The Desgn of 10109 Rural Development * . The Design of Rural Development The Design of Rural Development Lessons from Africa UMA LELE Published for the World Bank by THE JOHNS HOPKINS UNIVERSITY PRESS Baltimore and London Copyright © 1975 by INTERNATIONAL BANK FOR RECONSTRUCTION AND DEVELOPMENT 1818 H Street, N.W., Washington, D.C. 20433 All rights reserved Manufactured in the United States of America Originally published, 1975 Second printing, 1976 (paperback) Library of Congress Cataloging in Publication Data: Lele, Uma J. The design of rural development. Bibliography: p. 227. Includes index. 1. Africa-Rural conditions-Case studies. 2. Under- developed areas-Economic policy. 3. Underdeveloped areas-Social conditions. I. International Bank for Reconstruction and Development. II. Title. HN773.5.L34 309.2'63'096 75-10896 ISBN 0-8018-1756-0 ISBN 0-8018-1769-2 pbk. Foreword How to raise productivity among the rural poor in developing countries is one of the two or three most urgent questions confronting the international development community today. The scale of the problem is immense: * Of the 2,000 million people in the World Bank's developing member countries, roughly 1,500 million live on the land, and nearly half of those are entrapped in what can only be described as absolute poverty. * The number of the absolute poor increases and their enforced degrada- tion deepens with every passing year, despite more than two decades of ex- traordinary worldwide economic growth. The problem is not merely that the benefits of economic growth have been inequitably distributed, as they have, but that the poor themselves have been unable to contribute to that growth. The solution, then, lies in raising the productivity of the poor, so that their own incomes can rise as those of others in their societies do. But how? So far no one has the complete answers, but we in the World Bank are accelerating our efforts to find them, along with many others in governments, international agencies, and academic institutions around the world. This book represents an important element of our research in this direc- tion. It results from a major study of rural development policies and pro- grams in sub-Saharan Africa. From that experience, it attempts to draw lessons to guide our current and continuing operations in support of gov- ernment programs to raise the productivity of the rural poor. ROBERT S. MCNAMARA President, The World Bank Table of Contents I. Rural Develolpment Perspectives in Africa 3 Nature and magnitude of the problem 3 Design, timing, and methodology of the study 6 Overview of development programs 12 The concept of rural development 19 II. Nature of the Production Systems 1: Some Key Influences on the Development of Farming 22 Factors affecting labor flow into smallholder agriculture 23 The food constraint and specialization in agriculture 27 Mechanization in smallholder agriculture 33 Migration and the development of farming systems 38 Regional equity in rural development 45 III. Nature of the Production Systems 2: Diversification of Productive Activities 46 Diversificaltion in integrated rural development programs 46 Development of technological capability for diversification 50 Farming systems and intraregional equity 53 Livestock development in the traditional sector 56 Concluding remarks 60 IV. Agricultural Extension and Mass Participation 62 Different organizational approaches to extension services 64 Intensity of extension services in rural development programs 66 Costs implicit in intensity of extension services 68 Profitability of technology and effectiveness of extension 69 viii CONTENTS Incentives and performance of extension services 71 Local participation in extension activities 73 Neglect of women in agricultural extension 76 Implications of past extension experience for organization of future extension 78 V. Agricultural Credit 81 Some issues related to organization of agricultural credit 81 Potential for saving in the subsistence sector 82 Institutional smallholder credit and formal administrative requirements 84 Factors affecting repayment of credit 93 Implications of past credit components for design of future programs 96 VI. Marketing of Agricultural Output 100 Factors influencing marketing components 101 Effectiveness of marketing components in the development projects 102 Implications for the design of marketing components 112 VII. Social Services 116 Economic justification for social services 11 7 Discrepancies between demand for social services and resource availability 120 Implications for resource mobilization and increasing participation 124 VIII. Forms of Rural Development Administration 1: Autonomous Projects 127 Issues related to project authorities 127 Africanization of management in autonomous projects 130 Politics, management, and Africanization 134 Direction and coordination in the transfer of functions to the regular administration 138 Concluding statement 140 IX. Forms of Rural Development Administration 2: Nationally Planned Programs 142 The Special Rural Development Program in Kenya 143 Decentralization and the u]dmaa movement in Tanzania 151 CONTENTS ix X. Training for Rural Development 162 Training programs and the rural population 162 Training of field staff 170 Training of managers 171 XI. Summary and Conclusions 175 Some major issues related to designing rural development programs 175 Effect of national policies on past programs 178 Institutional development in past programs 186 Implications of the findings of ARDS for designing rural development programs 189 Glossary 193 Acronyms a:nd non-English terms used in the text 193 National currencies and exchange rates, 1972 196 Equivalents of weights and measures 196 Appendix: Project Reviews 197 Cameroon 197 Ethiopia 202 Kenya 204 Malawi 214 Mali 216 Nigeria 219 Tanzania 221 Bibliography 227 Background sources used 227 Additional basic reading 238 Index 239 Text Tables 1.1 Selected national statistics for the countries containing programs reviewed 4 1.2 Major features of the projects reviewed 8 3.1 Comparison of 1966 and 1971 crop production, Chilalo Awraja, E,thiopia 47 3.2 Milk purchased, in liters, by CADU, Ethiopia, 1967/68- 1971/72 47 3.3 A comparison of pre-project forecast for year 9 (maturity) and actual year 2 crop yields in quintals per hectare, WADU, Ethiopia 48 x CONTENTS 3.4 Projected and actual yields of maize and groundnuts in quintals per hectare, LLDP, Malawi, 1969/70- 1973/74 49 5.1 Amount of credit and number of loans extended in three rural development programs, Ethiopia, 1967/68-1972/73 86 5.2 Seed and fertilizer distributed on credit in three rural development programs, in quintals, Ethiopia, 1967/68-1972/73 87 5.3 Number and percentage of credit takers and distribution of gross benefits, by land tenure and size of holding, CADU, Ethiopia, 1968-71 88 5.4 Distribution of inputs on credit by tenancy and sizes of farms in nine minimum package areas, Ethiopia, 1971/72 89 5.5 Distribution of inputs on credit by tenancy of farms in the Wolamo Highlands, WADU, Ethiopia, 1971/72 89 5.6 Seasonal credit: amount loaned, number of borrowers, and default rates, LLDP, Malawi, 1968/69-1973/74 91 5.7 Total inputs supplied on credit and hectarage covered, LLDP, Malawi, 1968/69-1972/73 91 6.1 Grain purchases by the Marketing Division of CADU, Ethiopia, and additional benefits received by farmers, 1967/68-1971/72 103 6.2 Commodities handled by the Marketing, Credit and Cooperatives Division of WADU, Ethiopia, 1971/72 104 7.1 Estimated development/recurrent expenditure ratios for public services, Kenya, 1966-70 124 7.2 Estimated capital/recurrent expenditure ratios for education and health, Tanzania, 1973-80 124 Maps Cameroon-French technical assistance programs 198 Ethiopia-major rural development programs 201 Kenya-livestock areas 205 Kenya-rural development programs 207 Kenya-agricultural potential and patterns of rural migration 211 Malawi-Lilongwe Land Development Program 215 Mali-French technical assistance programs 217 Nigeria-small industrial development 220 Tanzania-tobacco and cotton areas 224 Preface What explains the very limited impact of past development programs on the low-income rural populations in Africa? Why- despite a great variety of approaches tried by donor and national agencies and despite a great amount of experience generated by these efforts- has the problem of rural poverty remained acute? If future rural development programs are to leave a more long-lasting and positive imprint on rural Africa and the rural world in general, what lessons has this experience to offer for the specification of target groups, the sequencing and phasing of activities, and the choice of policies and institutions? Concern about these questions and a search for more realistic operational guidelines resulted in intense discussion within the World Bank in 1971. This revealed how little theoretical framework there actually was for guid- ance in designing and administering programs in rural development and how little systematic, operationally focused analysis actually existed of the practical experience of the past. From this discussion emerged the basis for the present study. In carrying out a study of this kind, an international agency such as the World Bank is able, of course, to provide an operational focus; but equally important, it is able to draw upon comprehensive evidence developed from a diverse set of experiences in order to examine some of the most impor- tant policy and institutional issues of development faced by national gov- ernments and donor agencies. To exploit these opportunities, however, analysis of the development issues must go well beyond the use of formal analytical tools if an understanding is to be reached of the many administra- tive, technological, sociopolitical, and environmental factors that influence the quality of rural planning and that often explain the ineffectiveness of its implementation at the micro level. To analyze such issues, the present study draws on detailed evidence from seventeen rural dievelopment programs in sub-Saharan Africa. In each case existing survey data and analysis have been used in combination with interviews with a number of persons involved in planning, implementation, and evaluation of the programs. From the mass of micro evidence pro- duced, the study identifies the most basic factors that require attention if the gap between the overall objectives of rural development and the actual performance is to be reduced. Thus, the most important feature of the study lies not so much in a set of definitive solutions as in a way of analyzing the diverse sets of specific con- xii PREFACE straints and potentials that are encountered in rural areas. There neverthe- less emerges from the evidence much practical insight- about an appropri- ate balance between development of food and export crops, for example, and between productive and social activities, central direction and grass root involvement, and precision in planning and flexibility in implementa- tion. Most important, the study documents the need for, and the nature of, an overall policy and an institutional framework that are conducive to the objectives of rural development. The field investigations and analyses were carried out by twenty reviewers, whose names, along with salient features of the projects they reviewed, are contained in Table 1.2 (pages 8-11); references to their obser- vations also appear throughout the book. An appendix to the book contains additional description of these programs. The observations in individual studies are supplemented by my extensive field investigations in project areas and by review of a substantial literature on African rural develop- ment. There simply would not have been an effective substitute for the highly diversified field investigations and analyses, although administering such a large exercise and synthesizing its findings was not a task without frustrations. Formulation and completion of a study of this magnitude would not have been possible without the contributions and active support of a large num- ber of individuals and institutions. Members of the Bank's Eastern Africa Region- in particular Shahid Husain, K.G.V. Krishna, and Lyle Hansen- initiated the idea of an operationally oriented study. The governments, re- search institutions, and the rural people in the countries in which the pro- grams are located, as well as the donor agencies that have financed the pro- grams reviewed, provided much of the information analyzed in individual studies. Graham Donaldson and Robert Shaw participated in the project reviews and supervised eight project reports. Shirin Velji prepared the substantial bibliography. The final drafts of the manuscript were completed at Cornell University, where I was a senior research fellow at the Center for Interna- tional Studies and a visiting associate professor in the Department of Agri- cultural Economics. Milton Esman, Bemard F. Stanton, and Norman Uphoff were instrumental in providing this reflective environment. At Cornell, Arthur Goldsmith assisted me in the writing of the final drafts; his diligence and intellectual curiosity made plodding through thousands of pages of project reviews and related literature a surprisingly manageable task. Perhaps nothing better indicates the interest in the subject matter than the number of my colleagues at the Bank on whose operational experience I was able to draw and who read and contributed to improvement of the study at various stages. I am much indebted to John Cleave, Andrew Mercer, Bengt Nekby, Peter Nottidge, and Walter Schaefer-Kehnert for their input. Colleagues outside the Bank also took an energetic interest in the study. My many stimulating discussions with John Mellor were respon- sible for much improvement in the manuscript. Comments by Raj Krishna, Guy Hunter, Jon Moris, Michael Lipton, Carl Eicher, Bruce Johnston, Robert Chambers, Albert Waterston, and Ted Owens helped to broaden the scope of the issues. Of the many African nationals with whom I have had PREFACE xiii discussions, I am particularly grateful to Paulos Abraham and Philip Mbithi for the indigenous perspective that they provided. Other persons who com- mented are acknowledged throughout the book, by references to either per- sonal discussions or to their publications. Alun Morris and Rachel Anderson edited the manuscript and improved its overall presentation; the index was prepared by Annette Braver. The carrying out of a study such as this, which deals with a diverse set of issues and caters to a broad range of interests, is a rope trick difficult to per- form, even for an Indian. I particularly appreciate the unstinting support that I received from Flans Adler, Bernard Bell, Sir John Crawford, Price Gittinger, P.D. Henderson, James Hendry, Nathan Koffsky, Stanley Please, and Alexander Stevenson. Not only have they contributed substantively to the study, but also they have ensured its successful completion in numerous important ways. Despite these many contributions, I alone am responsible for the views expressed and for those deficiencies which may remain. UMA LELE Washington, D.C. June 1975 The Design of Rural Development CHAPTER I Rural Development Perspectives in Africa Nature and Magnitude of the Problem This study of rural development in Africa-the African Rural Develop- ment Study, or ARDS-grew out of a substantial interest within the World Bank in finding ways of designing relevant projects that could be ac- complished despite limited resources, particularly of money and trained manpower, and that would reach a large proportion of the low-income rural population. The World Bank's concern was prompted by the fact that between 85 and 95 percent of the nearly 310 million people living in sub-Saharan Africa live in rural areas. Most survive on a meager annual per capita income of less than $100,' based largely on low productivity agriculture and livestock (Ta- ble 1.1). Their production is oriented mainly towards subsistence. The mar- keted agricultural production comes largely from a small subsector of large estates, organized ranches, and commercial smallholders. Agriculture con- stitues the largest sector in the gross national products of most African countries (Table 1.1). Kenya well illustrates the importance of the rural sector in African economies. Recent estimates indicate that, in 1971, 90 percent of the Ken- yan population of approximately 11.7 million lived in the rural sector. Of these, 70 percent consisted of smallholders whose gross annual per capita income was rarely more than $70. Two percent were landless and 12 per- cent were nomads. According to recent estimates made by the World Bank, nomads con- stitute as much as 25 percent of the population in some Sahel countries (Chad, Mali, Mauritania, Niger, Senegal, and Upper Volta). To support their cattle, they depend on vast areas of low-potential land with an unreliable annual rainfall of 250-500 millimeters. Statistics on the levels of living of this traditional pastoral sector are even poorer than those on the subsis- tence agricultural sector. It is apparent, however, that nomadic living stan- dards are comparable with, if not lower than, those in the subsistence agri- cultural sector. 1. For purposes of comparison, unless otherwise specified, all monetary figures used in this book have been converted into U.S. dollars at official 1972 exchange rates and rounded. Table 1.1: Selected National Statistics for the Countries Containing Programs Reviewed Index of Population as Annual per capita percentage of Rural population Percent of jbod Total total population growth Per capita GDP originating production population population of as rate agricultural Per capita in agriculture, in 1972 mid-1971 sub-Saharan percentage 1965-71 land GNP, 1971 jbrestly, (base 1961-65 Country (millions) Afticaa of total (percent) (hectares) (US$) andfishing 100) (1) (2) (3) (4) (5) (6) (7) (8) Cameroon 5.8 2.4 80 2.1 2.6 200 40 114 Ethiopia 25.3 10.4 91 2.4 2.8 80 54 109 Kenya 11.7 4.8 89 3.3 0.5 160 34 105 Malawi 4.6 1.9 94 2.5b n.a. 90 52 122 Mali 5.1 2.1 88 2.1 7.7 70 43 85 Nigeria 56.5 23.3 76 2.5 0.8 140 50 86 Tanzania 13.2 5.5 92 2.8 4.1 110 40 140c n.a. Not available a. Excluding South Alrica, Rhodesia, Angola, and Mozambique. h. Recent estimates indicate a more-rapid growth rate in the 1970s, betweeni 3.0 and 3.5 percent per year. c. Estimates made by the World Bank's Agricultural and Rural Development Sector Survey ofTanzania, conducted in the fall of 1973, indicate a much smnaller growth rate of'agricul- tural production than implied in FAO data. According to the survey's estimates,which were based on statistics provided by the governmient of Tanzania, increases in agricultural production 1rom the 1968-72 period have barely kept pace with the rate of growth olf the rural population. The annual growth of agricultural production measured in constant prices declined from 3.9 percent over the period 1964-68 to an annual rate of 2.4 percent during the 1968-72 period. At the same time, the annual rate of growth of total GDP decreased from 6.3 percent to 4.3 percent over comparable periods. These figures indicate that virtually no net per capita gains in agricultural productivity were achieved. Sotirce: Columns 1, 2, 4, and 6 - World Bank Atlas (Washington, D.C.: World Bank, 1973). Colunmns 3, 5, and 7 - USAID, Statistics and Reports Division, AID Economic Data Book,fbr Afirica (Washington, D.C.: U.S. Government Printing OftEce, annual, various years, 1970-74). Column 8 - Food and Agriculture Organization of the United Nations, FAO Production Yearbook, 1972. vol. 26 (Rome: FAO, 1973). RURAL DEVELOPM.ENT PERSPECTIVES 5 Per capita cultivable land is higher in Africa compared to the densely populated parts of Asia as, for example, an average of 2.84 hectares in Ethiopia (Table 1.1) compared to 0.28 hectares in India. However, labor in- put tends to be lower and soils are generally poorer.2 Hence, average per hectare yields of principal crops may be lower in Africa than those attained in Asia.3 The overall per capita food production in sub-Saharan Africa has not shown much increase (Table 1.1). Contrary to much of the conventional wisdom on developrnent, which assumed that growth would lead to greater participation, increases in productivity in the already commercialized rural subsector have had relatively little effect on the living standards of the sub- sistence population. With annual population growth rates in the range of 2.5 to 3.5 percent, the population pressure on land has been increasing rapidly with all the con- sequent effects of poverty: hunger, unstable family life, poor nutrition, ill health, and little or no access to formal education. Understandably, the attraction of the city lights is considerable. Recent censuses show that between 1962 and 1969 the population of Nairobi grew at the rate of 15.2 percent per year. According to estimates reported by the World Bank, between 1957 and 1967 the migrant population in Dar es Salaam increased at an average rate of 8 percent per year. However, urban employment has not been increasing rapidly enough to absorb the flood of rural migrants, partly because of the particular enclave nature of industrial development. The manufacturing sector has included a relatively small indigenous entrepreneurial element and few linkages to domestic resources or to rural demand. Industrial production has been capital intensive, based largely on imported inputs, and caters to a small market of relatively high-income domestic urban consumers.4 In the absence of development of the subsis- tence rural sector, inadequate effective demand remains a major constraint to development of tlhe manufacturing and the service sector, which other- wise holds considerable potential for increasing employment.5 Improving living standards of the subsistence rural sector is important, not only as a holding operation until industrialization can advance sufficiently to absorb the rural exodus, but frequently as the only logical way of stimulating over- all development. This approach is also essential for purposes of improving the general welfare of an extremely large section of the low-income popula- tion. 2. Lawrence Dudley Stamp and W.T.W. Morgan. Africa: A Study in Tropical Development, 3rd ed. (New York: Wiley, 1972). 3. See Food and Agriculture Organization of the United Nations, FAO Production Yearbook, 1971, vol. 25 (Rome: FAD, 1972), Table 13B. FAO estimates indicate that per capita caloric in- take is also generally lower in sub-Saharan Africa than in South Asia, although fat and protein consumption is often higher. See Tables 136-38. 4. See International Labour Organisation, Employment, Incomes and Equality:A Strategy for Increasing Productive Employment in Kenya. report of an interagency team financed by the United Nations Development Programme and organized by ILO (Geneva: ILO, 1972). 5. See John W. Mellor and Uma J. Lele, "Growth Linkages of the New Foodgrain Tech- nologies," in Indian Journal of Agricultural Economics, vol. 28, no. I (January-March 1973), pp. 35-55; also Uma J. Lele and John W. Mellor, "Jobs, Poverty, and the 'Green Revolution'," in International Afifairs, vol. 48, no. I (January 1972). 6 DESIGN OF RURAL DEVELOPMENT Design, Timing, and Methodology of the Study Rural development has, therefore, received a great deal of attention in re- cent years in development literature, national plans, political platforms, and in the lending programs of most donors.6 This widespread interest is reflected in the considerable increase in the number of rural development programs being undertaken by the national governments and assisted by external donors. However, the knowledge as to how to bring about develop- ment of the subsistence rural sector is still very limited. The African Rural Development Study was carried out to improve such knowledge and to provide guidelines, based on analysis of past experience, for design and implementation of the World Bank's future rural develop- ment programs in Africa. ARDS builds on earlier Bank-initiated research into agricultural development in tropical Africa.7 However, as distinct from the previous efforts, the present study is marked by three new areas of con- cern. First, in recent years there has been an explicit interest in promoting participation of the lowest income groups through rural development pro- grams. Second, development has been viewed more broadly from the perspective of improved welfare as well as of increased agricultural produc- tivity. Consequently, there has been a keen interest in identifying ways of establishing priorities and of time-phasing between and among both pro- ductive and social service activities. Third. there has been a clear recogni- tion of the financial, manpower, and institutional constraints encountered in rural development. This has identified the need to find ways of making maximum use of these resources and of augmenting the existing rural po- tential over time. The focus of the study is, therefore, consciously operational-for exam- ple, to examine the factors which influenced the choice of interventions in planning past programs, including the capacity of these programs to iden- tify constraints and potentials in the course of planning and implementa- tion, and to assess the extent to which the programs were able to deal with circumstances not initially forseen. One of the main objectives of ARDS has, thus, been to investigate whether and how the reviewed projects and similar programs should have been designed, had there been more concern with broad participation, overall welfare, and the utilization of local finan- cial and institutional resources. Relatedly, the study also investigates whether there is a need to bring about institutional and procedural changes in planning and implementing rural development programs to increase their effectiveness in realizing the new objectives of development. For the comparative evaluations on which the major findings presented here are based, seventeen sets of rural development projects and programs involving the participation of a number of multilateral, bilateral, and na- tional agencies were carefully analyzed. These were selected from nearly sixty programs in various parts of sub-Saharan Africa initially listed for perusal. While no limited survey can claim to be exhaustive, the projects 6. See, for instance, Robert S. McNamara, Address to the Board of Governors, Nairobi, Sep- tember 24, 1973 (Washington, D.C.: World Bank, 1973). 