 1988), and Bonnie Campbell, "Inside the country next to Katsina, Kaduna, Bauchi, and Sokoto states. Miracle: Cotton in the Ivory Coast," in The Politics of Agriculture in Since the gap between demand and supply of cotton is Tropical Africa, ed. Jonathan Barker (London: Sage Publications, large (ranging roughly over 50 percent), the initiatives by the 1984). private sector described above show potential for success. 8. This section draws heavily from Ken Anthony, "Sub-Saharan Some caution is necessary to guard the captive farmer from Africa Agricultural Research Review: Cotton Research," unpub- the monopolistic/oligopolistic nature of these schemes. lished paper, September 1986; Fred E.M. Giliham, "A Review of Until now, however, since the cotton demand-supply gap is Cotton Production Research in Nigeria, Eastern and Southern large, the schemes are being implemented successfully, Africa," consultant report for the World Bank, August 1986; and increasing farmers' incomes and making a positive dent in Elton Thigpen, IECCM, the World Bank (personal communication the rural areas, saving the country valuable foreign with one coauthor). exchange through import substitution, and at the same 9. The program is summarized in Gillham, pp.35-38. time providing essential raw materials to textile mills, thus 10. Comments by M. Alikhan, World Bank Resident Mission, increasing their underutilized capacity. With the abolition of Nigeria, February 21, 1989. the Commodity Boards, a relatively close-to-free-market 11. A more detailed account of the recent development in the price system for cotton is currently working satisfactorily in Nigerian cotton industry is related by Alikhan in Annex 2. Nigeria. 12. SODECOTON's rural development costs in 1987/88-after subsidy costs-were CFAF 3.6 billion. This figure is to be contrasted to 5.8 billion spent in 1985/86. T & V Extension System and Research It should be mentioned that efforts have been made in Nigeria to reorganize extension and training under a single unified extension system, using the T&V system and the positive gains of research-extension linkages achieved through monthly technology review meetings. It would not be advisable to disturb the T&V system, which is being implemented vigorously. The private sector initiatives described in the earlier paragraph could play a significant complementary role to the public sector initiatives carried out under the ADP system in an effort to revive cotton production in Nigeria. 34 Bibliography Anthony, Ken. 1986. "Sub-Saharan Africa Agricultural Research Freud, Claude. 1988. Quelle Cooperation? Paris: Editions Karthala. Review: Cotton Research." Unpublished paper. Gagne-Gervais, Carole. 1984. "Cameroon: The Cash Crop Sector: Anthony, Ken. 1985. "U.K. Agricultural Research Aid to Kenya, Its Performance and Future Development Possibilities." Tanzania and Malawi." MADIA Study, World Bank. Washing- Yaounde, Cameroon: USAID. ton, D.C.: World Bank. Gautrand, Jacques. 1988. "Le coton africain ne veut pas mourir." Askari, Hossein and John Thomas Cummings. 1976. Agricultural Jeune Afrique Economie, no. 105. Supply Response-A Survey of Econometric Evidence. New York: Gillham, Fred E.M. 1987. "A Review of Cotton Production Research Praeger Publishers. in Nigeria, Eastern and Southern Africa." Consultant Report Balcet, lean-Claude and Wilfred Candler. 1981. Farm Technology prepared for the World Bank. Washington, D.C.: World Bank. Adoption in Northern Nigeria. Mimeograph. Washington, D.C.: Government of Tanzania/Government of the Netherlands. 1986. World Bank, West Africa Projects Department. "Plan of Operations. Assistance to the Cotton Sector in Baris, P., Claude Freud and J. Zaslavski. 1987. "La politique agricole WCGA Including Support for Pamba Engineering Ltd." Dar- du Cameroun de l'independence a nos jours." MADIA es-Salaam, Tanzania: Government of Tanzania. Study, World Bank. Washington, D.C.: World Bank. Hanak, Ellen. 1988. "The Politics of Agricultural Policy in Tanzania." Bates, Robert. 1981. Markets and States in Tropical Africa. Berkeley: MADIA Study, World Bank. Washington, D.C.: World Bank. University of California Press. Hayami, Y and VW. Ruttan. 1985. Agricultural Developmenkt An Beauvilain, Alain. 1985. "Remarques sur la situation pluvio- International Perspective. Baltimore: Johns Hopkins University metrique actuelle au Nord du Cameroun." Revue de Geogra- Press. phie du Cameroun 5 (no. 11:47-62. Howell, John, Adrian Hewitt, and Ken Anthony. 1987. "U.K. Beenhakker, H.L. and N. Bruzelius. 1984. "Transport and Marketing Agricultural Aid to Kenya, Tanzania, and Malawi." MADIA of Agricultural Products in Tanzania." Draft of report Study, World Bank. Washington, D.C.: World Bank. prepared for the Transportation Department, World Bank. International Cotton Advisory Committee (ICACI. no date. "Survey Washington, D.C.: World Bank. of Cost of Production of Raw Cotton." Washington, D.C.: Caisse Centrale de Cooperation Economique. 1988. "Proposition ICAC. pour un plan de redressement de le SODECOTON," Volume jammeh, Sidi. 1987. "State Intervention in Agricultural Pricing and I and 2, Paris: CCCE. Marketing in Senegal." Ph.D. dissertation, Johns Hopkins Campbell, Bonnie. 1984. "Inside the Miracle: Cotton in the Ivory University. Coast." In Politics of Agriculture in Tropical Africa. Ed. Jonathan and Chandra Ranade. 1987. "Agricultural Development and Barker. London: Sage Publications. Foreign Assistance to Senegal: A Review of the World Bank's Carr, Stephen. 1989. Back-to-Office Report. Internal World Bank Experience in Senegal." MADIA Study, World Bank. Wash- Document. Washington, D.C.: World Bank. ington, D.C.: World Bank. Cleave, John H. 1974. African Farmers: Labor Use in the Development of Krueger, Anne O., Maurice Schiff and Alberto Valdes. 1988. Smallholder Agriculture. New York: Praeger Publishers. "Agricultural Incentives in Developing Countries: Measuring Collinson, M.P 1974. "Cotton Development in Tanzania: A Review the Effect of Sectoral and Economy-Wide Policies." The World of the Cotton Program in Sukumaland." African Rural Bank Economic Review, 2 (September): 255-272. Development Study Background paper, World Bank. Wash- Lele, Uma. 1988a. "Comparative Advantage and Structural Trans- ington, D.C.: World Bank. formation: A Review of Africa's Economic Development _ . 1972. Farm Management in Peasant Agriculture-A Handbook for Experience." In The State of Development Economics: Progress and Rural Development Planning in Africa. New York: Praeger Perspectives. Ed. Gustav Ranis and T. Paul Schultz. New York: Publishers. Basil Blackwell Inc., pp. 187-221. Compagnie Francaise pour le Developpement des Fibres Textiles . 1988b. "Agricultural Growth, Domestic Policies, the External ICFDTI. 1987. "Rapport d'Activite pour l'Annee 1986." Paris: Environment and Assistance to Africa: Lessons of a Quarter CFDT. Century," in Colleen Roberts, led.). Trade, Aid, and Policy Cotton Lint and Seed Marketing Board (CLSMB). 1985. "Budgets Reform: Proceedings of the Eighth Agriculture Sector Symposium. for the Year Ist July 1985-30th June 1986." Appendix A: Washington, D.C.: World Bank. Income and Expenditure Budget. Nairobi: CLSMB. . 1975. The Design of Rural Development: Lessons from Africa. Cox, R.A.). and A.T. Spurling. no date. "Unit Farms in Malawi, The Baltimore: Johns Hopkins University Press. Organisation, Recording and Results of Experimental Unit . 1989a. "Sources of Growth in East African Agriculture." The Farms." Research Circular no. 4/72. Bvumbwe Research World Bank Economic Review 3 (January):119-144. Station. _ . 1984. "Tanzania: Phoenix or Icarus?" in Arnold C. Harberger, de Matharel, D. 1987. "Nigeria: Study of Cotton Marketing, Seed led.), World Economic Growth. San Francisco: Institute for Production and Quality Control." CFDT Report. Paris: CFDT. Contemporary Studies Press. Delgado, Christopher L. and John W Mellor. 