7. See the Bank-sponsored study by John C. de Wilde, Experiences with Agricultural DeLelop- ment in Tropical Afi-ca, 2 vols. (Baltimore: Johns Hopkins Press, 1967). RURAL DEVELOPMENT PERSPECTIVES 7 were selected to represent a very considerable diversity in design and imple- mentation as well as in the environment in which they are situated.8 The reviews were carried out with a view to raising a consistent set of questions about these programs and projects so as to facilitate comparison and to pro- vide the basis for obtaining operational insights.9 This work has also drawn on the rural sector surveys conducted by the Bank in Kenya in 1972 and in Tanzania in 1973 as part of the African Rural Development Study. These surveys analyzed development policies to evolve lending strategies for the two countries. The objectives and the methodology of the study were finalized after considerable discussions among various Bank staff as well as with persons outside the World Bank who had been involved in rural development. From these discussions a consensus emerged that, before collecting addi- tional information, it was necessary to analyze the existing information systematically and comparatively, to review the state of the knowledge on rural development, to improve operational decisions, and to develop in- sights into the priorities for further operational research. Given the considerable interest of the operational parts of the Bank in the findings of the study, the evaluations of the programs on which this re- port is based were completed in sixteen months between July 1972 and December 1973. All the reviews were based on data already collected by a number of agencies. With a few exceptions no field surveys to obtain addi- tional quantitative data were conducted. However, in all cases data analyses were combined with field investigations. These consisted of (a) interviews with persons who had considerable experience in design, implementation, supervision, and evaluation of the specific programs reviewed or similar other programs; (b) interviews with the rural people who were being reached through programs; and (c) searches for additional sources of data related to programs reviewed from other donor, government, and research agencies. 8. See Table 1.2 of this study. The following section of this chapter and the Appendix con- tain additional information and map locations of the programs reviewed. To facilitate com- parative analysis in conducting reviews, some of the programs were grouped together. Therefore, although seventeen sets of programs were reviewed, only thirteen project reports were prepared. The last column of Table 1.2.indicates the grouping of the programs, and the Appendix provides their corresponding descriptions. 9. In this report the terms "projects" and "programs" are used interchangeably. However, since the analysis is carried out with the objective of providing guidelines to the Bank Group for designing projects, the distinction between the two is worth noting at this stage. A project is generally understood to mean (a) a well-defined range of activities which lend themselves to rather precise specification of objectives to planning, financing, and implementing within an organized unit and which have a clear beginning, (b) a set of targets by which to judge their performance; and (c) an orderly specified end. A number of activities financed by donor agen- cies in the rural sector have usually been designed, appraised, implemented, and evaluated within this project concept. On the other hand, the ujtimaa movement and the Sukumaland cotton development in Tanzania or the spontaneous squatter settlements in Kenya are not projects in this sense. They have been prompted by broader sociopolitical or administrative considerations which do riot lend themselves to precise specification of objectives or to a pre- cise set of targets by which to judge their performance. Therefore, although many of the issues in this book are presented in terms of their implications for project design and imple- mentation, the analysis deals with broader questions related to development of the low-in- come subsistence sector. As will be apparent later in the text, these broader questions have substantial implications for what constitutes a project. It should also be noted that, despite their obvious limitations and for the sake of brevity, the terms "low income" or "subsistence" are used interchangeably in this study to describe the target populations. Table 1.2: Major Features of the Projects Reviewed First year of Country Project Categor implementation Project aim Cameroon Zones d'Action Priori- Regional inte. 1967 To integrate rural development taires Integrees (ZAPI) grated rural de- under independent, nongovern- velopment mental companies set up in each of seven selected areas in the East and South Central prov- inces; involving farmer organi- zations, extension and credit, and social development Cameroon Societe de Developpe- Regional settle- 1966 To integrate development of ment d u Nkam ment and inte- pioneer villages in a sparsely (SODENKAM)LandSet- grated mral de- populated region of 120000 hec- tlement Scheme velopment tares in the Littoral province: technical and cooperative mar- keting support for coffee and cocoa production; social devel- opment Ethiopia Chilalo Agricultural De- Regional inte- 1967 To provide a wide range of pro- velopment Unit(CADU) grated rural de- ductive and social services, velopment assisting participants to increase net per capita income Ethiopia Wolamo Agricultural De- Regional inte- 1970 As above-and to establish velopment (WADU) grated rural de- 1,750 settlers in two lowland velopment and areas of Wolamo settlement Ethiopia Minimum Package Pro- Miscellaneous 1970 To provide rudimentary facil- gram (MPP) ities, including credit, fertilizer, improved seeds, extension, cooperative development, and feeder roads for 100 areas in the fourteen highland provinces to be reached bv 1980 Kenya Kenya Livestock Devel- Miscellaneous 1970 To provide credit for develop- opment Project (K LDP) (Swynnerton Plan ment of individual, commercial, developed in mid- company, and group ranches fifties. Kenya Range and grazing blocks to promote Development Pro- rural employment and to in- gram began in 1964) crease production and export of livestock products Kenya Agricultural Finance Functional 1963 Tofinancealargenumberofac- Corporation (AFC) (Some functions tivities throughout the rural sec- begun under British tor, including purchase of land Land Transfer Pro- front European owners, live- gramn in 1961 and stock raising, procurement of the Land and Agri- farm equipment, and on-farm culture Bank in development 1931) Kenya Special Crops Develop- Commodity SCDA: 1960 (to To increase tea production and ment Authority (SCDA); 1964); KDTA: 1964 exports among smallholders Kenya Tea Development Authority (KTDA) 8 Authors of A verage sizec reports prepared Project cost' Financing agencies Projeclpopulationb ofholding under the study South Central prov- FAC and government Population of approxi- Less than 3 hectares ince: S2.48 million (to of Cameroon mately 1 75,000 in the for cocoa and I hec- 1972), East province: seven ZAPIs tare for food crops Initial financial assis- tance of $0.7 million provided to the proj- ect, excluding techni- cal assistance G. Belloncle $0.5 million(1972/73) Governnment of Settlementof l8,000 to 1.0-2.0 hectares D Gentil Cameroon 22,500 inhabitants by 1981 $19.3 million (to SIDA, with govern- Population within Chi- 3 hectares 1975) ment of Ethiopia lalo Awraja of 40033000 $5.1 million (to 1974) IDA, Urnited King- Population within 0.5 hectares dom, and World Food Wolamo Awraja of Program, with govern- 240,000; 460 old settle- ment of Ethiopia ments reorganized and about 207 new settlers T Tecle accommodated by Oc- T tober 1972 $30 million (to 1976) IDA.SIDA, and FAO. Approximately 5 mil- 0.7-2.0 hectares with government of lion people living in Ethiopia MPP areas in 1974: 19 million people to be reached by 1980s $11.3 million (to IDA, SIDA, USAID, 2,500-3,000 ranchers Individual ranches: H. Jahnke 1974) d with government of or pastoralists 600 hectares, com- H. Ruthenberg Kenya mercialranches:3,250 H. Thim hectares: group ranches: 18,000 hec- tares: grazing blocks: 700,000 hectares $33 million (total IDA, SIDA, British Covers the entire Not applicable J.D. von Pischke assets in 1971) e Land Transfer Pro- country gram, Federal Repub- lic of Germany, with governnient of Kenya S15.6 million (to IDA, C'ommonwealth 66,500 farmers culti- 2.7 hectares D. Sullivan 1971) Development Corpora- vating tea (1972) tion (C[)C), the Feder- al Republic of Ger- many, with govern- ment of Kenya (Table continues on the following page.) 9 Table 1.2 (continued) First year of Country Project Category implementation Project aim Kenya Special Rural Develop- Regional inte- 1971 To enable major strategic exer- ment Program ISRDP) grated rural de- cise by the Kenyan government velopment in decentralized and innovative planning, embracing numerous programs for increasing rural in- comes, employment and welfare Kenya Spontaneous Land Settle- Spontaneous Not applicable Substantial local self-help ment effort among squatters who have es- tablished settlements on land to which they have no title Malawi Lilongwe Land Develop- Regional inte- 1967 To increase production of major ment Program (LLDP) grated rural de- crops in a concentrated area of velopment 465,587 hectares through im- provements in rural infrastruc- ture, land reorganization, train- ing, credit, extension, market- ing, and livestock development Mali Compagnie Francaise Commodity 1952 To increase cotton production pour le Dvelopprement and exports des Fibres et Textiles tCFDT) Mali Bureau pour le Deve- Commodity 1967 To improve groundnut produc- loppement de Production tion and marketing and to in- Agricole (BDPA)- crease exports Operation Arachide Nigeria Small-scale rural indus- Miscellaneous IDC: 1962; SIC: To provide Finance, investment tries: Industrial Develop- 1966 guidance, training, and other ment Center (IDC) and services for new and expanding Small Industry Credit small industries Schemes (SIC) Tanzania Urambo Setilement Commodity Urambo: 1951, To incTease tobacco productioti Scheme: Tumbi Settle- Tumbi: 1954: World through settlement schemes at ment Scheme; World Bank: 1970 Urambo and Tumbi linked to Bank Flue-Cured To- farmers'cooperatives and provi- bacco Projeci sion of credit and extension and to increase tobacco production for export in 150 uj6maa villages in the Tabora region through a program linked to collective pro- cessing Tanzania Sukumaland Cotton De- Commodity 1950 To increase production ofcotton velopment for export on existing and new farms in the Sukumaland area of northwest Tanzania through government initiatives in hus- bandry, credit, extension, re- search, land clearance, and cooperatives Tanzania Ujjrimaa village develop- Regional inte- 1967 To create communally orga- ment grated rural and nized self-help villages, pro- social develop- moted through decentralized ad- ment ministration and the Tanganyika African National Union (TANU) a. The cost fgures are based on the best available information and converted to U.S. dollars at offtcial 1972 currency ex- change rates. In several cases where the project was of an informal and/or broadly defined nature, it was impossible to esti- mate costs. b. Except for SODENKAM, KLDP, KTDA, and Smallholder Tobacco Development, the population figures refer to the number of persons living within the project area, many or most of whom are not direct participants in program activities. These nonparticipants may, nonetheless, "beneftt" from the project's interventions, though the impact will, of course, be greater on the active participants. 10 Authors of' A verage sizec reports prepared Project costa Financing agencies Projectpopulationb ofholding under the study S1.6 million (in Government ol' Six districts with popu- Not applicable N. Bedi 1972/73) Kenya, witi USAID, lation of 826000 (1971) SIDA, the United Kingdom, NORAD, and the Netherlands Not applicable Not applicable The Kenyan squatter Not applicable P. Mbithi population is approxi- C Barnes mately 300,000(1969) $14.6 million (to IDA with government Population of 550,000 2.0-4.0 hectares B.H. Kinsey 1975) of Malawi (1973) within Lilongwe district S9.8 million (up to European Deselop- Population of 1.0 mil- 4.0-5.0 hectares 1971172) ment Fund and FAC lion within cotton zone (1971/72) W Anderson $19.7 million (up to FAC with project sup- Population of 0.75 mil- 3.83 hectares (1970) 1971/72) port by World Bank lion w i th i n the groundnut operation (1971/721 Not available USAID with govern- IDC northern and Not applicable H. Turner ment of Nigeria eastern regions: SIC: M. Shah northern region Urambo: $0.26 mil- U,ambo: originally Urambo: 2,400 farmers Urambo:2.0 hectares: M. Agarwal lion (in 1964/651: Tanganyika Agricul- cultivating tobacco Tumbi: 1.2 hectares: D. Linsenmeyer Tumbi: $0.10 million tural Corporation; 11970): Tumbi: 4,600 World Bank: 4.0 hec- (in 1964165), World Tumbt: originally Brit- farmers cultivating to- taresf Bank: S14.7 million ish American Tobacco bacco (1970): World (by 1978) Co.: World Bank: Inter- Bank: 15,000 to 30D000 national Development farmers cultivating to- Association (IDA), bacco (by 1978) with government of Tanzania Not available Govern ni aent ot Sukumaland popula- 3.1 hectares M. Collinson Tanzania and IDA tion of 2.5 million Not available Govern ment of 2 million ujzimaa vii- Not applicable F. Robinson Tanzania with assis- lage residents (1973); P. Abraham tance from various expected to cover en- agencies for specific tire rural population by projects the end of 1975 c. These figures should be taken only as approximations, as some refer to family holding and others to cropped area per farm Figures on a more comparable basis could not be obtained. d. A.second livestock project, costing $59.7 million, will enlarge and extend the project as reviewed in this study. e. Total cost of AFC could not be obtained. In the 1970/71 period AFC disbursed about $4.3 million. f. The World Bank tobacco project also called for the settlers to be provided with 12 hectares of timberland and 4 hectures of non-arable land. Source: Project reviews carried out under ARDS. 11 12 DESIGN OF RURAL DEVELOPMENT The availability of data varied considerably between programs. For those programs which either had a substantial data base and/or were reviewed by consultants who had earlier carried out considerable primary research on particular programs (as in the case of the Lilongwe Land Development Pro- gram in Malawi; cotton development in Sukumaland, Tanzania; or range development in Kenya), much systematic analysis has been possible through reviews. In other cases (e.g., ujamaa in Tanzania or the French-assisted cotton and groundnut schemes in Mali), even the benefits which are normally con- sidered to be easily quantifiable, such as increases in agricultural produc- tion, could not be estimated reasonably accurately. Cost records were in- sufficient in many cases. Wherever possible, conventional cost/benefit analyses were conducted. However, the major strength of this study lies in its examination of the broad institutional and participatory questions re- lated to rural development rather than in its conventional evaluations of projects. The methodological tools available for analysis of such institutional and distributional questions are, of course, rather crude- a problem that is com- pounded by the need for an interdisciplinary approach to such analyses. There is also a broad range of views on rural development and on the validity of some of the research results. Wherever possible, these various points of view have been presented to provide a balance to the arguments and to broaden the perspective. Substantial literature on African rural development has also been reviewed to support or oppose the observations made in the study. However, a strong consensus emerges on a surprisingly large number of issues. Overview of Development Programs The programs selected for analysis may quite justifiably be classified in several alternative ways as, for instance, by degree of integration, breadth of objectives, size of target population, or by funding agencies. To emphasize their diversity in design and implementation and to point to the gradual shift in objectives, they are classified into the following five categories: Commodity Programs The first investments in the development of low-income subsistence ag- riculture in Africa were prompted by the relatively straightforward objec- tive of increasing production of export crops in the smallholder sector. They were financed by European commercial companies, development cor- porations, or the governments of former colonies. Four sets of such pro- grams were selected for analysis. To increase tea production and exports among smallholders, the Kenya Tea Development Authority (KTDA, first known as the Special Crops De- velopment Authority) was established by the Kenyan government in 1960. By 1972 the project had brought 24,900 hectares in ten districts under tea cultivation and produced almost 313,000 quintals of green leaf. The net return from a hectare of tea (1970/71) was approximately $543. To put rural income disparities in Africa in a proper perspective, it is important to note RURAL DEVELOPMENT PERSPECTIVES 13 that, even with a substantial improvement, the per capita average annual net cash income from tea production was only about $41, compared to the per capita cash income of $7 in the pre-program era."' The smallholder sec- tor presently accounts for 21 percent of Kenya's total tea production. KTDA is known as one of the most successful and profitable rural develop- ment projects. Development of smallholder tobacco was undertaken by the Tanganyika Agricultural Corporation (TAC) in the Urambo region of Tanganyika in 1951. The scheme is now managed by the Urambo Farmers Cooperative Society. A similar scheme, presently managed by the Tabora Tobacco Growers Cooperative Society, was started by the British American Tobacco Company (BAT) in the nearby Tumbi region in 1954. Combined tobacco production in the two schemes grew from about 1,110 quintals in 1960 to nearly 36,300 quintals by 1970. The two schemes produced about a third of Tanzania's flue-cured tobacco (flue-cured tobacco constitutes over 80 per- cent of Tanzania's tobacco exports). In the mid-1960s the per capita cash in- come of the tobacco-growing population was estimated to be $21 in Tumbi and $62 in Urambo, compared to $5-$10 for the non-tobacco-farming pop- ulation in the Tabora region." In Mali a cotton production operation was undertaken in 1952 with assis- tance from the Compagnie Francaise pour le Developpement des Fibres et Textiles (CFDT). A similar scheme for groundnuts-Operation Arachide-was started by the Bureau pour le Developpement de Produc- tion Agricole (BDPA) in 1967. In 1972 the CFDT project produced virtually all of Mali's cotton and the BDPA scheme about half of Mali's groundnuts. From 1961 to 1972 CFDT's cotton production increased from 6,000 to 68,000 metric tons. Between 1967 and 1972 groundnut production under BDPA increased from 18,000 to 75,000 metric tons. Participants in the CFDT project had net per capita cash incomes of approximately $14 in 1972. The figure for participants in BDPA was approximately $2-$5. In these export crop schemes, the agricultural services of extension, cred- it, inputs, marketing, and processing have been administered through proj- ect authorities or parastatal organizations. The project authorities have con- siderable autonomy, although technically they may be under the Ministry of Agriculture. Most cash crop schemes initially managed by expatriates are being more rapidly Africanized since independence. 10. The per capita income figures given here and below are only approximations and are not strictly comparable between projects, since some refer to net income and others to net cash income (i.e., excluding the value of subsistence production). More comparable figures could not be obtained. The existing income statistics are useful, however, in indicating levels of living in the program areas. The production and hectarage figures are likewise limited. Though based on the best available information, there are inevitable disparities in the quality and currency of the data. Sources and limitations of various data series are discussed in in- dividual project reviews. Where relevant for analyses, these questions have been discussed in the subsequent chapters. A flurther difficulty is encountered in estimating the number of per- sons reached by a specific project activity, which generally refers to a narrow definition of project benefits. References to population covered by a project imply the potential population that may eventually be reached according to the broader criteria. 11. See W. Scheffler, "Tobacco Schemes in the Central Region," in Smaliholder Farming and Smaliholder Development in Tanzania, ed. Hans Ruthenberg, Afrika-Studien no. 24 (Munich: Weltforum Verlag, 1968), and W. Scheffler, Baiierliche Produktion tinterAu sicht am Beispiet des Tabakanbaus in Tanzania, Eine Social-Okonomische Stwdie, Afrika-Studien no. 27 (Munich: Weltforum Verlag, 1968). 14 DESIGN OF RURAL DEVELOPMENT Development of export crops has also occurred through normal adminis- trative efforts. Cotton development in Sukumaland, Tanzania, is such a case. As a result of the Tanzanian government's extension program, tsetse fly clearance, and cotton marketing service, average cotton hectarage per farm in Sukumaland increased from 0.28 hectares in 1950 to 1.54 hectares in 1966. Total cotton production in the area grew from 40,000 bales in 1950 to 405,000 bales in 1970. The increase in cotton production was facilitated by the more than doubling of the world market price of cotton, which rose from $0.09 to $0.21 per kilogram between 1950 and 1962. According to th. detailed farm management surveys carried out by Collinson, the annual per capita cash income in the region increased from less than $2 to about $22 between 1950 and 1966. The major differences between the export crop developments under- taken by the indigenous administrations and those undertaken with com- mercial or foreign assistance and administered through autonomous project authorities are the latter's greater administrative coordination and use of trained manpower and financial resources. For example, while KTDA had one extension worker for every 120 farmers and CFDT one for every 175 farmers, the ratio in Sukumaland has been 1:1,020. KTDA's total operating expenses per beneficiary, excluding capital costs for factory construction, road building, and so forth, were estimated to be $200 by 1971, or $18 per beneficiary per year. Given the 1.7 million to 2.0 million farm familes in Kenya, if similar investment were undertaken on a national scale, it wouid mean an annual recurrent expenditure of approximately $30-$36 million for agriculture services alone. The Kenyan government's total recurrent ex- penditure on agriculture averaged $24.3 million between 1968 and 1971.12 Obviously, to apply the intensity of KTDA's extension effort on a na- tional scale, the government would have had to increase its spending on ag- riculture by only 25 to 50 percent. However, it is not so much the total resources allocated which determine effectiveness as the way in which the resources are used. In this regard KTDA has been extraordinarily effective. The reasons for KTDA's success are analyzed later in this study. Regional Rural Development Programs The regional programs reviewed under ARDS may be classified as two types: (a) integrated rural development programs undertaken mainly on the initiative of donor agencies and planned and administered by expatriates and (b) regional development programs undertaken with substantial initia- tive and participation by national governments. Three sets of integrated programs of the first type and two sets of the second were selected for anal- ysis in this review. Programs of the first type include (i) the Chilalo Agricultural Develop- ment Unit (CADU) and the Wolamo Agricultural Development Unit (WADU), both in Ethiopia; (ii) the Lilongwe Land Development Program (LLDP) in Malawi; (iii) the Zones d'Action Prioritaires Integrees (ZAPI) 12. USAID, Statistics and Reports Division, "Central Government Finances: Kenya," in AID Economic Data Book br Africa. revision no. 286 (Washington, D.C.: U.S. Government Printing Office, October 1972). RURAL DEVELOPMENT PERSPECTIVES 15 and the Soci&te de Developpement du Nkam (SODENKAM) settlement scheme (originally Operation Yabassi-Bafang), both in Cameroon. These regional programs have been based on the assumption that a criti- cal minimum effort is necessary to make a noticeable impact on the target * populations in a relatively short period. The integrated programs have, therefore, simultaneously provided a number of productive and social ser- vices, such as soil conservation measures, roads, general agricultural exten- sion, credit, marketing services, livestock development services, training, health clinics, water supply, and nutrition education. As in many agro-in- dustrial enterprises, the integrated programs have involved substantial con- centrations of financial and trained manpower resources in relatively small geographical areas (see Table 1.2). For example, the Lilongwe Land Development Program in Malawi origi- nally covered an area of only 238,460 hectares. During negotiations with IDA the program expanded to include an estimated 465,550 hectares. LLDP was expected to increase the per capita net annual income of partici- pants in the program from approximately $10 to the range of $17-$35, de- pending on how income was defined and estimated. By the 1972/73 period LLDP had constructed 851 miles of roads, fifteen major bridges, and 340 boreholes. In the same year the program trained 25,000 farmers and dis- tributed 23,000 seasonal credit loans worth a total of approximately $457,000. In Ethiopia the Chilalo Agricultural Development Unit, located in Chilalo Awraja, covers about one million hectares. The project was ex- tended to assist participants in attaining a net annual per capita income of approximately $33. In the 1972/73 period CADU distributed 13,000 loans totaling about $420,000. The program also carries out substantial staff train- ing as well as developing agricultural technology, marketing cooperatives, feeder roads, and an improved water supply. LLDP and CADU each employed over thirty expatriates during their first few years of program im- plementation. Also in Ethiopia the Wolamo Agricultural Development Unit in Wolamo Awraja is more modest in scale. In its first phase the program was expected to raise the per capita cash income of members of 7,750 participating farm families from $11 a year to $18-$24. In 1972 the program disbursed 7,400 loans worth $132,000. WADU has also conducted road and water develop- ment programs and has settled or reorganized 600 families in sparsely pop- ulated lowland areas. At the time the program was reviewed, WADU had carried out little staff training. The seven integrated priority action zones, or ZAPIs, located in the South Central and East province of Cameroon, were planned in 1966 by Compa- gnie d'Etudes Industrielles et d'Amenagement du Territoire (CINAM), a private French firm. Participants in the three East ZAPIs were expected to increase their annual per capita income from $19 to $47. In 1972 the four South Central ZAPIs loaned about $363,000 as working capital for cocoa production. ZAPIs have also involved an intensive extension effort (one advisor per 150 farmers), women's extension programs, and the establish- ment of farmers' organizations. The programs carried out by ZAPIs in the East province were simnilar. In 1972 they disbursed approximately $45,700 in loans. 16 DESIGN OF RURAL DEVELOPMENT The SODENKAM settlement scheme, located in the Littoral province of Cameroon, was designed by Societe Centrale pour les Etudes du Ter- ritoire-Cooperation (SCET-Cooperation) in 1964. By 1972, 1,180 pioneer familes had been settled in the project area. To date SODENKAM has dis- tributed 400,000 coffee seedlings and 300,000 cocoa seedlings and has built ten schools and six health dispensaries. The program also has a cooperative marketing component. Because the existing indigenous administrative capacity in Malawi, Ethiopia, and Cameroon was inadequate for the implementation of com- plex rural development programs, the integrated programs were adminis- tered through parallel administrative structures established along the line of agro-industrial ;nterprises. In creating such autonomous administrative units, it was assumed that the improvements in the indigenous administra- tive systems essential to achieve similar results on a wider scale could often be initiated more effectively by the demonstration effect of the integrated projects than by the alternative of improving the inadequate, uncoordi- nated, and ill-trained existing administration. In the short run the use of a large number of expatriates in managerial positions made possible by the creation of autonomous project authorities has also substantially increased the capacity of the recipient countries to invest in regional development. Unlike the integrated programs, the two nationally initiated efforts reviewed in the study, namely, the Special Rural Development Program (SRDP) in Kenya and the ujeimaa movement and associated decentraliza- tion of administration in Tanzania, have been undertaken largely with the objective of improving the domestic capacity for planning and implementa- tion. It is hoped that with this approach the indigenous administrations will develop the capability to carry out effectively the ultimate objective of in- creasing the incomes and welfare of the mass of the subsistence rural popu- lation. SRDP, conceived in 1966, was a major administrative effort to improve interministerial and interdepartmental coordination. Implicit in the SRDP concept was the delegation of planning and implementing responsibility to the lower levels of administration so as to reduce the planning and imple- mentation gap and thus improve the effectiveness of rural administration. The program involved the initiation of a variety of rural development schemes, including a master farmer project, the development of hybrid maize and cotton, a cattle development program, youth training, road con- struction, and home economics extension. These schemes were to be planned and administered by divisional and local-level administrators. In 1971 SRDP covered six districts with 8 percent of Kenya's rural population. Apart from improving administrative effectiveness in the districts, SRDP aimed at producing administrative innovations that would be a step towards greater administrative decentralization on a national scale. Thus, SRDP is essentially an exercise in deconcentration of administrative au- thority.13 In Tanzania, by contrast, there has been an attempt to bring about 13. See David K. Leonard, "Communication and Deconcentration," in DevelopmentAdmin- istration: The Kenyan Experience, eds. Goran Hyden, Robert Jackson, and John Okumu (Nairobi: Oxford University Press, 1970), pp. 102-11 and Chapter IX of this book. RURAL DEVELOPMENT PERSPECTIVES 17 devolution of administrative power by fostering participation of the peasant masses in the planning and implementation of rural development pro- grams. Ujamaa village settlements constitute a major part of the develop- ment strategy. It is expected that ultimately all activities in the ujamaa vil- lages will be undertaken on a cooperative basis. Between the Arusha declaration of ujamaa policy in 1967 and the year 1973, more than 5,500 uj6maa villages were formed. The degree of local participation and collec- tive cultivation varies substantially among these villages. Since 1972 pri- mary administrative authority for the planning and implementation of the rural development effort has been vested in the regional offices rather than in central ministries. The village and ward-level committees, which repre- sent smallholder interests, and the Tanganyika African National Union (TANU), which is the national political party, have also been assigned a greater role than they enjoyed previously in the allocation of budgetary resources and the planning and implementation of local programs. Functional Programs While the regional programs involve a multifaceted approach to rural de- velopment, the functional programs are undertaken to remove a single con- straint that is considered to be particularly critical for getting rural develop- ment under way. Investment may be directed only to the development of a national network of adaptive agricultural research, to training extension services, to the construction of feeder roads, to the provision of agricultural credit, or to improving the agricultural marketing network. Functional pro- grams are appealing because of their relatively clear and limited objective and their potential for benefiting large numbers of rural people. In one sense functional programs are also somewhat easier to implement than in- tegrated programs, since they involve only the development of a single ma- jor institutional structure. Substantial investment in a particular functional program can also be an effective means of emphasizing the program's criti- cal importance, of generating a long-term national commitment to it, and of ensuring that sufficient national resources are allocated for its continuance. Functional programs rnay also be effective in stimulating other develop- ments-e.g., demand for fertilizer may be stimulated by profitable tech- nologies, and production for market may be prompted by feeder roads. The agricultural credit administered by Kenya's Agricultural Finance Corporation (AFC) constitutes such a functional program. Though the bulk of AFC's operations have been aimed at large-scale farmers, the agency has had programs for smallholders since 1965. Most significant of these has been the World Bank's smallholder credit project (IDA 105), for which IDA provided $3.6 million in 1967. Aimed to assist 8,100 farmers with per capita cash incomes of approximately $30-$35, by 1972 the project had approved 10,000 loans averaging $470.14 Nevertheless, during the fiscal year ending in March 1971, only about 30 percent of AFC's total disbursements, or $1.3 million, were classified as small-scale loans (i.e., loans under $1,400). 14. The cotton and tobacco development schemes in Tanzania discussed above have also been assisted through similar credit programs financed by loans from IDA. These funds have been administered through the Tanzanian Rural Development Bank. 18 DESIGN OF RURAL DEVELOPMENT AFC's activities cover the entire country and involve financing a large number of activities in the rural sector, including the purchase of land from European owners, livestock raising, and the procurement of farm equip- ment. AFC's interest rates ranged from 3.5 percent per annum on loans for land purchases to 7.5-8.0 percent for seasonal loans to smallholders. By contrast, the interest rates charged by traditional sources of credit, such as friends, relatives, and local moneylenders, vary substantially. They are, however, generally presumed to be far higher than those charged by credit institutions, such as AFC, and often exceed 100 percent per year. Miscellaneous Category of Planned Programs Production of specific commodities may be developed by providing a number of services related to the development of a specific subsector or region. The Kenya Livestock Development Project (KLDP), the Industrial Development Centers (IDC) in Nigeria, and the Minimum Package Pro- gram (MPP) in Ethiopia constitute such cases of subsectoral development. The livestock program undertaken by the government of Kenya involves the development of commercial, company, individual, and group ranches in Kenya's Central, Eastern, and Rift Valley provinces. In addition, USAID has been assisting in the creation of grazing blocks on marginal land in the Northeast province. By 1972 thirty commercial, eight company, thirty-nine individual, and twenty-five group ranches were financed. Loans to com- mercial ranches averaged $46,000 per ranch. For company, individual, and group ranches, the average loan sizes were $106,000, $5,570, and $4,250, respectively. It was expected that these loans would generate additional employment for 400 persons on commercial ranches and for 422 persons on company ranches. Loans to individual or group ranches were not expected to create additional employment opportunities but only to support the ex- isting pastoral population. The development of grazing blocks involves direct grants rather than loans. Total investment in the blocks is expected to be $1.5 million. In 1962 Nigeria established an Industrial Development Center in Zaria. Another IDC was initiated in Owerri the following year. The purpose of these centers was to appraise applications of small industries for financing, to provide advice on investment opportunities, and to sponsor training in a wide range of technical and managerial skills. Between 1969 and 1972 loans of $2.7 million were granted to 294 small enterprises (employing less than fifty persons and with fixed assets worth $90,000 or less) in such industries as baking, sawmilling, printing, and motor vehicle repairing. To spread the benefits of agricultural development more broadly than is possible in the integrated programs, the Minimum Package Program (MPP) was undertaken by the Ethiopian government in 1971 on the initiative of the Swedish technical experts involved in the Chilalo Agricultural Develop- ment Unit. The program is being implemented by the Evaluation and Pro- gram Implementation Department of the Ministry of Agriculture. As its name suggests, MPP involves the provision of only those services con- sidered to be critical for rural development. In this case they include agricul- tural extension, credit, cooperative development, and feeder roads. Plans call for the development by 1981 of one hundred MPP areas in the fourteen highland provinces. Some twenty-eight of these areas were in operation in RURAL DEVELOPMENT PERSPECTIVES 19 1974. Each MPP area is intended to contain about 10,000 farming families. If developed according to plan, MPP will cover about half of the geographic area of Ethiopia and will involve the entire agricultural population, exclud- ing the nomadic tribes. Spontaneous Efforts Unlike Tanzania, which has adopted an explicit policy of fostering the broad participation of the lowest income groups in development, some countries have not taken adequate steps to encourage participation of the lowest income rural populations. Nevertheless, even in such situations, considerable development effort may be under way through local initiative. It is estimated that in Kenya, as a result of the landlessness, unemploy- ment, and the lack of alternative economic opportunities, 300,000 people of various tribal origins have established squatter settlements on land to which they have no title. These settlements receive few, if any, government services. However, they have displayed a notable degree of self-help in the establishment of schools and health clinics. Three representative squatter settlements, the Kibwezi-Mtito-Chylulu hill zone in southeast Kenya and Ngoliba B and D and the Muka Maku Cooperative Society in central Kenya were reviewed as part of ARDS. The Concept of ]Rural Development It is apparent that the programs reviewed together involve a very consid- erable effort to extend development to low-income subsistence populations, although the intensity of the effort varies widely among them. The diver- sity in the programs is not confined, however, to the degree of effort. It ex- tends to a variety of other factors. For instance, the programs vary consid- erably in their emphasis on the realization of specific objectives. They also illustrate the considerable range in potential and constraint that exists among regions and countries in terms of policies, institutions, physical and human resources, and technological possibilities. Consequently, they dem- onstrate in a concrete way the possible divergence between the objectives of rural development and the physical and institutional potential for their realization. Further, the programs reflect the varying degrees of knowledge as to the actual potential and the most effective ways of developing it. They, therefore, manifest the desire of the planners to experiment with alterna- tive approaches. Thus, apart from the genuine diversity, the programs reflect the gaps between the facts and the assumptions on which rural de- velopment planning is often based. Such gaps frequently explain differences between expected and actual project performance. The past pro- grams also differ in tlaeir flexibility in implementation. They, therefore, dis- play varying ability to minimize losses from unanticipated responses and to exploit unforeseen potential. The analysis of these past efforts should pro- vide useful insights into the nature of interactions and conflicts among these numerous and immensely variable factors. It should also illuminate the preconditions necessary for ensuring the effectiveness of alternative ap- proaches and, thus, provide useful lessons for the formation and implemen- 20 DESIGN OF RURAL DEVELOPMENT tation of future rural development strategies. The thrust of the study is, therefore, positive. It is not meant to be a postmortem undertaken to deliver a verdict on past performance. But before commencing with the discussion of these issues and particu- larly in view of the very broad range of objectives, it is essential to outline the new objectives or concept of rural development. In this study rural development is defined as improving living standards of the mass of the low-income population residing in rural areas and mak- ing the process of their development self-sustaining. This simple definition has three important features with substantial implications for how rural de- velopment programs are designed and implemented: 1. Improving the living standards of the subsistence population involves mobilization and allocation of resources so as to reach a desirable bal- ance over time between the welfare and productive services available to the subsistence rural sector. 2. Mass participation requires that resources be allocated to low-income regions and classes and that the productive and social services actually reach them. 3. Making the process self-sustaining requires development of the ap- propriate skills and implementing capacity and the presence of institu- tions at the local, regional, and national levels to ensure the effective use of existing resources and to foster the mobilization of additional fi- nancial and human resources for continued development of the sub- sistence sector. Self-sustenance thus means involving, as distinct from simply reaching, the subsistence populations through development programs. The achievement of these aims requires consideration of the interaction of a variety of specific issues which have a profound impact on the design and performance of individual programs. These issues may be classified in three general categories: 1. National policies: i.e., land tenure systems, commodity pricing and marketing systems, wages and interest rate structures, and so forth. 2. Administrative systems: i.e., the degree of centralization-decentraliza- tion in the governmental structure. 3. Scope for institutional pluralism: i.e., the distribution of development responsibility among the normal government structure; semi- autonomous government structure; private, commercial, and tradi- tional institutions; and elective bodies. This study, however, was not meant as an analysis of policies and institu- tions per se. Nor was it intended with a view to influence changes in policies and institutions in the specific cases under consideration. Subse- quent to the conduct of the study, as a result of domestic sociopolitical fac- tors, policy changes have already occurred in some cases, as, for instance, in the case of the land tenure situation in Ethiopia or the collectivization in Tanzania. The purpose of the study has been to analyze, among other things, the effect of such overall policies and institutions on programs and to draw general lessons from these experiences for designing future pro- grams. The chapters are, therefore, organized around the issues faced in planning and implementing programs directed towards the subsistence sec- tor. Chapters 11 and 1II discuss development of the farming systems. Chap- RURAL DEVELOPMENT PERSPECTIVES 21 ters IV, V, and VI describe the organization and delivery of agricultural ser- vices. Chapter VII examines the issues related to the provision of social ser- vices. Chapter VIII analyzes the performance of the parallel administrative structures, and Chapter IX examines the performance of administrative de- centralization in Kenya and Tanzania. Chapter X points out the role of training in rural development programs. Chapter XI summarizes the find- ings of the study and points out their implications for the design and imple- mentation of future rural development programs. CHAPTER 11 Nature of the Production Systems 1: Some Key Influences on the Development of Farming Numerous interrelated choices have to be made in planning the develop- ment of subsistence agriculture. For instance, there are questions related to the appropriate production mix. Should programs involve promotion of a single, obviously profitable crop with an assured export market, as in the earlier export crop schemes; or should they emphasize development of the overall farming system as is being attempted in some of the more recent rural development programs? There are also questions related to an ap- propriate factor mix. Under what conditions may capital-intensive produc- tion processes be promoted? Choices with regard to capital or labor inten- sity are, of course, closely related to a production mix. Several other questions with regard to factor proportions also need resolution. For instance, under what circumstances does development of farming necessitate settlement of populations on new land as distinct from intensification of already cultivated land? And further, under what condi- tions should development of farming be brought about by concentration of financial and administrative resources in a few regions, as is being at- tempted in the integrated rural development programs; and when may equitable allocation of resources between regions be justified, as in a mini- mum package program? This chapter examines some of the factors that influenced choices re- lated to these issues in the programs reviewed, the effectiveness of the in- terventions adopted on the basis of these choices in fulfilling the pro- nounced objectives of the programs, and some of the unanticipated conse- quences of the interventions of past programs. As a background for discussion of these issues, Section 1 discusses some of the main factors that appear to influence the flow of labor input into tra- ditional African agriculture. Section 2 points out the close interaction be- tween the labor constraint and the food constraint, the effect of the joint food-labor constraint on farm-level efficiency and on rural welfare, and the implications of the joint labor-food constraint for the choice of a production mix. Section 3 discusses implications of the labor constraint for mechaniza- tion of agriculture and assesses the effectiveness of past efforts at mecha- nization. Section 4 discusses the nature of rural migration, its interaction PRODUCTION SYSTEMS 1 23 with land policies and labor supply, and the implications of these various factors for regional allocation of resources for agricultural development. Section 5 examines some of the conditions under which geographical con- centration or diffusion of resources may be necessary in early stages of de- velopment to ensure regional equity. Factors Affecting Labor Flow into Smallholder Agriculture Evidence suggests that, compared to Asia, labor availability in Africa is frequently a greater constraint to increasing agricultural productivity than is the availability of land.' As a result, technological innovations, which re- quire intense labor input, often may not result in the full realization of po- tential production increases. At the same time, however, there is some prima facie evidence of surplus labor in African agriculture. On an average, African smallholders devote only about 1,000 man hours per adult per year to agricultural activities, compared to up to 3,000 man hours in several Asian countries.2 Thus, there is the seeming paradox of a farm labor short- age combined with an apparent labor surplus. The explanation for this con- tradition lies in the way the labor time is allocated in African agricul- ture-i.e., (a) in the seasonality of labor usage; (b) in the allotment of time between farm work, nonfarm activities, and leisure; and (c) in the division of labor between the sexes. Because these three factors influence the amount of labor allocated to agricultural work at any given point in time, they frequently pose a constraint to increasing productivity, limiting the effectiveness of interventions introduced through programs. The effect of these three factors on the supply of labor to agriculture, therefore, needs elaboration. Seasonality of Labor Usage The seasonality of labor usage is perhaps the most important element in the labor constraint. For example, cotton farmers in Sukumaland, Tanzania, experience heavy demand for labor in January and February when their crops must be weeded and again during the harvest period in June and July. By contrast, little labor is needed from August through November.3 Norman observes a similar seasonality in labor use in northern Nigeria. Most agricultural work occurs during the wet season with peak labor requirements during June and July. The dry season, November through April, is mriarked by little farm activity.4 Due to the variability in 1. See, for example, D).W. Norman, "The Organisational Consequences of Social and Eco- nomic Constraints and Policies in Dry-Land Areas," paper presented to the Second Interna- tional Seminar on Change in Agriculture, Reading, U.K., September 9-19, 1974; and M.l. Kolawale, "The Role of Agricultural Labor in Nigerian Rural Development." paper presented to the Seminar on Rural Development, Ife, Nigeria, November 1973. 2. See John H. Cleave, Aftican Farmers: Labor Use in the Development of Smaliholder Agricul- ture (New York: Praeger, 1974), p. 34. 3. Ibid., p. 92. 4. D. W. Norman, "Inter-disciplinary Research on Rural Development: The Reru Experi- ence," paper prepared for a Development from Below Workshop, organized by the Overseas Liaison Committee of the American Council on Education and the Association for the Ad- vancement of Agricultural Sciences in Africa, Addis Ababa, Ethiopia, October 12-20. 1973. 24 DESIGN OF RURAL DEVELOPMENT ecological conditions, the seasonal labor usage, of course, varies substan- tially between regions.5 While the peaks and troughs in the demand for agricultural labor are pri- marily a result of climatic factors, cropping patterns obviously affect the specific profile of the labor schedule. To reduce seasonal labor bottlenecks, new crop mixtures have to be such as to permit flexibility in the timing of labor input and to allow the use of labor to be spread more evenly throughout the year. Such crop mixtures as, for instance, maize, legumes, and cotton in Sukumaland or sorghum, millet, cowpeas, groundnuts, and cotton in northern Nigeria, have proven popular in many parts of Africa. The response to new crops is, however, substantially influenced by the extent to which the new crops complement or compete with the existing cropping pattern and the seasonal labor use. Norman observes, for exam- ple, that farmers in northern Nigeria have responded quite favorably to the introduction of tobacco, because demand for the bulk of labor for tobacco production does not conflict with the seasonal demands of major crops in the area. Best results for tobacco production are obtained when seed beds are prepared in early July and the tobacco seedlings transplanted about six weeks later. Harvest takes place towards the beginning of October. By con- trast, there seems to be less scope for fostering tomato production, also being promoted in the same area. The tomato crop is quite demanding of labor time (approximately 296 man hours per hectare). It must be planted between May and June and harvested between August and October. The labor input is, thus, required at a time of traditional peak labor demand, ag- gravating rather than supplementing the existing demand for labor. The relatively greater ease in increasing tobacco compared to tomato pro- duction in northern Nigeria has, of course, also been a result of the higher returns to the former crop.6 The question of differing returns to different activities leads to the second important factor that influences the labor available to agricultural production-namely, the allocation of time be- tween farm and off-farm activities and leisure. Allocation of Time According to Cleave, adult family members typically work only twenty to forty hours a week on their land, even during peak months. On an annual basis they may spend only three to four hours a day on agricultural work-such as planting, weeding, or harvesting. Of the remaining daylight hours, on a yearly basis, often somewhat less than half are devoted to off- farm activities-e.g., income-generating odd jobs, domestic chores, social obligations, and so forth. Leisure takes up the remainder of the time.7 It is 5. See John C. de Wilde, Experiences with Agricultural Developmnent in Tropical Afiica, 2 vols. (Baltimore: Johns Hopkins Press, 1967) for an excellent description of the ecoclimatic condi- tions and farming systems in Africa. 