1984. "A Structural . 1989b. "Structural Adjustment, Agricultural Development View of Policy Issues in African Agricultural Development." and the Poor: Some Lessons from the Malawian Experi- American Journal of Agricultural Economics 66 (December):665- ence." MADIA Study, World Bank. Washington, D.C.: World 670. Bank. Eicher, Carl. 1982. "Facing Up to Africa's Food Crisis." Foreign Affairs and L. Richard Meyers. 1986. "Agricultural Development 61 (Fall): 151-174. and Foreign Assistance: A Review of the World Bank's . 1985. "Agricultural Research for African Development: Experience in Kenya, 1963-86." MADIA Study, World Bank. Problems and Priorities for 1985-2000." Paper prepared for Washington, D.C.: World Bank. a World Bank conference on Research Priorities for Sub- and L. Richard Meyers. 1987. "Growth and Structural Saharan Africa, Bellagio, Italy. February 25-March 31, 1985. Change in East Africa: Domestic Policies, Agricultural FAO/World Bank Cooperative Programme Investment Centre. 1987. Performance and World Bank Assistance, 1963-1986, Parts I "Tanzania: Agricultural Sector Review Mission, Working and I." DRD Discussion paper, nos. 273 and 274. Washing- Paper 3: Export Crops." Rome: FAO. ton, D.C.: World Bank. 35 and Vishva Bindlish. 1988. "How Important are the Relative "Situation et perspectives du coton en Afrique de l'Ouest et du Effects of Economy Wide and Sector Specific Policies in Centre." 1987. Afrique Agriculture, no. 148, December. Explaining the Past Performance of Nigerian Agriculture?" Societe de Developpement des Fibres Textiles (SODEFITEX). MADIA Study, World Bank. Washington, D.C.: World Bank. 1987. "Note de Presentation." Dakar, Senegal: SODEFITEX. -, Robert Christiansen, and Kundhavi Kadiresan. 1989c. Societe de Developpement du Coton du Cameroun (SODE- "Issues in Fertilizer Policy in Africa: Lessons from Develop- COTON). 1986. "Rapport Annuel, Campagne 1985/86." ment Programs and Adjustment Lending, 1970-87." MADIA Garoua, Cameroun: SODECOTON. Study, World Bank. Washington, D.C.: World Bank. _ . 1988. "Note concernant le redressement de la filiere _____ and others. Forthcoming. "Nigeria's Economic Develop- coton." Garoua, Cameroun: SODECOTON. ment, Agriculture's Role and World Bank Assistance, 1961- United States Department of Agriculture. 1987. "World Cotton 1986." MADIA Study, World Bank. Washington, D.C.: World Situation." Foreign Agricultural Service, Circular Series. Bank. Washington, D.C.: USDA. Lipton, Michael. 1987. "Limits of Price Policy for Agriculture: Which van der Sleen, Max. No date. "The Manawa Project: A New Saw Way for the World Bank?" Development Policy Review (London) Ginnery in Mwanza Region, Financial Feasibility Study." 5 (lune):197-215. Report prepared for Netherlands Economic Institute. Marketing Development Bureau (MDB). 1987. "Annual Review of Mwanza, Tanzania: Tanzania Cotton Marketing Board. the Cotton Industry." Draft of 1987 Report. Dar-es-Salaam, Weber, Michael T. et al. 1988. "Informing Food Security Decisions Tanzania: Ministry of Agriculture. in Africa: Empirical Analysis and Policy Dialogue." American Ministere de la Cooperation. 1987. "Le coton en Afrique de Journal of Agricultural Economics. 70 (December): 1044-1051. l'Ouest et du Centre, situation et perspectives." Paris: World Bank. 1981. "Accelerated Development in Sub-Saharan Ministere de la Coop6ration. Africa: An Agenda for Action." Washington, D.C.: World Bank. Ministry of Agriculture. 1987. "National Agricultural Research Plan." _ . 1987. "Aide Memoire: Kenya-Cotton Processing and Lilongwe, Malawi: Ministry of Agriculture. Marketing Project." joint Government/IDA Supervision Ministry of Agriculture. 1987. "Report of the Task Force on the Mission, 16-30 November 1987. Washington, D.C.: World Programme for the Rehabilitation of Ginneries." Summary Bank. Report. Dar-es-Salaam, Tanzania: Ministry of Agriculture. . 1973. "Appraisal of Geita Cotton Project, Tanzania." Ministry of Agriculture/Ministry of Cooperative Development/ Washington, D.C.: World Bank. Cotton Lint and Seed Marketing Board. 1987. "Policies for . 1987. Cameroon: Country Economic Memorandum." Wash- the Development of the Cotton Subsector." Nairobi, Kenya: ington, D.C.: World Bank. Ministry of Education. . 1988. Operations Evaluation Department. "Cotton Develop- Ministry of Agriculture, Animal Husbandry and Water Resources. ment Programs in Burkina Faso, C6te d'lvoire and Togo." 1962. African Land Development in Kenya, 1946-1962. Nairobi, Washington, D.C.: World Bank. Kenya: Ministry of Agriculture, Animal Husbandry and Water . 1983a. "Eastern Senegal Rural Development Project." Resources. Appraisal Report. Washington, D.C.: World Bank. Mundlak, Y 1988. "Capital Accumulation, the Choice of Tech- _ . 1984. Operations Evaluation Department. "Geita Cotton niques, and Agricultural Output." In Agricultural Price Policy for Project (Tanzania)." Project Performance Audit Report. Developing Countries. Eds. John W Mellor and R. Ahmed. Washington, D.C.: World Bank. Baltimore: Johns Hopkins University Press. "Nouveaux records pour le coton en Afrique noire. 1987." Afrique l1985. "Kenya: Cotton Processing and Marketing Project." Agriculture, no. 144, July. Joint Government/IDA Mid-Term Review Mission, 4-22 Price Waterhouse Associates (Africa). 1987. ,,ODECOTON: Mission November 1985. Washington, D.C.: World Bank. Price Waterhouse Associates (Africa). 1987. SODECOTON: Mission _ . 1982. "Kenya: Cotton Processing and Marketing Project." de Diagnostic." General Report, Annexes. Staff Appraisal Report. Washington. D.C.: World Bank. Ray, Anandarup. 1988. "Response to Lipton's (June 1987) Review of Staff "Mal Lor Shire D.C.: Worldurank. WOrld Development Report 1986." Development Policy Review _ 1983b. "Malawi: Lower Shire Valley Agricultural Develop- (London) 6:211-222. ment Projects." Impact Evaluation Report. Washington, D.C.: (London) 6 :211-222. ~~~~~~~World Bank. Republique du Cameroun, Mission de Rehabilitation des Entre- . Bank. prises du Secteur Public et Para-Public. 1988. "SODE- . 1986. "Malawi: Shire Valley Agricultural Consolidation COTON, Diagnostic Rapport Provisoire," Yaounde, Camer- Project." Project Performance Audit Report. Washington, oun: Mission de Rehabilitation des Entreprises du Secteur D.C.: World Bank. Public et Para-Public. ___. 1979. "Malawi Shire Valley Agricultural Development Republique du Senegal, Ministere du Developpement Rural. 1986. Project." Project Performance Audit Report. Washington, "Conseil interministeriel sur les resultats de la campagne de commercialisation 1985/86; 1'etat d'avancement de la . 1987. "Nigeria: Agricultural Sector Review." Washington, campagne agricole 1986/87; note sur la filiere coton au D.C.: World Bank. Senegal." Dakar, Senegal: Minist&re du Developpement . 1978. "Shire Valley Agricultural Consolidation Project." Staff Rural. Appraisal Report. Washington, D.C.: World Bank. Republique du Senegal, Ministere du Developpement Rural/ _ . 1983b. "Tanzania: Agricultural Sector Report." Washington, Societe de Developpement des Fibres Textiles. 1987. D.C.: World Bank. "Programme de Sauvegarde et d'Ajustement de la Filiere _ . 1984. "Toward Sustained Development in Sub-Saharan Coton du Senegal." Dakar, Senegal: Ministere du Africa." Washington, D.C.: World Bank. Developpement Rural. Yayock, 1.Y and Vijay Kumar. 1988. "Cotton Production in Nigeria: Roupsard, Marcel. 