6. Returns per hectare (excluding labor costs) are $222.30-S259.35 for tobacco and $172.90 for tomatoes. For a further discussion of these points, see D.W. Norman, "The Organisational Consequences of Social and Economic Constraints and Policies in Dry-Land Areas." 7. For instance, as Cleave reports, D. Pudsey found that, among fifteen male farmers in Kakangi, Uganda (1964-65), the typical work day was broken into 3.6 hours of farm work, 3.5 hours of off-farm activities, and 4.9 hours of leisure. Among nine male tea farmers in Kyarusozi, Uganda (1965), the pattern was 4.1 hours of farm work, 3.5 hours of off-farm ac- tivities, and 4.1 hours of leisure. It is interesting to note that the tea farmers devoted longer hours to their crops, presumably because the returns to tea production were attractive. It is PRODUCTION SYSTE.MS 1 25 clear that some slack time exists in the off-farm activities and leisure periods that can be transferred to agricultural work. It would be incorrect to assume, however, that the time-use pattern com- mon to smallholder agriculture is inherently wasteful or inefficient. Off- farm activities in subsistence agriculture are frequently of legitimate eco- nomic or social value, although the relatively large proportion of time spent on visiting neighbors and on traditional and similar activities may seem in- compatible with a highly productive farming system.8 The large amount of leisure also seems necessary because of the low caloric intake of smallholders and the often arduous nature of manual fieldwork, although systematic evidence is still very limited on this particular con- straint.9 Finally, the seemingly short period devoted to farm work is fre- quently justified by the marginal disutility of labor in technologically un- derdeveloped agriculture in which returns from labor are frequently too low to prompt further substitution of labor from nonfarm to farm activities.'0 The allocation of time between various daily activities is, however, quite flexible and, therefore, amenable to change, provided appropriate interven- tions are introduced. With men, an increase in agricultural activity tends to be at the expense of various off-farm activities, while with women the ten- dency is to sacrifice leisure time."1 One explanation for this pattern may be that in some cases women's chores and duties are less strenuous than the men's; hence, women require shorter recuperative periods. However, an equally if not more irnportant reason seems to be that male nonagricultural work is frequently less critical than are the typical female household responsibilities, allowing men to transfer their time from off-farm to farm work more easily than the women.12 The important fact is that there is con- siderable scope for increasing the labor input to agriculture, though "the ex- istence of a range of [competing] demands on time... .raises the opportunity cost of agricultural work and reduces the readiness with which the flow of labor into agriculture can be increased."'3 Significant shifts uin the daily work pattern can, therefore, be brought about only where profitable innovations are available that increase the marginal productivity of agricultural labor sufficiently to offset the attrac- tion of other household, social, or income-earning activities. There are numerous examples that support this assertion. For example, farmers in Shinyanga district in Sukumaland, where farm income is about $99 per hec- tare, devote approximately 160 hours to their crops during the busiest month of the season. Farmers in nearby Ukerewe district, where income per hectare is only about $52, devote a mere 45 hours to farm activities dur- ing the peak month.'4 Cleave's summary of labor use patterns in a number also noteworthy that the tea farmers seem to require less leisure time. See Cleave, African Farmers: Labor Use in the Development of Smaliholder Agriculture, pp. 152-56. 8. Cleave points out triat ceremonial duties, often considered an important drain on pro- ductive activities, are scheduled for periods of low labor demand. See ibid., p. 180. 9. Ibid., pp. 178-79. 10. John Mellor, "The Use and Productivity of Farm Family Labor in Early Stages of Agri- cultural Development," irn Journal of Farm Economics, vol. 45, no. 3 (August 1973), pp. 517-34. 1 1. Cleave, Aftican Farmers: Labor Use in the Development of Smailholder Agriculture, p. 180. 12. Belloncle and Gentil point out the many competing demands on the women's time aris- ing from their household and productive chores. 13. Cleave, Afnican Farmers: Labor Use in the Development of Smallholder Agriculture, p. 180. 14. Ibid., p. 20, Table 2.2 and p. 124, Table 5.2. The figures are for the late 1960s. 26 DESIGN OF RURAL DEVELOPMENT of areas in Africa also indicates that the labor input to agriculture tends to be higher in areas where commercial crops make up a large proportion of the crops grown.15 Agricultural innovations must, of course, also be profitable relative to in- come-earning nonfarm activities if they are to lead to a greater labor input to agriculture. For example, although the introduction of tractors among maize farmers surveyed in western Nigeria resulted in a net saving of 12.4 man days per hectare, almost all of the time saved was spent on nonfarm activities like trading, carpentry, and tailoring.'6 Evidently, the marginal return to farm labor was less than the return to other kinds of labor. Sexual Division of Labor The division of labor between men and women is the third major factor that influences labor availability in smallholder farming. In traditional African agriculture there frequently tends to be a marked distinction in the roles of the sexes, although this division is bewilderingly variable in various parts of Africa. Cash crops and heavy work such as plowing have tended to be a male responsibility. Food crops and lighter work like weeding are more often the women's charge. Women are also responsible for the bulk of the domestic work-such as food preparation, wood and water gathering, and child rearing. The total amount of labor input by sexes is often quite dis- proportionate. A recent article by the United Nations Economic Commis- sion for Africa observes that women provide 60 to 80 percent of the agricul- tural labor used in parts of Africa. According to ECA, the case of the West Lake region in Tanzania is typical: men spend about 1,800 hours annually on agriculture while women spend about 2,600 hours.'7 Cleave cites numerous similar examples which suggest that, compared to men, African women often work longer both in farm and nonfarm activities.'8 This ob- servation is also strongly supported by Belloncle and Gentil in their review of the ZAPI and SODENKAM programs in Cameroon. A rigid division of labor on a sexual basis in which men refuse to perform "women's work" therefore places a limitation on the amount of labor available for agricul- tural production. There is a good deal of evidence, however, that the sexual division of labor is flexible and may be breaking down in many areas. Where new tech- nologies have been introduced, it is not uncommon to find men and women 15. For example, in southwest Nigeria (c. 1967) male farmers in the Alade area, where com- mercial crops account for 74 percent of the crops grown, spend about 1,610 hours per year on agricultural work. By contrast in the Akumazi area, with 39 percent commercial crops, males spend only about 730 hours a year on farm work. Of course, the trend towards increased labor in response to commercialization is only general in nature; and there are many exceptions. For instance, in the same part of Nigeria, male farmers in lloffa devote 2,135 hours a year to agricultural work, though only 32 percent of their crops are commercial. Obviously, other fac- tors, such as soil and climatic conditions, are also important determinants of the amount of labor time that is devoted to farming. See ibid., p. 32, Table 3.1. 16. M.I. Kolawole, "An Economic Study of Tractor Contracting Operations in Western Nigeria," Ph.D. dissertation, Cornell University, 1972. 17. See United Nations Economic Commission for Africa, Human Resources Develop- ment Division, "Women: The Neglected Human Resources for African Development," in Canadian Journal of Afiican Studies, vol. 6, no. 2 (1972), pp. 359-70. 18. Cleave, Afiican Farmers: Labor Use in the Development of Smallholder Agriculture, pp. 171-73. PRODUCTION SYSTEMS 1 27 performing the same. tasks during periods of intense labor requirements. For instance, the review of KTDA points out that much of the hired labor in smallholder tea cultivation in Kenya is female. In areas where a market exists for food crops., men seem to be willing to assist in food production. Nevertheless, as indicated by ZAPI and SODENKAM, when women are expected to do a disproportionate share of the agricultural work, improve- ments in farm technology may have the unforseen consequence of increas- ing the already heavy labor burden on women.19 Steps that lighten the household tasks of women should, therefore, facilitate labor availability for farm work. Fo' example, innovations, such as improved food preparation techniques or thve introduction of wheelbarrows for more efficient water carrying, can reo.i-e the time required for domestic tasks signifi- cantly.20 These que: 'ns are discussed further in Chapters IV and VII of this study. To summarize, the labor problem in African smallholder farming is largely an allocative problem. New cropping patterns and labor-saving in- novations can facilitate reallocation and more efficient use of labor, pro- vided such interventions are based on an adequate understanding of the complex economic, sociocultural, and physiological factors that determine the labor utilization pattern in traditional agriculture. The Food Constraint and Specialization in Agriculture To what extent were these and other factors that influence labor avail- ability taken into account in planning the commodity projects? This section will illustrate that labor availability in smallholder agriculture is also closely related to the desire of subsistence producers to ensure domestic food needs. The food constraint can be attributed to two factors frequently noted in the traditional African agriculture: (a) the generally high risk and low profitability of food technology, which ties up a substantial amount of labor in food crop production; and (b) the fragmented market systems for food crops which necessitate priority on food production to ensure supply for domestic consumption. The export crop projects reviewed under ARDS il- lustrate how these two factors can restrict the amount of scarce labor that can be released for the production of more profitable export crops, affecting efficiency in resource use at the farm level. Relatedly, the reviews also indi- cate how promotion of export crops in the absence of a more profitable food technology may reduce the supplies available for domestic consump- tion, thus affecting rural welfare. 19. Also see Ester Boserup, Women s Role in Economic Deielopment (New York: St. Martin's Press, 1970), p. 21, Table 1. Alternatively, some technological innovations may decrease the role of women in agriculture and, hence, their participation in economic development. 20. Vail estimates that traditional head portage of family water in Tanzania typically re- quires a labor-time input of 312 hours per year. The introduction of a $10 wheelbarrow, which can carry a much larger quantity of water, results in the reduction of labor time of 208 hours per year. If the time saved were transferred to farm tasks, the resulting net increase in agricul- tural production of perhaps $20 would more than cover the cost of the wheelbarrow. See David J. Vail, "Technology for Socialist Development in Rural Tanzania," mimeographed, n.d., background paper for the Seventeenth Annual Meeting of the African Studies Associa- tion, Chicago, November 1974. 28 DESIGN OF RURAL DEVELOPMENT Export Crop Projects and the Labor-Food Constraint The commodity projects have frequently had a rather limited objective. Apart from aiming to increase export earnings of specific export crops, the narrow focus of the projects was prompted by a variety of other considera- tions. In addition to assured markets, profitable technologies existed for ex- port crops due to the substantial early investment in research.2" Innovations related to export crops were, thus, frequently far easier to pro- mote than those related to food crops. Further, as will be illustrated in Chapters IV through VI, organization of effective delivery of extension, inputs, and credit has been relatively less difficult, both administratively and financially, in the case of export crops. Not only was it easier to recover costs of delivery of services provided for export crop production, but it was also easier to mobilize substantial addi- tional resources through a combination of pricing and marketing policies. It must be emphasized, therefore, that the deficiency of an export crop ap- proach may frequently lie in its failure to evolve adequately over time to meet the needs of the smallholder producer effectively and not in its initial concentration on highly profitable export crops. The diverse needs of the small farmers and the necessity to broaden the scope of services to meet such needs is quite evident in the commodity project areas. For example, in the tea-growing districts of Kisii and Kericho, where the Kenya Tea Development Authority has been in charge of pro- moting production of tea in the smallholder sector since the early 1960s, the increased incomes from tea production have stimulated considerable in- terest among the smallholder tea farmers in maintaining dairy cows. Although a high-grade cow cost as much as KShl,800 in 1973, purchases of dairy cows constituted an important item of expenditure among tea growers. However, institutional credit for dairy cows became available through the Department of Agriculture only in 1967 under the IDA credit program. By September 1972, of the KSh26.7 million in total loans dis- bursed, dairy cattle alone accounted for 40 percent; livestock and comple- mentary facilities comprised 75 percent. In recent years KTDA has also provided extension for maize. However, the multiplicity of administrative channels has led to a number of difficulties in effective implementation of extension for activities other than tea. These are discussed in Chapter IV. ZAPI in Cameroon provides a similar example of the diverse needs of small farmers. Belloncle and Gentil observe that the very careful analysis of the farming system in the south central ZAPI by agronomists of the Institut de Recherches pour l'Agronomie Tropicale (IRAT) has consistently pointed out the steps necessary to improve productivity of the food and livestock activities. Despite a few minor efforts in this direction discussed in later chapters, the main emphasis of the projects, however, continues to remain on the production and export of cocoa. 21. According to figures compiled by the Organization for Economic Cooperation and De- velopment, 81 percent of agricultural research expenditures in Africa went to export crops and only 14 percent to food crops in 1961. By 1971 the proportions had become approximately equal with 45 percent of research expenditures on export crops and 47 percent on food prod- ucts. Samuel Kassapu, "Depenses de Recherche Agricole en Afrique," unpublished memo- randum (Paris: Organisation pour Cooperation et Developpement Economique, September 14, 1973), p. 46. Table I. PRODUCTION SYSTEMS 1 29 The earlier agricultural policy measures in Sukumaland, Tanzania, were also oriented mainly to production of an export crop (in this case, cotton). Since late 1940s there has been tsetse fly clearance, provision of domestic water points, establishment of a cotton research center and a cotton-mar- keting infrastructure, and provision of extension services to encourage adoption of improved techniques in cotton production. During the period of cotton expansion, there occured a considerable shift in food production from sorghum to maize and, more recently, from maize to rice. For in- stance, between 1950 and 1960 production of maize per family had in- creased from 0.96 quintals to 5.62 quintals, while that of sorghum had declined from 8.85 quintals to 1.2 quintals. However, the total production of foodgrains per family declined from 15.81 quintals to 9.18 quintals during the same ten-year period. If innovations are profitable and involve low risk, participants in the ex- port crop projects respond as favorably to the introduction of services for food crops as they have to livestock development in the tea areas of Kenya. This is indicated by the experiences of the tobacco schemes in Urambo and Tumbi in Tanzania. Though initiated in the early 1950s, it was not until 1964 that the schemes provided services for food production. Since then, credit has been made available to smallholders for purchase of seed and am- monium sulphate for maize and paddy. According to the Urambo Farmers Cooperative Society's annual crop realization accounts, from the 1966/67 period to the 1969/70 period, hectarage under paddy and maize in Urambo increased from 132 hectares to 896 hectares and from 931 to 1,058 hectares, respectively. In the Tumbi tobacco scheme between 1965 and 1970, market- ing of paddy increased from 5,440 quintals to 78,930 quintals. Marketing of maize grew from 480 quintals to 2,340 quintals during the same period. The Food Constraint, Farm-level Efficiency, and Rural Welfare Such interest in food and livestock development among participants in export crop projects rnay often be explained by their need to ensure mini- mum supplies of food for domestic consumption and the consequent restriction on increasing overall agricultural production that is often im- posed by their traditional labor allocation pattern. Collinson's analysis of the Sukumaland experience, however, highlights the complex interactions between labor allocation and risk minimization that frequently influence the adoption of new innovations. To maximize yields, the extension service recommends that cotton be planted in December and maize in January. The early planting of cotton results in higher yields, while late-planted maize responds favorably to fertilizer. Further, since maize is often a host for cotton pests, it is advantageous to plant cotton before maize. However, farmers have generally preferred to spread the planting of both crops over a five-month period from October to February. While this does not lead to the full realization of potential increases in per hectare yields, especially of cotton, staggered planting (a) allows greater flexibility in the use of labor time, (b) assures a continuous supply of maize over a longer time period, and (c) reduces the risk of a total failure of the maize crop due to temporary emergencies such as drought or disease. For these various reasons, von 30 DESIGN OF RURAL DEVELOPMENT Rotenhan observed that the Sukuma "prefer lower but more secure returns instead of maximum yields."22 A similar emphasis on securing domestic food needs is observed by Belloncle and Gentil in the SODENKAM settlements in Cameroon. They point out the inordinate amount of labor time spent by women in the pro- duction of food crops. Because productivity of food plots tends to be very low, women frequently have to cultivate several distant plots to ensure ade- quate food supplies for domestic consumption. Obviously, a low-risk tech- nology that will increase productivity of food crops is critical, not only for increasing welfare but also for releasing labor from food crop production for allocation to production of high-value export crops. Belloncle and Gen- til also note that, because of the extractive history of export crop produc- tion, attention to food crop development is critical if rural development ad- ministrators are to gain the confidence of the populations being reached through programs. A different manifestation of the food constraint is noted in the tendency among smallholders to sell inputs provided for cash crop production, as has occurred in the tobacco schemes in Tanzania. Smallholders in the tobacco areas shifted a large portion of their land and scarce labor from food crops to tobacco production.23 As a result, small farmers producing tobacco in Urambo began to experience a deficit in food supplies in the off-season. They, therefore, sold the inputs purchased from the Urambo Cooperative Society (in particular, the fertilizer they received for tobacco production) to large farmers in return for the much-needed food. The fertilizer sales affected not only yields of tobacco on small farms but also the repayment of credit borrowed for the purchase of inputs. Apart from its adverse effect on allocative efficiency, the neglect of food crop production may also have an adverse effect on nutrition of the rural population, as is suggested in some export crop schemes. For instance, as pointed out earlier, despite the substitution of high-yielding maize for sorghum, the overall production of food crops in Sukumaland declined sub- stantially. Consequently, purchases of food crops increased. However, field investigations carried out in Sukumaland and evidence 22. D. von Rotenhan, "Cotton Farming in Sukumaland," in Smallholder Farming and Smallholder Development in Tanzania, ed. Hans Ruthenberg, Afrika-Studien no. 24 (Munich: Weltforum Verlag, 1968), p. 60. According to von Rotenhan, cotton planted in January (about 30 percent of all cotton hectarage in Sukumaland) mav have yields only one-third as large as cotton planted in December. (See pp. 62 and 82.) Norman has suggested that in northern Nigeria, where a similar food-labor constraint limits cotton production, efforts be made to de- velop a late cotton that can be planted after food crops and still produce maximum yields. See Norman, "The Organisational Consequences of Social and Economic Constraints and Policies in Dry-Land Areas." 23. Although tobacco hectarage varied fairly substantially from one year to another, it reflected a general upward trend. Tobacco Hectarage Per Farmer (Urambo Settlement Scheme, Tanzania, 1956-70) Year 1956 1958 1960 1962 1964 1966 1968 1970 Hecarcage 0.40 1.40 1.64 0.91 1.10 1.34 1.22 1.18 Source: Data for 1956 and 1964 arebased on Linsenmeyer's interviews with WalterScheffler: for 1956-62, see government of Tanzania. 'Urambo Annual Report for 1962-63." Tanganyika Agricultural Corporation, n.d.; and for 1966-70, see government of Tanzania, 'Urambo Production Report," unpublished, Ministry of Agriculture and Cooperatives, 1970. PRODUCTION SYSTEMS 1 31 from marketing studies elsewhere suggest that the rural food market is fre- quently fragmented.24 Despite the very considerable spontaneous increase in exchange between rural households reflected in the houshold budget data analyzed by Collinson, there is also evidence to indicate that there are pockets of deficit-foocl regions which do not always receive adequate and timely supplies of food from surplus regions.25 The adverse effect of early specialization on rural food supplies is often reinforced by variations in climatic conditions that lead to considerable year-to-year variability in domestic food availability. This is because greater profitability of export crops frequently leads to only the minimum alloca- tion of domestic resources for the production of food crops. Von Rotenhan's study suggests that the shift of hectarage to cotton, combined with the substitution of maize that is considerably more susceptible to rain- fall than the hardy millet and the sorghum grown earlier, may have had such an adverse effeclt on domestic food supplies in Sukumaland.26 This is supported by Collinson, who observed that poor weather in the late 1960s brought a revival of sorghum and millet as well as a plateau in the upward trend in cotton production. Anderson, like Collinson, in his review of the CFDT cotton development scheme in Fana, Mali, also observed an adverse effect of export crop specialization on food availability in the program area and the consequent readjustment of the farming system to ensure domestic food supplies. As a result of cotton promotion, the ratio of food crop area to cotton area was less than 2:1 in 1971. With the failure in rainfall during the 1971/72 season, food crop production declined considerably. Farmers discovered that they did not have enough food to carry their families and the migrant seasonal labor through the off season. In the 1972/73 period the ratio of the food crop area to the cotton area increased 3:1. To maintain an adequate supply of food, the project's extension services also began to advocate that every hired laborer grow at least one-half hectare of millet for every one and a half hectares allotted to cotton. The problem of ensuring food supplies in rural areas is frequently exacer- bated by the pricing and marketing policies followed by governments towards food crops. Food crop pricing is particularly susceptible to sociopolitical influences that are often not conducive to domestic food pro- duction. The adverse effect of government policy on the diversification of agricultural production and, in particular, on food crops is evident in BDPA's groundnut scheme in Mali. Mali has been experiencing cereal deficits in the urban areas. During 1972 and 1973 the government imported large quantities of American grain. Office des Produits Agricoles du Mali (OPAM), the government agency charged with buying cereals to supply the cities, was unable to fulfill its an- nual domestic procurement target of 40,000 metric tons. 24. See the discussion on marketing and distribution problems in P.L. Raikes, P.R. Law- rence. and L.G. Saylor, D. Warner, "Regional Planning in Tanzania: An Economic View from the Field," Economic Research Bureau Paper 68:8 (Dar es Salaam: University College, 1968). 