1987. Nord Cameroun: Ouverture et Developpement. Prospects, Problems and Recommended Action for Devel- Ph.D. Dissertation, Paris: University of Paris. opment." Report prepared for the World Bank. Lagos, Santley, J.C. 1987. "Tanzania: Agricultural Export Rehabilitation Nigeria: World Bank. Project, Cotton Processing Industry." Draft of consultant report prepared for the World Bank. Washington, D.C.: World Bank. 36 THE MADIA STUDY Although many generalizations have been made about the agricultural crisis in Africa, relatively few detailed country and cross-country studies of African agriculture based on systematic data analysis have been conducted. Similarly, although foreign aid has constituted a large part of total government expenditures in Africa for close to fifteen years, there has been little analysis of the role of external assistance in African countries that goes beyond political criticism of official assistance or the alleged self- serving objectives of donors. The impetus for the study "Managing Agricultural Development in Africa" (MADIA) was to begin the process of filling this gap and to explain the nature and sources of the agricultural crisis, particularly the extent to which it originated in resource endow- ments, historical and contemporary events, external and internal policies, and the economic and political environment. The MADIA study involved detailed analysis of six African countries- Kenya, Malawi, Tanzania, Cameroon, Nigeria, and Senegal. In addition to the World Bank, seven donors, USAID, UKODA, DANIDA, SIDA, the French and German governments, and the EEC participated in the study. The analysis of country policies and performance during the last 20-25 years was carried out with the benefit of substantial input from the governments and nationals of each of the countries represented. The study had three main areas of focus: (1l the relationship between domestic macroeconomic and agricultural policy and agricultural performance, (2) donors' role in the development of agriculture, and (3) the politics of agricultural policy. The MADIA study was the result of encouragement and support from many people. Anne Krueger, former Vice President for Economic Research Staff in the World Bank, encouraged the establishment of these studies on aid and development in 1984. Gregory Ingram, former Director of the Development Research Department, provided unstinting support for the study. During the reorganization of the World Bank in 1986, the strong support from Benjamin King, then acting Vice President for Economic Research Staff, proved invaluable. Barber Conable, President of the World Bank, and Mr. Edward V K. Jaycox, Vice President for the Africa Region, have played a key role by ensuring support for the study's completion, as did Stanley Fischer, the Vice President for Development Economics. Yves Rovani, Director General of the Operations Evaluation Department, was particularly helpful as the MADIA study drew heavily on the works of OED. A special debt of gratitude is owed to the World Bank's Research Committee, which provided the initial funding for the study, and to the MADIA Steering Committee. In particular the strong support of the chair of the Steering Committee, Stephen O'Brien, has been of critical importance. Finally, without the active and continued encouragement of many African policymakers and donor officials, including numerous colleagues in the World Bank, this study would not have provided new perspectives. This support has taken the form of numerous reactions to written and oral presentations, and refinement of the analysis to identify the areas of consensus and continuing controversy. rD < The World Bank Headquarters 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. (D Telephone: (202) 477-1234 Facsimile: (202) 477-6391 Telex: WUI 64145 WORLDBANK RCA 248423 WORLDBK Cable Address: INTBAFRAD WASHINGTONDC European Office 66, avenue d'lena a 75116 Paris, France Telephone: (1) 40.69.30.00 Facsimile: (I) 47.20.19.66 Telex: 842-620628 ) 0 Tokyo Office 0 Kokusai Building 1-1, Marunouchi 3-chome > Chiyoda-ku, Tokyo 100, Japan - Telephone: (3) 214-5001 _. Facsimile: (3) 214-3657 Telex: 781-26838 CD 0-8213-1 323-1