25. Also see Chapter VI of this study for further discussion on this question. 26. See von Rotenhan, "Cotton Farming in Sukumaland," p. 57. The Sukuma apparently prefer the taste of maize to cassava or millet. Thus, despite the greater reliability of the latter crops, they emphasize production of maize. 32 DESIGN OF RURAL DEVELOPMENT Since the government is sensitive to urban demands for stable food prices, it has seemed to be reluctant to raise the price of cereals. The wide gap between the cereal price and the prices of groundnuts and cotton reduces the incentive for farmers to grow more food crops than they need locally. Besides, neighboring countries offer much higher prices for cereals, attracting the surpluses that exist in the rural areas and promoting "illegal" channels of marketing. Further, the weakness of OPAM as a marketing organization makes it difficult to implement a price policy, even if an effective policy were devised. OPAM lacks the personnel and funds to enter the cereals market early. Private merchants buy up the stocks of cereals at low prices and resell their stocks in July and August as the prices rise. An analysis by the Minis- try of Production estimated that in 1970 such a "black market" carried on 58 percent of cereal sales and 67 percent of cereal purchases. To summarize, the initial concentration of administrative and financial resources on a single export crop may have been justified in many of the past commodity projects, given the limited resources available for invest- ment, the often low level of technology available for food production, and the frequent lack of price incentives necessary to increase cereal produc- tion. However, past experience also indicates that, in the long run, such concentration may be detrimental to equity as well as to allocative efficien- cy. Rural development projects need to devote increasing attention to diversified farming and especially to food crop production. It is important to stress, however, that development of farming systems may not necessarily mean promotion of farm-level self-sufficiency in food production. Nor must it mean generation of massive food surpluses through promoting early regional specialization in food production. The lat- ter strategy may have to be resisted in early stages of development, since both low effective demand and inadequate marketing systems frequently pose a constraint to disposal of surpluses, an issue discussed in detail in Chapter VI. What is often needed is the development of technology and the provision of generalized delivery systems-in particular, extension and in- puts for food and export crops as distinct from specialized delivery systems for export crops alone. The need for the improvement of marketing facilities, including the de- velopment of feeder roads and communications between rural habitats, also cannot be overemphasized. As pointed out by the example of Sukumaland, spontaneous exchange of food in rural areas is quite substantial as rural communications and food surpluses increase. The project management of LLDP in Malawi also observed that, in the absence of traditional trading channels, the project would have been burdened with large surpluses of maize and groundnuts not procured by the Agriculture Development and Marketing Corporation (ADMARC). Kinsey's estimates show that as much as 60 percent of the maize traded in the Lilongwe project area may have been handled through traditional channels. However, because the capacity of the traditional marketing systems and the effective rural demand for food are limited in the early stages, services for food crop development, in many instances, have to be provided over a wide geographical area and their intensity increased gradually to ensure regional equity. The geographical spread of agricultural services needs to be PRODUCTION SYSTEMS 1 33 accompanied by the development of an infrastructure to link markets among regions so as 1.o keep pace with the gradual increase in effective de- mand for food from within the rural areas and from surrounding regions. Such an approach has been adopted in the Minimum Package Program in Ethiopia. However, even in this there has been a considerable adverse effect on prices in the initial years of implementation, particularly in the areas where marketable surpluses are increasing rapidly. This is because construction of feeder roads has not made the anticipated progress. The regional dispersion or concentration of resources must also depend on the relative agricultural potential between regions and on the scope for population migration between regions. These questions of regional alloca- tion of resources will be discussed after examining the scope in subsistence agriculture for removal of labor bottlenecks through mechanization. Mechanization in Smallholder Agriculture One of the frequent solutions to the joint food-labor constraint has been the introduction of mechanical implements to increase the output per unit of labor. These implements vary from simple hand-powered devices to sophisticated engine machinery. The discussion below is focused on effec- tiveness of two important mechanized innovations, tractors and ox plows, that are linked to modern multicultivators and weeders and are frequently introduced through programs. It illustrates the nature of interactions be- tween mechanization and labor bottlenecks on the one hand and between mechanization and complementary innovations on the other. It is the inad- equate understanding of these interactions which frequently explains the failure of the attempts to mechanize traditional agriculture. Tractors Tractors have often been viewed as a symbol of modern agriculture in Africa and, consequently, have been promoted through many rural devel- opment programs. Though tractors have generally proven well adapted to large-scale Western-style commercial farming, their role in development of smallholder agriculture has been less clear for a variety of reasons. First, since the labor bottleneck in peasant farming is largely seasonal, tractor ser- vices are needed at only the few critical periods of peak labor demand. Sec- ond, tractors are most useful in field preparation but often are of little utili- ty in weeding or harvesting. Tractors may, therefore, only postpone rather than break the labor bottleneck while doing little to alleviate the problem of underutilization of farm labor at periods of slack agricultural activity. Under some circumstances tractorization has aggravated labor bot- tlenecks rather than relieving them. For example, mid-season weeding tends to be a highly labor-demanding activity in much of peasant agricul- ture."7 When tractors are used for land preparation but not for weeding, as 27. For example, in the Zaria area of northern Nigeria, Norman reports that 26 percent of total agricultural labor time is devoted to weeding, the bulk of which occurs during June and July. Similarly, Ruthenberg indicates that in Sukumaland in 1963 weeding made up 26 percent of the labor input for maize and sorghum production and 29 to 48 percent for cotton. See Nor- man, "The Organisational Consequences of Social and Economic Constraints and Policies in Dry-Land Areas": and Hans Ruthenberg, Farming Systems in The Tropics (Oxford: Clarendon Press, 1971), p. 71. Table 4.4. 34 DESIGN OF RURAL DEVELOPMENT frequently tends to be the case, tractorization increases weeding require- ments. This is because tractorization allows expansion of cultivated hec- tarage and, hence, of the area to be weeded. Since the mid-season weeding period is already a time of peak labor demand, an even tighter labor bot- tleneck may be created. Similarly, if land is prepared in advance of standard planting dates to make greater use of available tractor time, frequently weed growth becomes well established by the time crops are sown. This, too, cre- ates additional weeding requirements.28 Tractorization can, therefore, be successful only if the necessary comple- mentary innovations are introduced to alleviate the labor constraint, especially for weeding. These may take the form of additional capital in- put-e.g., herbicides or mechanical weeders. Or they may be new planting techniques which reduce weed growth or facilitate weed removal as, for ex- ample, the close planting of groundnuts in Operation Arachide or the ridg- ing of cotton in Sukumaland. However, perhaps the most useful step in this regard is to alter cropping patterns to spread the need for labor more evenly through the season. As pointed out earlier, the shortage of labor available to farming activities is largely allocative in nature. New crop mixes and restructured labor patterns can, therefore, provide farmers with the neces- sary extra labor availability to cope with the possible labor bottlenecks cre- ated by tractor use. Several other problems are frequently encountered in the use of tractors in smallholder farming, including (a) the high capital costs of the equip- ment and herbicides,29 (b) the high overhead involved in maintaining and servicing the underutilized machinery,30 (c) the administrative problems in sharing a limited number of machines among a large number of diverse users, and (d) the technical difficulties of employing tractors under certain topographical and soil conditions. Of course, none of the above problems is insurmountable. If tractoriza- tion is accompanied by substantial increases in productivity, the high capital and recurrent costs of mechanized farming can be justified. Mainte- nance problems can be partly alleviated by training local people in proper machine use and repair. Administrative problems as well as the need for scale may be corrected by the introduction of effective group or cooperative farming or through the use of cooperative or commercial tractor-hire ser- vices. And lastly, research can single out the areas most amenable to tractor usage in addition to providing insights into new ways of adapting mecha- nization to divergent physical conditions. Much of the past evidence, however, suggests that successful introduc- 28. Of course, tractors may permit an expansion of hectarage sufficient to outweigh the po- tential decline in per hectare yields resulting from inadequate weeding. This appears to have occurred in some of the sparsely populated areas of Sukumaland, where extensive tractor- assisted cotton farming has resulted in yields that are low on a per hectare basis but high rela- tive to the labor input. See de Wilde, Experiences with Agricultural Development in Tropical Africa, vol. 2, p. 430. 29. According to de Wilde, a tractor and a minimum number of accessories would have cost $4,000 to $5,000 in mid-1960 prices. Investment in a large number of such machines can place a severe strain on the limited foreign exchange reserves of a developing nation. See ibid., vol. 1, pp. 113-19. 30. Cleave estimates that tractors may require 1,200 revenue earning hours per year to break even. Tractors are frequently used only half as many hours. See Cleave, African Farm- ers: Labor Use in the Development of Smaliholder Agriculture, p. 201. PRODUCTION SYSTEMS 1 35 tion of tractors is difFicult in smallholder agriculture. The experience with mechanized farming in Sukumaland illustrates the problems. In 1964 the Tanzanian government introduced a cotton block scheme which was hinged on the use of centralized tractor facilities. The Sukumaland region seemed suitable for mechanization since much of the land is flat, the grow- ing season is short, and the peak labor requirements are high. Many farmers in the early 1960s were already hiring tractor services.3' Under the scheme, blocks of between 121. .5 and 202.4 hectares were to be cultivated by tractors. Individual plots of 1.62 hectares would then be allocated to farmers for cot- ton production. The farmers were to raise the crop under close supervision but would retain ultimate responsibility for field maintenance and harvest activities. The scheme never achieved the scale envisaged. Rather than the pro- posed ratio of 121.5 hectares per tractor, in the 1964/65 period each tractor worked an average of only 27.1 hectares. The success of the cotton blocks was linked to the assumption that intensive farming on the blocks would lead to an increase in yield per hectare more than sufficient to cover the high cost of the program. Yet, rather than the expected 540 kilograms of cotton per hectare in the 1964/65 period, yields averaged only 180 kilograms per hectare during that period. These yield levels were similar to those achieved by farmers not participating in the program. However, the cost per hectare was approxirnately six times as high as the cost of production of nonparticipants.32 By the end of the cotton block scheme's first season, costs of mechanical cultivation, fertilizer, and spraying (mostly aerial) amounted to TShl.13 million. Only about one-fifth of these expenses were recovered.33 The Sukuma experience indicates that careful attention has to be paid to ensure that in practice the increases in production resulting from the in- troduction of tractors will justify their high cost. Where labor is inexpen- sive, tractorization may only mean substituting high-cost capital for low- cost labor.34 Though there often is scope for mechanized input to farming, the introduction of rnanual or animal-powered equipment may be a more feasible alternative than tractors. The recognition of this fact is reflected in Tanzania's ujtmaa policy, which emphasizes self-reliance among small- holders and "less spectacular forms of mechanization" that are consistent with local resources.35 Despite such a recognition in principle, Vail's study notes that in practice the promise of tractor services is frequently used as an inducement for peo- 31. In the Maswa district 50 percent of the cotton farmers surveyed by Collinson in 1963 used ox plows on their farms, and 50 percent used tractor-hire services. See M.P. Collinson, "Farm Management Survey No. 3, Luguru Ginnery Zone, Maswa District," unpublished background paper, Ukiriguru. Tanzania, 1963. 32. According to Collinson, the low yields on the Cotton Block Scheme were partly the result of substitution of labor from cotton to food production. 33. In the Cotton Block Scheme's first year, less than 5,670 hectares were cleared, of which only about 3,240 hectares were planted. For a further discussion of this project, see de Wilde, Experiences with Agricultural Development in Tropical Africa. vol. 2, pp. 437-41. 34. Collinson estimates that a Sukuma cotton farmer could increase his net cash income 14 percent through an additional outlay of TShl 1.5 on tractor and labor costs. Yet the same addi- tional outlay, if spent only on labor, could increase net cash income by 26 percent. 35. See government of Tanzania, SecondFive Year Plan, July 1969-June 1974, vol. 1, Gener- al Analysis (Dar es Salaam: Government Printers Office, 1969), pp. 37-38. 36 DESIGN OF RURAL DEVELOPMENT ple to form ujamaa villages.36 Vail attributes this partly to the identification of tractor use with modernity, both by administrators and policymakers in Tanzania and partly to the government's desire to promote its ujamaa pro- gram rapidly among a sometimes-reluctant rural population. The alterna- tive of fostering intermediate technology of the kind discussed below also seems to be creating a new rural elite of artisans and craftsmen, many of whom seem to use their spare time and the government-provided equip- ment to produce and service farm implements for private profit.37 Such a development conflicts with the egalitarian principles underlying ujtmaa. Thus, though there may often be strong economic arguments for limiting the introduction of advanced technology such as tractors, compelling politi- cal considerations may constrain the use of alternate strategies. Particularly given that several difficulties are encountered in getting production under way in newly formed ujamaa villages, there is a real danger that such politi- cal considerations may lead to a repetition of past mistakes with regard to tractorization. Ox Plows The introduction of ox plows to increase labor productivity often seems an attractive alternative to tractorization. Since much of the necessary equipment can be produced and the oxen raised locally, this type of mecha- nization avoids strain on foreign exchange. The cost of investment in a pair of oxen and a plow is, of course, much lower than that of a tractor. This lessens the need for scale and alleviates the administrative difficulties fre- quently associated with tractorization. There are also significantly fewer maintenance problems in the case of ox plows. However, the use of ox plows faces several of the same difficulties en- countered with tractors. The first is the fact that, as with tractor use, ox- plow cultivation may aggravate the seasonal labor bottleneck. Von Rotenhan reports that farmers in Sukumaland using hoe cultivation devoted about 1,490 hours a year per hectare of cotton, of which 370 hours and 430 hours, respectively, were spent on land preparation and weeding. Farmers using ox-plow cultivation spent about the same amount of time per hectare (i.e., about 1,520 hours). However, among the ox-plow-using farmers, land preparation accounted for only 120 hours, while labor input for weeding increased to 700 hours. The time saved in the early part of the season as a result of ox plowing was, thus, more than compensated by the increased labor required for weeding later.38 Because of the heightened weeding bottleneck that may result from using draft animals for field prepa- ration, yields per hectare are sometimes no higher for ox-plow cultivation than for hoe cultivation.39 The major gain arises from expanded area under cultivation. 36. Vail, "'Technology for Socialist Development in Rural Tanzania." For a further dis- cussion of the problems encountered in forming uiamaa villages, see Chapter IX of this study. 37. Vail, "Technology for Socialist Development in Rural Tanzania." 38. The explanation for this fact is that Sukuma farmers do not use an ox-drawn weeder or row cropping in conjunction with ox plows. See von Rotenhan, "Cotton Farming in Sukuma- land," p. 75, Table 6. 39. See Hans Ruthenberg, Agricultural Development in Tanganyika (Berlin: Springer-Verlag, 1964), p. 185. PRODUCTION SYSTEMS I 37 Thus, there may be a need, as in the case of tractors, to introduce new crops or crop mixtures that can increase productivity through inserting flexibility in the timing of farm operations. The replacement of broadcast sowing with row planting can also reduce labor bottlenecks. And, of course, there is a wide scope for the employment of low-cost implements that com- plement ox plows. For example, in the BDPA groundnut scheme in Mali, farmers using traditional methods devote 194 man days per crop of groundnuts and millet. The introduction of ox-drawn multicultivators and seeders can result in a reduction of fifty man days in the labor time per crop. The additional use of a 100-kilogram ox-drawn cart can result in a further saving of twelve man days. Such an ox cart increases costs of mech- anization by only 19 percent. Ox-drawn seeders are another potentially effective complement to ox plows. The Tanzania Agricultural Machinery Testing Unit in Arusha has developed an effective inter-row weeder that can be produced at a cost of $7 to $15. This instrument can result in a two- thirds to three-quarters reduction in weeding time for cotton.40 Demand for such relatively simple farm implements is frequently sub- stantial in rural areas, particularly when the introduction of such imple- ments is accompanied by other yield-increasing agricultural technologies. In the BDPA groundnut scheme, the number of multicultivators in service increased from 89 to 466 between the 1970/71 period and the 1971/72 period. The number of seeders and carts also increased substantially. An- derson observes that an increasing number of farmers are attempting to ob- tain several sets of oxen and implements with the goal of expanding their cultivated hectarage. In Tanzania also, as Vail points out, ox plows and seeders are greatly in demand among farmers exposed to the new imple- ments. The demand for such intermediate technology indicates a strong poten- tial growth linkage between the agricultural sector and the small-industry and service sectors of the rural economy. Promoting use of oxen, plows, weeders, and carts can create employment for craftsmen, mechanics, and cattle producers through the multiplier effects arising from increased agri- cultural productivity and incomes as well as from increased rural industrial employment and incomes. However, as noted above, political and other considerations may limit the promotion of intermediate technology and thereby restrict these growth linkages. The promotion of draft equipment is, however, also limited by its cost, which, though far lower than motorized equipment on a per unit basis, still represents a considerable investment for farmers. Anderson estimates that the average cost of animal traction in the BDPA groundnut scheme is about $48 per year, including the purchase of the equipment, feed costs, and depreciation. The true costs are probably somewhat higher, since the imple- ments are subsidized, the amortization periods are quite long, and it is assumed that farmers will use young oxen that can be sold as beef cattle for more than their purchase price.4' Because the maximum projected revenue 40. See Vail, "Technology for Socialist Development in Rural Tanzania." 41. According to Anderson, a pair of oxen cost about S117. A multicultivator, seeder, and cart represent a combined cost of$160. Feed, upkeep, and taxes amount to about $23 per year. Because the oxen can presumably be resold at a higher price and because the equipment is amortized over a period of five to seven years, the annual cost for introducing animal traction is quite low compared to the initial capital investment. 38 DESIGN OF RURAL DEVELOPMENT per hectare of groundnuts for farmers following BDPA-sponsored tech- niques (early sowing, fungicides, fertilizer, and so forth) is about $70, the average farmer must grow at least an extra two-thirds of a hectare of groundnuts to cover the cost of the animal-drawn equipment. This should be possible because the equipment allows a reduction of about one-third in the man days required to cultivate a hectare of groundnuts. However, the revenue figures assume yields of 12 quintals per hectare. In the 1971/72 period average groundnut yields in the BDPA operation were only 8.1 quin- tals per hectare. At lower yields the economic viability of introducing animal traction becomes progressively more questionable.42 This is why even an intermediate level of agricultural mechanization and the related rural industrialization frequently cannot be successful without yield-in- creasing agricultural technologies. In summary, it seems that both tractorization and ox-plow cultivation have a potential to increase productivity of smallholder agriculture, pro- vided the appropriate associated inputs and innovations are introduced si- multaneously. However, in most conditions ox-plow cultivation may be preferable because of its relatively greater flexibility, lower cost, and greater growth linkages with other sectors of the rural economy. Migration and the Development of Farming Systems The availability of labor in smallholder agriculture is also strongly in- fluenced by demographic patterns and intra-rural migration which, in turn, are related to the relative return to investment in agriculture in various regions. Surplus and underemployed labor may be attracted to areas with profitable agriculture and obvious labor bottlenecks. An influx of popula- tion to such areas may lead to a rapid increase in production and employ- ment. Conversely, areas with unattractive income and employment oppor- tunities may suffer a net outflow of population, creating labor shortages and economic stagnation. Migration may be temporary or permanent and, in either case, may frequently add substantially to the income of the migrants. On the other hand, a rapid population influx may also lead to a decline in wages and aggravate servitude, rural poverty, and social tensions in the regions of immigration. Intra-rural migration, therefore, poses yet another set of difficult ques- tions for rural development policy: Should programs be concentrated in areas of high potential to promote employment and attract migrants; or, alternatively, should resources be spread more equitably between regions to reduce inter-regional disparities and discourage migration? There can be no simple solution. In each case, the choice of policy must depend largely upon social and political realities, in particular upon the distribution of land rights among tribes and different economic classes and upon the government's 42. There is additional evidence to indicate that the cost of animal-drawn cultivation is often high, relative to return. Collinson estimates that at Usmao, Sukumaland, the relative cost of ox-plow cultivation is 17 percent greater than for hoe cultivation by hired labor. Besides, the quality of the work performed by ox plows is usually lower. However, relative costs for ox-plow cultivation were still 45 percent lower than for hired tractor services. Cited in Ruthenberg, Agricultural Development in Tanganyika, p. 184. PRODUCTION SYSTEHMS 1 39 willingness to correct inequities.43 Existing patterns of migration are, of course, also of considerable importance in determining policy. For exam- ple, does migration involve settlers or transient laborers; is it into areas of high or low population density; and is it spontaneous or the result of con- scious government policies? Finally and importantly, it is necessary to con- sider the relative cost of developing various areas. Is an area underpopu- lated because the land is of low potential or merely as a result of historical accident? Can it be opened up to increased migration through relatively simple steps, like the elimination of trypanosomiasis and malaria; or would there have to be heavy additional investments in physical infrastructure such as roads, irrigation, and soil conservation measures? The experiences related to migration in five countries illustrate the immense complexity of the issues and the difficulty in arriving at a few straightforward guidelines for planning rural development programs. These experiences, however, do provide useful insights as to the type of information needed for assessing the possible choice of interventions. Tanzania Population movernent in Sukumaland, Tanzania, over the last twenty years illustrates the ease in improving rural living standards when migra- tion involves movement to underpopulated areas of high potential. Tradi- tionally, the Sukuma were organized into independent chiefdoms. Inter- necine warfare led to demographic concentration, as large border areas sep- arating chiefdoms were too insecure to be farmed and, therefore, left un- populated. The continued abandonment of these "no man's lands" was re- inforced by the prevalence of the tsetse fly. These restraints on territorial expansion were largely removed during the colonial period. The govern- ment's tsetse fly clearance effort, combined with a decrease in particularist sentiment, encouraged the spontaneous settlement of many new areas. As a result, despite rapid population growth, the average farm holding in Sukumaland remained fairly constant between 1950 and 1970.44 The Sukuma were fortunate in that the underpopulated areas to which they moved tended to be of relatively high potential. Between 1955 and 1970 farmers in the newly settled areas were able to increase their cotton yields per square kilometer by 346 percent as opposed to an increase of 113 percent in the older settled areas.45 Out-migration is, of course, limited by the availability of empty land. Besides, population pressure may soon lead to a decline in productivity and living standards. The situation in the Ukerewe district in Tanzania is ex- 43. One of the important factors explaining the Tanzanian government's emphasis on regional equity in the allocation of resources and on the discouragement of private enterprise in the cash crop sector is a dislike of the social implications of the employee-employer relation- ship in the agricultural sector 44. Tanzanian census figures bear out the magnitude of out-migration in Sukumaland. Be- tween 1948 and 1967 population in the older settled areas increased by only 1.3 to 1.5 percent per year. By contrast in newly settled areas, population growth rates averaged 3.2 percent and ranged as high as 5.4 percent. See government of Tanzania, Recorded Population Changes, 1948-196 7 (Dar es Salaarn: Central Statistical Bureau, n.d.). 45. According to Collinson's evaluation of the available data on Sukumaland cotton pro- duction, it should be noted that cotton yields were still 2.5 times greater in the older areas (9.12 bales versus 3.62 bales per square kilometer), probably the result of more intensive cultivation. 40 DESIGN OF RURAL DEVELOPMENT emplary. With 178 persons per square kilometer, it is by far the most dense- ly populated area in Sukumaland (average 29.3 persons per square kilometer). Though its population increased relatively little between 1955 and 1970, cotton yields fell from 8.44 bales to 6.41 bales per square kilometer. In 1967 per capita net value of production was only TSh99, com- pared to TSh378 in the more sparsely settled Shinyanga district.46 Collinson observes that Ukerewe may be a harbinger of future trends in Sukumaland. Kenya Although population movement in Kenya has also been marked by out- flow into sparsely settled areas, intra-rural migration in Kenya presents a sharp contrast to the pattern experienced in Sukumaland. It illustrates the overriding importance of tribal and political factors which must receive at- tention in planning rural development programs. Whereas the Tanzanian government's land policies have encouraged population movement in Sukumaland, Kenyan policies have frequently tended to discourage a high rate of out-migration. Consequently, much illegal migration and settlement has occurred in Kenya with antecedent social, economic, and political effects. Like Tanzania, Kenya's pre-colonial period was marked by communal territorial claims and fluid boundaries between tribal homelands. Fixed property rights of the European type were first introduced during the co- lonial era, when supposedly unoccupied areas were classified as Crown lands. Much of this land, the so-called scheduled areas, was distributed to white settlers. The remaining land was set aside as native reserves and allo- cated to various tribes. According to Mbithi and Barnes, these policies resulted in the loss of much of the best land to Europeans and in the dis- placement of many indigenous communities.47 Further, traditional flex- ibility with respect to land usage was disrupted, threatening the ecological balance between man and his natural environment. As a consequence, Kenya's demographic pattern is marked by pockets of concentration which are unrelated to the carrying capacity of the land. Much of the high-potential land, particularly in the former White High- lands in Rift Valley province, is relatively unpopulated.48 Large portions of the former native reserves, which are often of low potential, are too 46. These problems seem to be connected to three factors: (a) Though holding size has re- mained constant, the number of persons supported on each holding has increased. (b) Pressure on the land means that a smaller proportion of the holding can be left fallow. In low- density Shinyanga an average of 65 percent of each holding was left fallow (1967); in high-den- sity Ukerewe the proportion was only 49 percent. (c) Similarly, a large share of the holding must be devoted to food crops-e.g., 32 percent in Ukerewe as opposed to 18 percent in Shi- nyanga (1967). Figures are based on Collinson's evaluation of available data. 47. According to Ruthenberg, however, "the greater part of the high-potential land-the estimates vary between two-thirds and more than four-fifths-and definitely almost all the best land remained available for the Af-rican peasants and herdsmen." See Hans Ruthenberg, A.fican Ag,icoltn,a/ Development Policy in Kenya, 1952-1965 (Berlin: Springer-Verlag, 1966), p.3. 48. High-potential land is defined as those areas receiving more than 86.4 centimeters of rain per year, medium-potential as areas receiving 61 to 86.4 centimeters, and low-potential as areas receiving less than 61 centimeters. According to Mbithi and Barnes, the first two catego- ries make up 11.5 percent and 5.5 percent of Kenya's land area, respectively. Von Pischke re- ports that about half of' the former scheduled areas receive more than 101.6 centimeters of rain per year. PRODUCTION SYSTEMvS 1 41 crowded in terms of the land's supportive capacity. This is particularly true in the areas surrounding the Highlands, such as the districts of Kakamega, Kisii, and Murang'a. Still other reserve areas, such as Naruk and Samburu districts, are underpopulated, largely because tribal animosities constrain the settlement of outsiders. And, of course, many people whose ancestral homelands were expropriated are without legal claim to any land. Popula- tion pressures in crowded areas squeeze additional people into landlessness. Except for removing white hegemony in the former scheduled areas, the present Kenyan government, which inherited this inequitable system of land distribution, has riot made a significant effort to correct the imbalance through a systematic land distribution policy.49 Indeed, the government's policies in the agricultural sector may be exacerbating the rate of landless- ness. In achieving rural development, unlike Tanzania, Kenya has explicitly opted for the Western model of private capitalism and individual entrepreneurship. According to Mbithi and Barnes, land registration pro- grams and progressive farmer schemes have hastened the decline of com- munal property and other traditional social arrangements. Unequal access to land through these policies has been accompanied by unequal access to new inputs and, in particular, to credit.50 Consequently, a new rural elite has emerged that has been able to expand its holdings and increase its incomes. At least partially the growth of this rural elite occurs at the expense of the remaining "less progressive farmers," many of whom join the growing ranks of the landless. In response to these realities, large numbers of land-hungry peasants have established illegal residence in unoccupied areas, especially in the central and coastal regions. Unlike the formal settlement schemes fre- quently tried in Africa, such spontaneous squatting is, of course, a relatively inexpensive ad hoc solution to the problem of malallocated land.51 Its viability, however, seems questionable. Though the land occupied by squat- ters is sometimes of good quality-such as in the Chylulu Hills bordering on Tsavo National Park-the squatters generally have limited technical knowledge of agricultural practices. This results in low productivity. Frequently, bush burning, charcoal making, poaching, and other exploita- tive practices are followed, thereby reducing the carrying capacity of the land. Where squatters are of different ethnic stock from the surrounding community, tribal tensions and conflicts are not infrequent. Finally, because the settlements are illegal, the settlers do not have recourse to gov- ernment services which could assist in their economic development. 49. The order preventing natives from owning land in the scheduled areas was revoked in 1961. Efforts to transfer land from European to African ownership were started shortly thereafter. Most important among these were (a) the Million Acre Scheme, (b) the British Land Transfer Assisted Owner's Scheme, and (c) the Stamp Purchase Plan. 50. The bulk of the credit extended by Kenya's Agricultural Finance Corporation has gone to land purchases and other large-scale uses. According to von Pischke, as of March 31, 1971, loans outstanding for land purchase and large-scale farming and ranching amounted to $25 million or about 88 percent of the loans outstanding. Though AFC has recently attempted to reach more small farmers, in financial year 1970 land purchase and development loans still amounted to 61 percent of the credit extended. See Chapter VI for further points on AFC's credit program. 51. For example, Kenya's Million Acre Scheme cost about 570 million between 1962 and 1970 or over $2,000 for each of the 34.000 families that were settled. According to Mbithi and Barnes, most of these settlers have not achieved their target incomes of $196-$280 per farm per year. 42 DESIGN OF RURAL DEVELOPMENT Mbithi and Barnes point out that, quite to the contrary, squatters are often harassed as trespassers and that, on occasion, their huts are burned by gov- ernment agents attempting to drive off illegal residents. Thus, in the ab- sence of an overall land policy, the squatters remain a smoldering-and perhaps an explosive-problem in Kenya.52 Ethiopia The crucial importance of land policy for rural development strategy is even more strongly evinced in Ethiopia where, until the recent political upheaval, feudalistic land ownership patterns and insecurity of tenure con- strained production increases and employment generation. Outside of the northern provinces-where communal land rights have prevailed-land ownership has been highly concentrated in the landed gentry, the Ethiopian Orthodox Church, and the central government.53 Although statistics on land ownership have not been very reliable, existing estimates indicate that until recently in these areas approximately 50 percent of the farmers were tenants, while perhaps 35 percent of the land area was owned by absentee landlords.54 The tenants were subject to arbitrary eviction, and in many in- stances the introduction of profitable farming techniques had led landlords to revoke tenant leases. For example, in CADU the introduction of tractor services, which increased the economies of scale in farming, allowed land- owners to expand their own hectarage, forcing large numbers of tenants off the land. Cohen estimates that in Chilalo alone, some 6,500 tenant house- holds may have been evicted between 1966 and 1972 as a result of trac- torization.55 By contrast, the tenants have had little incentive to adopt in- novations themselves because their rents were uncontrolled and any in- crease in productivity frequently led to a corresponding increase in rent. Similarly, there have been no rules concerning compensation for improve- ments made on rented land; so there has been little inducement for tenants to make capital investments in their holdings. Land policies and the associ- ated rural social organization have also affected access to agricultural ser- vices, including credit and extension. These questions are discussed further in Chapters IV and V. Despite frequent government policy statements concerning the need for tenancy reform, the landowner-dominated Parliament56 had been able to forestall effective remedial legislation. As recently as 1973 Parliament blocked enactment of a bill which would have restricted rents and secured 52. However, as Mbithi and Barnes point out, even a thorough land reform could not sup- ply enough land to support all those who want it. For example, they estimate that if the Rift Valley province were developed to its maximum capacity, it would be able to absorb only about 1.4 million immigrants bv the year 2000. Yet, within the same time frame, there is ex- pected to be an overspill of 12.5 million persons from the crowded Central, Western, and Nyangira provinces. 53. Approximately 70 percent of Ethiopia is considered arable. In 1972 the Church owned perhaps 20 percent of this arable land and the government perhaps 12 percent. See John M. Cohen, "Ethiopia After Haile Selassie," in Aflican Alfairs, vol. 72, no. 289 (October 1973), pp. 369 and 373. 54. Ibid., p. 370. 55. See John M. Cohen, "Effects of Green Revolution Strategies on Tenants and Small- Scale Landowners in the Chilalo Region of Ethiopia," in Jowrnal of DevelopingAAreas, vol. 9, no. 3 (April 1975). 56. This section was written before the political changes that occurred in Ethiopia in early 1975. PRODUCTION SYSTENIS 1 43 limited tenant rights to land and improvements. S7 Because of the political difficulties in alleviating the plight of tenants on privately owned land, in recent years increasing attention had been given to settling smallholders on unoccupied government land. However, powerful interest groups con- strained this alternative as well. In the past the government's vast hold- ings-perhaps half of the land area in Ethiopia-had been distributed on a patronage basis. An estimate indicates that between 1942 and 1970 the gov- ernment may have made land grants to private individuals totaling almost 5 million hectares. However, 80 percent of this land went to civil servants, military and police officials, and other elite, while only about 20 percent went to landless peasants or unemployed persons.58 Under the political structure that existed until recently, there was little impetus to change this system of land distribution. Thus, for instance, CADU, in attempting to relocate the large number of evicted tenants in Chilalo, was unable to find available government land, despite an estimated 40,000 hectares held by the government in the area.59 Nevertheless, there were a number of experiments to establish settle- ments on unoccupied government land. Many of these were in relatively marginal areas, however. For instance, the WADU settlements located in the lowlands of Abela and Bele are subject to drought, extreme heat, and periodic flooding.60 Though the WADU settlements are potentially produc- tive, a large capital investment is required for land clearance, drainage ditch construction, development of the infrastructure, and so forth; and the proj- ect must also provide food to the farmers during the initial period of settle- ment. These costs raise questions as to the establishment of such settle- ment schemes on a wide scale.6' Where settlement of unoccupied areas is an expensive proposition, the improvement of tenancy rights in previously settled areas is clearly a better economic alternative. In the absence of such land reform, minimization of settlement costs may warrant development of areas which, like the Kenyan squatter settlements, are already undergoing 57. Cohen, "Effects of Green Revolution Strategies on Tenants and Small-Scale Land- owners in the Chilalo Region of Ethiopia." The political situation has, of course, changed dra- matically subsequent to the mutiny and coup d'etat of 1974. 58. See Cohen, "Ethiopia After Haile Selassie." The bulk of the government land is located in the southern part of the country and was acquired during the territorial expansion of the empire in the latter part of the nineteenth century. 59. CADU was successful in relocating ninety-five tenants from Gobe to government land in Assassa. The move was required to make room for a livestock multiplication center. See J.D. MacArthur, The Development ol Policy and Planning lor Land Resettlement in Ethiopia, mimeographed (Bradford, U.K.: Project Planning Centre, University of Bradford, September 1972), p. 106. 60. Tecle notes that the prevalence of malaria was one of the primary reasons why these lowlands were unpopulated. After this disease was eliminated in the late 1950s and early 1960s. Wolamo tribesmen remained reluctant to migrate there, presumably because of their attachment to their traditional homeland. 61. According to the data collected by Tecle, preparation of the plots alone costs about $335 per farm family. Total Phase I costs for the settlement scheme amount to between $1,125 and S1,475 per farm family, depending on how many settlers are established. Only about half of these costs are considered recoverable. Other dry-land settlement schemes also seem to involve high establishment costs. MacArthur provides figures for settlements at Assassa, Bako, and Leka. On a per farmer basis, establishment costs are $280, $355, and $800, respectively. The Assassa figure is proba- bly too low, since the cost figures do not include surveying, land preparation, the construction of roads, or the provision of water supply. See MacArthur, The Development of'Policy and Plan- ning for Land Resettlement in Ethiopia, p. Ill, Table 4.1. 44 DESIGN OF RURAL DEVELOPMENT spontaneous settlement. However, even though this alternative is econom- ically desirable, it is frequently unpalatable politically. Malawi Population movement in Malawi is somewhat different from the pre- vious examples in that it often represents migration into areas that are al- ready heavily populated. Malawi has traditionally had a highly mobile pop- ulation. Kinsey reports that perhaps 13 percent of the Malawian population resides outside the country. Many Malawians have provided labor to mines in neighboring countries. Internal migration is also substantial. In 1966 the average district in Malawi was experiencing net out-migration of 1.1 per- cent, and much of this movement was to districts containing major rural development programs. Lilongwe-site of LLDP-is typical of these dis- tricts, experiencing net annual immigration at the rate of 6.7 percent.62 Ac- cording to estimates, in 1973 the population density in Lilongwe was over twice as great as the national average (119 persons versus 53 persons per square kilometer). Much of the available data suggest that Lilongwe offers favorable eco- nomic opportunities. The district is marked by moderately rich soil and am- ple water supplies. Kinsey reports that, even prior to LLDP, 58 percent of the population derived cash income from the sale of their own produce as opposed to the national average of 36 percent. Similarly, the proportion of persons with no cash income is significantly lower (28 percent) than the na- tional averdge (37 percent). The presence of LLDP has certainly added to the attractiveness of the area. A great deal of the recent population growth, however, seems most at- tributable to the construction of the new national capital at Lilongwe City. Recent estimates indicate that the population of the city is growing by 16 percent per annum with a projected population of 180,000 by 1980.63 The at- traction of this massive construction project can be explained partly by the fact that an unskilled worker in Lilongwe City can earn an annual income that is about three-fourths the average annual farm income in the area. However, many of these employment opportunities are temporary. If addi- tional employment opportunities do not develop in the area, large numbers of people may find themselves unemployed when the construction boom is over. There is also the danger of an impending food shortage. Kinsey esti- mates that the Lilongwe area must increase food production by 9 percent per year if adequate food supplies are to be produced locally for the growing urban population. Whether resources are concentrated in a few regions or spread more evenly must, thus, depend on the complex interaction of these and other factors discussed in later parts of this book. Mali The preceding discussion has concerned semi-permanent migration and settlement. Seasonal population movement is also common in Africa and 62. See government oi Malawi, Department of Census and Statistics, Malawi Population Census, 1966, n.p., n.d. 63. See government of Malawi, Capital City Development Corporation, Industrial Lilongwe, n.p., n.d. PRODUCTION SYSTEMS 1 45 poses an entirely different set of problems in administering rural develop- ment programs, as illustrated by the example of the CFDT cotton zone in Mali in which the number of transient laborers has increased dramatically in recent years. Though this population influx reflects a measure of success for the program, it has created complex problems in ensuring high produc- tivity and a steady food supply for the rural population. According to Anderson, the itinerant workers typically devote three to four days a week to a farmer's land. In return they are fed and given access to plots on which they cultivate cotton. At first this increases crop revenues in the program area; however, because the migrants tend to be inadequate farmers, their yields are low. Reaching these individuals through extension is also more difficult, since migrants usually leave the area within one to three years. Repeated cotton growing on the same land exhausts the soil, compounding the productivity problem. As noted earlier, the presence of seasonal workers in the project area also causes food shortages. In the Fana sector in the 1971/72 period, shortages were so severe that millet prices doubled. Such problems will continue as long as the influx of transient labor remains high and policies related to food production continue to be neglected. Regional Equity in Rural Development As the experiences in Tanzania, Kenya, Ethiopia, Malawi, and Mali indi- cate, the interdependent factors influencing population movement in Africa make policy decisions regarding rural development complex. Where the income-earning opportunities in an area are substantial and the political roadblocks are limited., efforts to stimulate population influx in specific regions through further investment of resources may be warranted. However, as illustrated by the spontaneous settlements in Kenya, too great an influx may also prove counterproductive. Therefore, care needs to be ex- ercised not to exceed the supportive capability of the local natural resources. In regions or localities where this has already occurred, it may be prudent to stimulate out-migration by investing resources in the develop- ment of areas with declining or stagnant populations. In all cases adequate attention to land use and land rights is essential for success. Each of these interdependent factors has to be considered in weighing the costs and bene- fits of investment in different geographic areas. Additional considerations related to regional equity are discussed in Chapters VII and VIII. CHAPTER III Nature of the Production Systems 2: Diversification of Production Activities In this chapter some of the additional issues faced in planning the devel- opment of the traditional farming systems are discussed. For instance, were the more recent rural development programs, such as CADU, WADU, LLDP, and SRDP, effective in achieving the goal of increasing overall agri- cultural productivity of the target population? To what extent are these pro- grams developing the technological capability necessary to identify and remove the diverse set of constraints arising from the complex interactions between the physical, demographic, and social variables that affect im- provement of the farming system? Alternatively, if the programs do not have a technological component, to what extent have they benefited from the research capacity being developed elsewhere? Lastly, to what extent have the areas had the benefit of profitable technological packages suited to diverse farming systems within program areas so as to promote equity be- tween various classes and regions and among farmers within a region? In addition to these various questions related to development of the farming systems, this chapter also discusses the problems faced in planning the de- velopment of the traditional livestock sector. Diversification in Integrated Rural Development Programs Table 3.1 shows the estimated increase in yields, hectarages, and produc- - tion of the principal crops in Chilalo Awraja, Ethiopia, prior and subse- quent to the establishment of CADU. CADU has a relatively well-devel- oped research unit carrying out adaptive research on a broad range of prob- - lems, including the adaptation of improved seeds and such simple labor-in- tensive farm implements as an improved plow, an animal-drawn harrow, a new hand hoe, and a stationary thresher. CADU is also conducting research on upgrading milch cattle. CADU's research unit has adapted three im- proved wheat varieties to suit ecological conditions in the Chilalo area. Ac- cording to the data collected by CADU's evaluation unit, wheat yields in Chilalo Awraja increased from 1.33 to 2.0 metric tons per hectare from 1966 to 1971. Average yields of beans have also increased to some extent. However, in the first few years, despite efforts at diversification, CADU's PRODUCTION SYSTEMS 2 47 major impact has been on wheat production. This has led to considerable wheat surpluses in the Chilalo area within three to four years after the proj- ect got under way. As a result, CADU has faced several problems related to marketing and pricing of wheat. The effect of these on adoption of innova- tions is discussed in Chapters V and VI. There appears to have been a decline in the production of milk in Chilalo in the 1971/72 period as shown by the rather sharp decline in the milk received at CADU's purchase cen- ters (Table 3.2). The decline has been attributed to the substitution of pastureland to cultivation of more profitable wheat. However, CADU has also attempted to improve the quality of local dairy cattle. According to Cohen, in the 1972/73 period the purchases of milk had increased by 70 per- cent over the previous year. He attributes this to CADU's sustained efforts on the dairy front.' Table 3.1: Comparison of 1966 and 1971 Crop Production, Chilalo Awraja, Ethiopia 1966 Crop 1971 Crop Yield Yield A rea Volume (metric tons Area Volume (metric tons (hectares) (metric tons) per hectare) (hectares) (metric tons) per hectare) Barley 67,900 108,565 1.60 79,300 126,880 1.60 Wheat 23,650 31,457 1.33 51,000 102,000 2.00 Flax 18,950 11,932 0.63 25,300 15,939 0.63 Peas 8,950 11,734 1.31 6,400 8,320 1.30 Corn 4,900 14,863 3.05 17,700 53,985 3.05 Beans 3,900 10,591 2.71 7,800 25,136 3.22 Teff 1,500 1,895 1.25 7,600 9,500 1.25 Solirce: G. Bergman and H. Lindqvist, CreditSituation in Chilalo Awrqia, CADU Minor Research Task no. 3 lAsella. Ethiopia Chilalo Agricultural Development Unit, July 1969); and CADU Crop Sampling Surveys. Table 3.2: Milk Purchased, in Liters, by CADU, Ethiopia, 1967/68-1971/72 Year Milk purchased 1967/68 4,000 1968/69 136,000 1969/70 318,000 1970/71 159,000 1971/72 147,113 Source: Government of Ethiopia, Chilalo Agricultural Development Unit, Marketing Divisional Files; and Cost/Benefit Analysis of CADUfor the Period 1967168-1973174 (Asella: CADU, 1971). WADU appears to have been somewhat more successful than CADU in increasing crop yields. The highlands of Wolamo in Ethiopia are suited to grow a range of crops, including maize, wheat, and teff. Two years after the establishment of WADU (i.e., in the 1970/71 period), maize yields per hec- tare in the highlands are estimated to have increased from 12 quintals to 24 1. Based on personal discussions with John Cohen. 48 DESIGN OF RURAL DEVELOPMENT quintals. The project appraisal team of the World Bank had predicted an in- crease of only up to 13 quintals per hectare in the highlands and up to 18 quintals per hectare in lowland settlement areas by the ninth year of opera- tion. Table 3.3 shows yields of the major crops in WADU. The number of farmers receiving production credit increased to 7,070 by the 1972/73 period, whereas the appraisal had predicted that the number would be 4,000 in that year. The existing data thus indicate that, in the first few years of im- plementation, the actual rate of return on investment in WADU may have been significantly higher than that expected. The rapid surpassing of the projected targets in Wolamo has been at- tributed by the observers of WADU to a combination of the very fertile Wolamo soils, the meticulousness of the Wolamo farmers, and the effec- tiveness of the extension system. Unfortunately, compared to CADU, WADU's technological research component has been modest and confined mainly to fertilizer trials. Therefore, how effectively WADU will tackle the problems arising from the introduction of more complex innovations in the future still remains to be seen. Table 3.3: A Comparison of Pre-project Forecast for Year 9 (Maturity) and Actual Year 2 Crop Yields in Quintals per Hectare, WADU, Ethiopia Yield Forecast at A ctual maturity a verage Crop Pre-project (year 9) (year 2) Highland Maize 8.0 13.0 20.0 Wheat 9.0 12.0 17.0 Teff 5.5 8.5 7.0 Lowland Maize 12.0 18.0 24.0 Cotton 1.4 5.6 10.0 Chilies 1.0 5.0 9.0 Source: Government of Ethiopia, Wolamo Agricultural Development Unit, Annual Repornlor Year 2 Pro- grams (Soddo: Ministry of Agriculture, May 1972). The Lilongwe Land Development Program in Malawi is aimed at increas- ing productivity of all the major crops grown in the area: namely, maize, groundnuts, and tobacco. During the first phase of program activities, maize production was expected to increase from 66,200 metric tons to 184,- 000 metric tons through increase in yields as well as in the area under culti- vation. However, the major economic return was to be realized through an increase in the yields and the production of high-value groundnuts. Because of the restriction on exports of tobacco, the project did not provide for an increase in tobacco production. Rather, the aim was to increase tobacco yields so as to permit diversification of hectarage to other crops. The project also aimed at increasing productivity of livestock. Data collected by LLDP's evaluation unit suggest that average maize yields during the first five years of the program, 1969/70 to 1973/74, conformed roughly to projected yields, although there has been substantial annual variability (Table 3.4). The hec- PRODUCTION SYSTEMS 2 49 tarage under improved maize increased from approximately 8,620 (1969/70) to 20,650 (1971/72). Highlighting the inadequacy of the pro- duction data, Kinsey points out that during this period total maize produc- tion may have been anywhere from 46,300 metric tons to 136,000 metric tons a year, depending on which of the available estimates af farm size, pro- portion of hectarage under maize, and yields per hectare are used in the cal- culations.2 However, LLDP's main shortfall seems to have been in groundnut pro- duction. Both the original and the second World Bank appraisal team en- visaged a far greater increase in the yield and the hectarage of groundnuts than seems to have been realized in the first six years. The available data suggest that during the 1972/73 period groundnut yields in the project area declined for the fourth consecutive year and were substantially lower than the yields outside the project area. If the shortfall in groundnut yields is ac- cepted and given the problems encountered in development of the livestock discussed later in this chapter, the rate of return on LLDP during the first six years may have been somewhat lower than projected. Table 3.4: Projected and Actual Yields of Maize and Groundnuts in Quintals per Hectare, LLDP, Malawi, 1969/70-1973/74 Maize Groundnuts Season Projected Actual Projected Actual 1969/70 14.1 11.5 5.9 6.2 1970/71 14.8 13.4 6.0 5.5 1971/72 15.5 15.0 6.2 5.2 1972/73 16.2 13.0 6.3 2.7 1973/74 17.0 13.1* 6.4 4.5* *Preliminary results. Source: Based on the Phase I Appraisal of LLDP and the data collected by LLDP's evaluation unit. More important, however, is the fact that, despite the attempt at diver- sification, LLDP's major success so far lies in increasing maize production leading to substantial increases in marketable surpluses. Consequently, LLDP, like CADU, has encountered many problems in marketing maize and in maintaining the necessary price incentive. Unless the pricing and 2. Kinsey notes that estimates of the total hectarage cultivated in LLDP vary by as much as 26 percent, while estimates of the proportion planted with maize differ by up to 21 percent. Depending on which estimates are used and in what combination, maize hectarage can vary by as much as 78 percent. Estimates of mean maize yield per hectare range between 11.2 quin- tals and 19.1 quintals. Thus, the available data provide estimates of total production that vary by 120 percent. LLDP's example is instructive in an important respect, especially as substan- tial data have been collected in this project. For most other programs the data for baseline and subsequent years are usually very much poorer. In such cases, judging the program mainly on the basis of precise realization of targets has the danger of diverting attention away from the more basic questions related to the project design, implementation, and reliability discussed in this study. There is no question that far greater attention needs to be devoted to collection of reliable data for the baseline and subsequent years to be able to judge project performance ob- jectively. At the same time, in assessing performance, the weight attached to the realization of short-run production targets that are frequently based on the poor initial estimates also needs to be reduced so as to examine the overall program performance in a proper perspective. This and the subsequent chapters outline in considerable detail a point of view as to what such a perspective may be. 50 DESIGN OF RURAL DEVELOPMENT marketing problems are tackled, it seems that it will be difficult for these in- tegrated projects to be able to continue the rate of expansion of cereal pro- duction that they realized in the first few years. The integrated programs, therefore, raise a dual question as to the efficacy of the marketing compo- nents and the prudence of generating massive supluses in small regions in a short period without regard to effective demand. These questions are dis- cussed later. However, to summarize, during the first short period of five years in which the integrated programs had been under way, the goal of diversification had not yet been fully realized. Development of Technological Capability for Diversification The difficulties encountered in achieving diversification are, of course, closely related to the efforts to develop suitable technologies. Although, in principle, the role of a profitable technological package is generally well rec- ognized, in practice most programs reviewed seem rather poorly equipped to develop an understanding of the interactions in the existing farming systems and the changes that are generated in the overall system by in- troduction of one or more innovations. An effective link of research units with rural development programs is one of the critical steps in adapting research results to particular farming systems. Where such effective research capability does not exist, it is neces- sary to allocate finances and manpower for developing appropriate tech- nological packages. Such research needs to be distinguished from the agronomic trials conducted on individual crops that frequently fail to take account of the system effects. Further, it is essential that there be a close link between adaptive research and extension. Far too often there is either not a profitable technological package that takes account of specific local constraints faced in applying re- search results to farming systems; or there is not effective training of exten- sion agents, making much of the extension effort ineffective. Often the ex- tension service also fails to appreciate its role in identifying the farm-level constraints which can be removed through research. There is, therefore, lit- tle emphasis on an adaptive approach to research so as to make the neces- sary improvements in the package with which the extension service is equipped. Apart from helping to create technological packages, local re- search stations need to facilitate a two-way flow between research and ex- tension. Admittedly, such research on multiple-cropping systems is complex; and the task of establishing an effective research-extension link is demanding, both in time and effort. However, its importance cannot be overstated, as is best illustrated by the evidence from the programs reviewed. The decline in groundnut yields in LLDP exemplifies the paucity of knowledge of the complex interactions involving agronomic and economic factors. Analysis of LLDP highlights two points. The ongoing research effort, conducted outside the purview of the program at the Chitedze gov- ernmental station, seems to have been inadequate in developing adaptive technological packages and in identifying and dealing with the constraints introduced by the program's other activities. Further, because of the em- PRODUCTION SYSTEMS 2 51 phasis on early realization of production targets, far too little attention could be given to developing the right package prior to its dissemination. According to Kinsey, based on a limited number of research trials, it was decided to issue 4.5 kilograms of sulphur dust to all farmers receiving credit for groundnuts in the 1970/71 season. The sulphur was intended to control leafspot. Before the results of this season were known, the appraisal report had opted in favor of supplying sulphur to growers who were expected to achieve yields of 147 kilograms per hectare. This recommendation was made, although evidence existed to suggest that such a recommendation was premature and probably inappropriate. The premature introduction of sulphur and the low rate of application stemmed from technical interrelationships which created a conflict of ob- jectives. On the one hand, sulphur was introduced because sulphate of am- monia fertilizer had been phased out and because it was expected that sulphur deficiency inr maize would become a problem. On the other hand, the rate of application was kept low because the relationship between soil pH and the use of sulphur was unknown and because it was assumed that heavy applications of sulphur would lower the pH. These technical prob- lems dominated all other considerations, including the effect on the pro- gram's credibility of promoting an innovation that may prove to be inap- propriate. In the 1971/72 period the groundnut credit package included 22.4 kilograms of sulphur for one hectare. The amount of sulphur was raised to 89.6 kilograms per hectare in the 1972/73 period. Kinsey observes that the effect of these rates of application on soil pH in the program area is still an open question and may explain the declining yields of groundnuts. It is also true, however, that a number of central African countries appear to be suffering from a similar decline in groundnut yields, in particular in areas where new maize varieties have been introduced. It has been observed that, given the very considerable labor required for weeding, harvesting, and shelling of groundnuts, the limited labor is probably being allocated to the production of high-yielding maize. Although these two explanations of declining groundnut yields have been put forward, the precise agronomic or economic factors that explain the decline are not known. And yet the knowledge as to whether the pH factor, the labor constraint, the relative prices, or some other factor is affecting groundnut yields is critical in deter- mining whether to undertake agronomic research; whether to avert labor constraint through a different crop rotation, mechanization, or both; or whether to change price relationships. Such problems are faced all too fre- quently in rural development programs. And yet, they have received relatively little attention in agricultural research. On the basis of evidence from their own and various other studies of agri- cultural research in Kenya, Mbithi and Barnes make a similar point in their review of the Kenyan squatter settlements. In Kenya the government re- search stations serve as the main source of information on maize tech- nology. These stations are usually self-contained institutions which have few or no links with the extension service. Research carried out on farms near Embu and Katumani research stations shows that adoption rates are even lower on these farms than those studied 2 to 100 miles away. The farmers feel that the research station is for the wazungu (Europeans); that 52 DESIGN OF RURAL DEVELOPMENT the stations use wazungu dawa (fertilizers), tractors, and aerial irrigation; keep too many books and records; and employ many people in their fields. Such an attitude implies that the farmers do not perceive the work on the research stations as being suited to their situation. A follow-up on some of the farmers who had attended a field day at one of the stations indicated once again how keenly they felt the lack of relevance of the large-scale operations at the research station to their small farms. The programs are not divisible or adaptable, and the emphasis is on optimal combinations of resources to maximize output without consideration of the scarcity of some of these resources at the small-farm level. Physical science research has been conducted on a very limited scale-e.g., exhaustive soil surveys and full-scale fertilizer trials have been carried out on only two soil types in Trans Nzoia. Mbithi and Barnes point out that, in addition to these gaps, research is lacking in a strong subsistence or a small-crop base. Little attention has been given to major food crops (such as pigeon peas and bullrush millet) and to small-scale cash crops (such as coriander, green grams, and castor). De- velopment of food crop mixes for different ecological regions has also been neglected in agricultural research. A similar situation exists in the SODENKAM land settlement in Cameroon. According to Belloncle and Gentil, to ensure that the results achieved on the Nkondjok research farm are suitable for practical adoption, the experimental conditions should faithfully reproduce the conditions under which the technical innovations developed are to be applied in the future. The Nkondjok farm is situated on a fine basaltic plateau, completely cleared, whereas the pioneers' farms are usually in the middle of a forest on fairly steep slopes and have very poor soils. Regardless of the results achieved on the program's Technical Support Center, one may well ques- tion how they can be adopted by the pioneers, considering the different and largely artificial conditions under which they have been obtained. The review further notes that usually there is no direct link between the Nkondjok experiements and SODENKAM's extension programs in the vil- lages. Nor has much attention been given to seeking solutions to the prob- lems faced by the pioneer villages. Rather than developing a coherent ap- plied research program, there has been a series of attempts entrusted to different research institutes with neither a clear definition of the objectives pursued nor an examination of their connection with SODENKAM's policy. This is particularly obvious in the testing of food crops, which have a fundamental bearing on the operation. The effort should obviously have been directed toward research into the types of rotation possible in the area so that the results could be made available to pioneers. In stock raising, too, the "experimental" program should have been initi- ated on the basis of the actual conditions in which the cattle would live in the villages; and the desired objective should have been clearly defined. It is certainly a pleasant sight to see the Ndama herd on the farm peaceably graz- ing on 12 hectares of pasture, but that does not seem to be a sufficient reason to continue on the course adopted by the program in the field of live- stock development. The failure of research to be oriented towards removing or adapting to critical constraints and, hence, to improving performance of the agricultural PRODUCTION SYSTEMS 2 53 systems has been observed repeatedly in the reviews of the uj6maa, SRDP, the Sukumaland cotton development, and several others, It has also been emphasized in the previous agroeconomic research on Africa, most notably by de Wilde in his earlier work commissioned by the World Bank.3 And yet development of effective national research systems remains one of the crit- ical gaps in realizing the objectives of rural development in Africa. Several steps need particular attention in future rural development pro- grams. There is a substantial need for an explicit commitment among na- tional policymakers and donors to the task of developing agricultural re- search systems. This commitment needs to be translated into action by allocation of substantially more finances and experienced manpower than in the past to carry out research and, in particular, to train the indigenous manpower in the various disciplines. Such effort is already under way at the international level in the form of the various international research in- stitutes.4 However, particularly because the ecological variability is so overwhem- ing in Africa, there is a need to make a conscious effort to develop a net- work of national, regional, and local research systems and to create an in- digenous manpower capability to operate such systems. Only then will the international effort be able to emphasize adaptive agricultural research. This is where a major gap now exists, despite the fact that a plethora of re- search institutions and experimental stations already exists in many African countries. Frequently these institutions are ill-staffed, ill-financed, and their efforts poorly coordinated with each other to deal with the many problems that extend across regional and national boundaries. Finally, the linking of agricultural research with action programs cannot be over- emphasized. It is only when an effective two-way dialogue between re- search and extension is established that the current rural development effort will have a noticeable impact.5 Farming Systemls and Intraregional Equity As evident from the preceding discussion, the objective of rural develop- ment programs may not be only to increase overall productivity in a region but also to increase participation of low-income farmers in agricultural pro- duction and incomes and thus to reduce income disparities between classes of farmers. To realize the latter objective, interventions have to be based on the knowledge of the variation in the resources within regions and between 3. See John C. de Wilde, Experiences with Agricultural Development in Tropical Africa, 2 vols. (Baltimore: Johns Hopkins Press, 1967). 4. In 1974 the World Bank allocated $34 million for support of international agricultural re- search, an increase from $15 million in 1972 and $23 million in 1973. In 1975 the World Bank hopes to spend $45 million for this purpose. For a summary of the activities of the institutions supported by the World Bank, see Consultive Group on International Agricultural Research, International Research in Agriculture (New York: World Bank/FAO/UNDP, 1974). 5. For a general discussion of some of the problems faced in African agricultural research, see St. George C. Cooper, Agricultural Research in Tropical A,fica (Nairobi: East African Literature Bureau, 1970). Concerning the role of interdisciplinary research, see D. W. Nor- man, "Interdisciplinary Research on Rural Development: The Reru Experience." Paper pre- pared for a Development from Below Workshop, organized by the Overseas Liaison Commit- tee of the American Council on Education and the Association for the Advancement of Agri- cultural Sciences in Africa, Addis Ababa, October 12-20, 1973. 54 DESIGN OF RURAL DEVELOPMENT classes and on the factors that explain the existing allocation of resources, levels of productivity, and variations in it. How incomes may vary substantially within a target area by crops is il- lustrated well by the Lilongwe Land Development Program. A study6 of the Tsabango, Chinyama, and Kabuthu areas in the Lilongwe district shows that the gross return per hectare of groundnuts and tobacco is, respectively, twice and three times that of maize. Given that tobacco growers tend to have holdings that are far above average in size, these returns result in dis- parities in incomes of up to 300 percent between tobacco growers and non- tobacco growers. This observation holds, irrespective of whether income variation is considered on the basis of a holding or on a per capita basis. Only 30 percent of the farmers in the current program area sell tobacco from 11 percent of the cropped hectarage. The distribution of income is, therefore, highly skewed in favor of this group. According to Kinsey's estimates for more recent years, there has not been a significant change in relative incomes of tobacco and non-tobacco growers subsequent to the establishment of the program. The variation in income is a result of differences, not only in crops grown, but in soil fertility within program areas. There is, of course, a close relation between soil characteristics and cropping patterns. In Lilongwe the fertile soils existing in some units are also the most suited to tobacco cultivation. These varia- tions lead to the difficult conflicts in achieving the program's objectives of growth and distribution. Kinsey's analysis illustrates the point. The mean cash value of production in the area covered by the program's extension service is some 22 percent less than in other areas planned for the program but as yet outside the current program area. In 1971 the geographi- cal variation in value of production was 94 percent. One area was 58 percent below the mean value, while another was 23 percent above. LLDP must raise the value of production by more than 50 percent in the less productive areas if the geographical imbalance in value of production is to be elimi- nated. The geographical imbalance in income also has important implications for the program's objective of raising the average farm family's net cash in- come by 70 percent. It is far easier to reach this target by concentrating in- puts and extension in the high-value production areas rather than in the low-value production areas. To a large degree the amelioration of these income differentials between tobacco and non-tobacco growers and, hence, between those areas well suited and ill suited to tobacco production is beyond the project's direct control, since the granting of quotas to grow tobacco is administered by ADMARC, the government marketing agency. Moreover, the principal to- bacco production occurs in areas where farm size is above average, thus per- mitting the growing of both food crops and tobacco. In areas where high population pressures have reduced the average size of holding, tobacco pro- duction has declined and is limited by the hectarage available, once food crop needs have been satisfied. 6. H.K.F. Hoffman, "Case Studies of Progressive Farming in Central Malawi. Report on a Socio-Economic Survery Conducted in Selected Areas of the Lilongwe Plateau" (n.p.: govern- ment of Malawi, July 1967). PRODUCTION SYSTEMS 2 55 If balanced, broad-based growth is an objective of rural development, the question arises as to whether, in parts of Lilongwe where farms are sub- divided and submarginal, the provision of agricultural credit and extension is the right strategy or whether an entirely different approach, such as the consolidation of holdings and/or the creation of off-farm employment, should be followed. Data on intraregional income disparities arising from differences in quality of physical resources7 are, however, too limited in most cases to per- mit such judgments in the course of planning. Often, even if the data were available, the prograrns are ill-equipped to incorporate such information in planning the strategy or in modifying it as new information becomes avail- able during the course of implementation. This is illustrated by the evalua- tion of SRDP's Vihiga Maize Credit Program carried out by Hay and Heyer of the Institute of Development Studies (IDS) in Nairobi.8 They point out the diversity in the Vihiga district in ecological conditions, in cropping pat- terns, and in population density. According to Hay and Heyer, the neglect of the diversity in physical conditions has led to an almost exclusive em- phasis on credit for maize. For example, there is substantial migration of male members as a result of availability of off-farm employment. Survey data show that 30 percent of the family heads were absent during the survey period. Consequently, despite the apparent high density of population, there is often considerable shortage rather than surplus of labor, affecting labor availability at peak periods of farming. The planning of the Vihiga Maize Program did not take into account this constraint and its effect on adoption. The amount of maize grown also varies substantially between farms. In fact, a considerable proportion of the land is allocated to crops other than maize, including coffee, tea, certified potatoes, sweet potatoes, beans, sorghum, and millets. The question, therefore, arises whether excessive emphasis on maize self-sufficiency is the appropriate strategy for either specialization or im- provement of income distribution and nutrition in Vihiga. The implications of the objective of maize self-sufficiency have also not been spelled out clearly in the strategy adopted. For example, it is not clear whether maize is to be promoted on each farm or whether self-sufficiency is to be realized in each area in the division or in the division as a whole. These various alter- native objectives and the way they affect strategy have substantial implica- tions for distribution of benefits among target populations within a program area. Hay and Heyer emphasize that, though considerable information became available on the Vihiga area, it was not utilized by the American planners who modified the initial plans after USAID decided to fund the Vihiga Maize Program.9 The constraints imposed by limited technology, manpower, and informa- tion currently available for planning rural development programs are real. 7. The disparities in distribution of benefits arising from institutional constraints (such as land tenure or access to inputs) as distinct from physical constraints (such as soil fertility or rainfall) are discussed in Chapters V, VI, and VI]. 8. F.G. Hay and J. Hever, "The Vihiga Maize Credit Package," in An Overall Evaluation of the Special Rural Development Programme, 1972, Institute for Development Studies Occasional Paper no. 8 (Nairobi: University of Nairobi, 1973), Appendix F, p. 5. 9. Ibid., p. F-I 7. Also see Chapter IX of this study for further discussion of effect of plan modifications on the program's performance. 56 DESIGN OF RURAL DEVELOPMENT Given these constraints, a very difficult dilemma is frequently encoun- tered. Should rural development programs involve use of substantial plan- ning resources at the outset to develop different development strategies so as to fulfill the objective of equity in a region, but probably at the cost of de- priving other regions of the country of these limited resources? Or, alter- natively, should initial efforts be concentrated on simple innovations, such as the use of improved seeds, fertilizer, row planting, and weeding, which can be introduced across the program area with a reasonable degree of modification? Considerations of equity on a larger scale may often necessi- tate the latter. It is apparent, however, that the program administrators ought to be cognizant of the likelihood of intra-program disparities, begin to collect the necessary information from the very initial stages of implemen- tation, and maintain enough flexibility to deal with the problem of dis- parities during the course of implementation. In subsequent phases of pro- gram implementation, the program may then develop more specific tech- nological packages for the groups that are otherwise likely to be bypassed by general programs. Livestock Development in the Traditional Sector Development of livestock may be as critical for improving rural produc- tivity and welfare as is improvement of the crop-farming system. In pastoral areas cattle raising has traditionally been a major activity, since limited rainfall has precluded crop production. Therefore, in many of these areas, livestock development is the only means of raising living standards. In non-pastoral areas stock raising is an effective complement to crop rais- ing. Improved planting techniques can provide increased supplies of fodder for livestock, which in turn can increase the dairy and meat products avail- able for domestic consumption. In addition, the animals provide manure for enriching the soil and can also be sold for cash.'" The program reviews indicate, however, that improvement of the pro- duction of livestock is beset with many difficulties, both in pastoral and non-pastoral areas. Livestock development involves a complex interaction of technical, economic, and sociocultural factors, leading to considerable uncertainty as to the response to innovations, a phenomenon best illus- trated by performance of the livestock components in the programs reviewed. In Malawi the increase in the domestic production of beef has not kept pace with the rapidly rising domestic demand." Introducing a livestock 10. Marticou, an agronomist working for ZAPI in Cameroon, pointed out that the "pigs, sheep, and goats which presently cause more damage than they bring in profits, could be the prime agents in agricultural progress. Instead of wandering around all day tearing up fields and plantations, they could be corralled under supervision, given food which, in any case, they would have taken in the fields, and the planters could take advantage of this domestication to gather the dung which is now lost. This would provide work for idle farm labor, would use the cassava which is at present frequently far beyond the farmer's needs,. . . and would constitute a new source of income." H. Marticou, Les Structures Agricoles dans le Centre Cameroun (Yaounde: Ministre de l'Agriculture, Direction des Statistiques, 1961). 11. Between 1966 and l971, domestic consumption of beef increased by 152 percent, while domestic production increased by only 41 percent. Government of Malawi, Economic Report, 1972, Budget Document no. 4 (n.p.: Office of the President and Cabinet, Economic Planning Division, n.d.). PRODUCTION SYSTEMS 2 57 component in LLDP, therefore, seemed to be a step towards improving domestic production of livestock as well as a means of increasing the cash incomes of smaliholciers. The World Bank's Phase 11 appraisal of LLDP noted that eighty aniimals per year were being stall-fed successfully in the program area and urged the creation of a ranch to increase the supply of feeder steers available to program farmers. Stall-feeding is a highly profit- able enterprise; gross returns per steer compare favorably with the returns per hectare of maize, groundnuts, or tobacco. Further, because stall-feeding is nonseasonal, farmers can earn income at times of the year when no alter- native employment is possible."2 LLDP started the [)zalanyama Ranch in 1970 to produce cattle for sale to non-cattle-owning farmers interested in stall-feeding. In addition, it was hoped that the existing cattle owners would be attracted to the new tech- nique of stall-feeding. The ranch, which covers an area of 65,182 hectares, was to acquire 2,500 head of cattle for each of the four years of its develop- ment. By 1974 it was intended to produce annually (a) 1,800 high-quality steers for fattening, (b) 700 improved heifers for either milk production or fattening, and (c) sorne 1,000 additional animals to be sold for slaughter. The ranch has not performed as expected. By the end of 1972, the cattle population was only 2,250 head, or less than half the projected number. The quality of the animals was poorer than expected. The rate of return has ranged between 4.5 and 5.0 percent. The ranch has been unable to provide many animals for stall-feeding. From October of 1971 to the end of 1972, less than 400 steers were sold or issued on credit. These shortcomings are explained by the ranch's inability to purchase the number of cattle antici- pated from local sources. Only 1,800 head were acquired during the first two years of operation, and these were more expensive than foreseen. The average price paid for cows in the local markets has been K43, 20 percent above the expected price of K36. The stringent cattle supply is largely explained by the complex interac- tion of the physical, economic, and sociocultural factors that influence the nature of the market in the traditional sector. When the World Bank's ap- proval team estimated the number of cattle that could be obtained from local auction sales, the project area had experienced a series of poor crop years. Farmers were selling off large numbers of stock to compensate for lost income. Based on1 this high sales volume, it was assumed that sufficient cattle would be available. After the ranch was established, however, there was a series of favorable seasons. Consequently, farmers have been reluc- tant to market cattle, preferring instead to build their own depleted stock. Further, LLDP competes for the existing cattle, not only with private buyers, but also with the West German project in Salima, which was estab- lished subsequent to the planning of the ranch. Cattle prices have been rising as a result of the shortage of available cattle and the competitive bidding between buyers. LLDP has been reluctant to pay the price commanded by the top-grade stock. Consequently, the limited number of cattle that the ranch has been able to purchase tends to be older 12. Gross margins for 320 of the credit steers issued by LLDP averaged K42 per head. Average cash receipts per grower in the project area from the sale of maize, groundnuts, and tobacco combined was K44 in the 1971/72 period. Thus, a smaliholder who fattened only one steer a year could potentially double his annual cash income. 58 DESIGN OF RURAL DEVELOPMENT and less productive. Apart from the desire to keep procurement costs low, according to the project management, maintenance of low prices is justiftied by the backward-sloping supply curve of herders. The available evidence suggests that Lilongwe farmers are reluctant to part with cattle except in times of economic hardship. Kinsey notes that most farmers sell or slaughter their animals only when they need immedi- ate cash. A 1972 survey conducted by LLDP's Livestock Section indicated that 60 percent of the program's cattle owners considered their stock to be primarily a form of insurance. Only 19 percent viewed cattle as an impor- tant source of income. Not surprisingly, the same survey found that, despite its profitability, only one percent of the 2,400 potential stall-feeders in the program area were actually being fattened."3 The program manage- ment has proposed further investigations to improve understanding of the supply response of Lilongwe farmers. The attitude toward cattle similar to that noted in LLDP is observed elsewhere in Africa, especially among nomadic tribesmen.'4 To rural people cattle is often not only an economic investment but a source of status and an important feature of the community's sociocultural activities. In the light of the economic and social significance of cattle and considering the high mortality rate in the herds, a desire to hold rather than to sell cattle may be natural in some circumstances.'5 It is, therefore, believed by LLDP management that, if prices of cattle rise, herders may be able to obtain cash with fewer sales or with sale of lower grade cattle. However, there is consid- erable conflict in the existing evidence as to the nature of the price respon- siveness of cattle herders. Some evidence indicates that, when incentive prices are offered for cattle, nomadic herdsmen behave like economic men.'6 A case, therefore, may be made for experimenting with higher prices to examine the extent to which farmers are willing to offer their cattle for higher prices. Further, since the project is a public sector investment, it is not clear whether the reduced benefits to LLDP as a result of the higher domestic cattle prices necessarily mean a net social loss. The higher prices will, after all, be received by local producers. The overall social benefit to the local population from higher prices may outweigh somewhat the disadvantages 13. Cossins noted similar attitudes toward cattie in the Shire Lowlands. The rural people kept stock (a) as a form of insurance to be sold in years when crops fail, (b) as a form of sav- ings to be drawn upon for special occasions, and (c) for the breeding of draft animals so a man could maintain or expand his holdings. Cattle were only of minor consideration as a source of food, and then usually for dairy rather than meat products. The only steady cash income re- ceived from cattle came from the sale of butter. See Noel J. Cossins, "A Study of People and Their Cattle in the Shire Lowlands," mimeographed study prepared for the World Bank on behalf of the Ethiopian Livestock and Meat Board, 1973, p. 79. 14. The Fulani in sub-Saharan Africa have a saying: "When my wife dies I am sad, but when my cow dies I cry." Ibid. 15. In the Shire Lowlands, for instance, mature females constitute 41 percent of the cattle population. About half of these cows give birth annually, and approximately one-third of the calves die. Taking into consideration the herd mortality rate of 12 percent, yearly increases will average only 2 percent. Of course, epidemics and drought threaten even this negligible growth. As Cossions observes: "One can understand the reasons why an owner of 40 animals, with an expected annual increase of only one animal, sells only when he is forced to." Ibid., p. 62. 16. See, for example, Polly Hill, Studies in Rural Capitalism in West Africa (Cambridge: Cambridge University Press, 1970). PRODUCTION SYSTEMS 2 59 of lower return to LLDP. However, it has been proposed that the ranch be stocked by imported cattle. Given the shortage of cattle in Kenya, a major exporter in Africa, it is not clear as to how this alternative compares with increased procurement of domestic cattle. Alternatively, it may mean pro- curing cattle from other exporters. The inadequate knowledge of the factors that affect the cattle market lends considerable uncertainty to the future of livestock development in Lilongwe. As the survey data indicate, if individual farmers are attempting to minimize their own risk by enlarging their herds, overgrazing and reduced carrying capacity on the land may result. Under such conditions initial improvements in the quality of cattle through use of high-priced im- ported cattle may prove self-defeating in the long run unless there is simul- taneous increase in stall-feeding and marketing of indigenous cattle. While promoting livestock production in non-pastoral areas like Lilongwe is difficult, the problems seem even more intractable in the tradi- tional pastoral sector where stock raising is the primary livelihood. The Kenya Livestock Development Project is a case in point. The pastoral areas covered by the project already suffer from serious overgrazing. The Masai are traditionally a herding people, and their herds have been built up to the point where the ecological balance of the area is endangered. The need for destocking is, therefore, urgent. The Masai tradition of common land, pri- vate cattle, however, constrains attempts to cut down the population of the herds. As long as the range is open, it is to each cattle owner's individual ad- vantage to enlarge his herd and get the maximum use of the public lands. This, of course, works to the disadvantage of the community as a whole, for the land is incapable of supporting great numbers of cattle.17 Destocking means reducing the human population in this region as well. Stock raising is capital intensive rather than labor intensive; it is most pro- ductive when undertaken on a large scale. There are already more Masai liv- ing on the range than can be supported by stock raising. Even if employ- ment opportunities were created in dairying, slaughtering, tanning, and other related enterprises, the potential for maintaining a large number of people in the pastoral sector is minimal. Range development must even- tually involve relocalion of a majority of the population and, hence, a dis- ruption of tribal life. Such change can only be brought about gradually as the attitudes of the Masai change with increased education.'8 Many of the past initiatives of the government have been received with considerable suspicion by the Masai. Since production problems in Masai land are related to the common land, private cattle-grazing system, two general solutions present themselves: to divide the rangeland in individual private holdings or to introduce collec- tive grazing. The overgrazing, overstocking cycle would then be halted. KLDP's efforts have concentrated on the former solution, largely through the encouragement of individual and group ranches. 17. Such problems have, of course, been compounded by the recent series of droughts in the Sahel region and the consequent, if temporary, southern expansion of the Sahara. 18. For a discussion of changes being wrought in Masai society, see Hans G. G. Hedlund, "The Impact of Group Ranches on a Pastoral Society," Institute for Development Studies Staff Paper no. 100 (Nairobi: University of Nairobi, June 1971). 60 DESIGN OF RURAL DEVELOPMENT Individual ranches are advantageous in that the owner's limited grazing land dictates a reduction in the size of his herd. The land can, thereby, be given a chance to regain productivity; and the quality of the cattle can be upgraded. The major drawback of these ranches is their size, for in pastoral areas ranches must be very large to take advantage of the erratic rainfall. The individual ranches assisted by KLDP average 600 hectares, which is too small to permit flexible use of the dry grazing land. This size limitation largely accounts for the low 6 percent rate of return on the investment realized on individual ranches.'9 Group ranches, on the other hand, are of a more efficient size, averaging 18,000 hectares.20 However, since only the land and not the cattle is group property in these ranches, the common land, private cattle problem per- sists. Setting individual grazing quotas has been an ineffective means of limiting herd size. Required culling offers a potential solution but remains largely untried to date. The long-range prognosis for these ranches is unclear, for it seems difficult for pastoral people to accept that livestock numbers should be kept fixed, irrespective of year-to-year fluctuation in rainfall and grazing conditions. The possibility does exist that cooperative group ranches could be estab- lished in which most livestock is in a communal herd. This would probably encourage destocking, but it is unlikely that pastoralists would willingly give up their right of free disposal over their cattle. In any event, this partic- ular form of ranching remains largely unexplored by KLDP. To summarize, it appears that, despite the numerous trials and errors in development of livestock involving traditional nomadic herdsmen, far too little is known about the precise response to such steps as development of a livestock market, provision of water dips, veterinary services, and con- trolled grazing through grazing blocks that may be undertaken for develop- ment of the traditional livestock sector. The knowledge is even poorer and uncertainty even greater than is the case with smallholder agriculture. Though investigation of and experimentation with such interventions must be expanded, the review of the livestock development programs suggests that the only long-term solution to the problem of range development in pastoral areas in Kenya may be for many of the pastoralists to be absorbed by the non-nomadic sector of the society. Concluding Remarks The analyses of these various problems related to the development of ag- ricultural and livestock systems indicate that the objectives of raising pro- ductivity in the subsistence sector and of ensuring a reasonable degree of 19. KLDP has also been granting loans to company and commercial ranches. The eight company ranches assisted by the program in Taita district averaged 23,000 hectares and were expected to generate a 20 percent rate of return on investment. The thirty commercial ranches receiving loans averaged 3,250 hectares, excluding the special case of the 300,000-hec- tare Galana Ranch. Their rate of return was expected to be 23 percent. Though they are relatively small in size, most of these latter are heavily involved in feedlot operations, which is a highly profitable enterprise, as in the case of the LLDP stall-feeder program. The internal rate of return from feedlot operations alone on these ranches was estimated at 25 percent. 20. Ecologists still feel that many of these ranches are too small. The rate of return from group ranches is 13 percent. PRODUCTION SYSTEMS 2 61 participation in growth have many dimensions. The programs reviewed have generally been remiss, not so much because they have failed to take into account these various dimensions at the outset, but because over time they have not improved the knowledge of these dimensions and of the in- teractions among them and, therefore, not developed the capability to tackle the equity problem through improved planning and implementation of programs. CHAPTER IV Agricultural Extension and Mass Participation Many of the projects reviewed are designed to ameliorate the problems that are so familiar to those acquainted with the traditional approach to ex- tension in Africa. Extension agents are few and far between, ill-paid, ill- trained, ill-equipped with a technical package, and consequently very poor in quality. That the farmer often knows more, at least about what is wrong with the new innovations, and that extension agents often do not follow their own advice have become parts of a folklore of extension in developing countries, Africa being no exception.' To counter the problems faced in traditional extension services, the pri- mary emphasis in virtually all the rural development programs reviewed is on higher intensity of extension-i.e., on increasing the number of exten- sion agents in a limited geographical area so as to increase the agent/farmer ratio. Many, although not all, new rural development programs also provide additional training to its staff and pay higher salaries. There are also consid- erable differences between the various programs in the organization of the extension system and in the approach to dissemination of innovations. Has the increase in the intensity of extension in these various programs been effective in alleviating the problems of the traditional extension ap- proach? In this chapter the effectiveness of various extension systems is assessed, not only in terms of their effect on increased production, but also in terms of their ability to disseminate a range of profitable innovations to a mass of the rural population, in particular to the relatively poorer sections of the farming population. 1. For an excellent discussion on problems of traditional extension, see J. Katorobo, Agri- cultural Extension in Nyahashenyi, mimeographed (Kampala: Makerere University College, 1966); J. Katorobo, "Agricultural Modernization: Kahoho Parish-Kigezi District," mimeographed (Kampala: Makerere University College, 1968); David K. Leonard, "The Ad- ministration of Kenyan Agricultural Extension Services," Institute for Development Studies (Nairobi: University of Nairobi, 1972); David K. Leonard, "Some Hypotheses Concerning the Impact of Kenya Government Agricultural Extension on Small Farmers," Institute for De- velopment Studies Staff Paper no. 71 (Nairobi: University of Nairobi, 1970), David K. Leonard, W. Opindi, E.A. Lucheme, and J.K. Tumwa, "The Work Performance of Junior Ag- ricultural Extension Staff in Western Province, Basic Tables," Institute for Development Studies Discussion Paper no. 109 (Nairobi: University of Nairobi, 1971); Jon Moris, "Farmer Training as a Strategy of Rural Development," in Education, Employment and Rural Develop- ment: The Proceedings of a Conference held at Kericho, Kenya, in September 1966, ed. J.R. Sheffield (Nairobi: East African Publishing House, 1967), pp. 322-65; M. Okai, "The Adequacy of the Technical Base for the Agricultural Extension Service in Uganda: A Case Study in Lango District," Rural Development Research Paper no. 6 (Kampala: Makerere University College Department of Agriculture, 1965); M. Okai, "Field Administration and Agricultural Development," mimeographed (Kampala: Makerere University College, 1966); T.M. Othieno AGRICULTURAL EXTENSION 63 Evaluation of the effectiveness of extension is by no means easy. It raises many methodological problems, some of which arise from the interactions of the extension service with other services such as credit and marketing, while others arise frorr the very considerable variability between farms in resource endowment and in climatic factors. Such variability may often be far more important in explaining yield differences than is the effectiveness of any particular intervention, including extension. Another important fac- tor is the social organization which, of course, has a profound influence on communications and, therefore, has implications for how extension pro- grams should be organized and administered. For example, under certain circumstances members of the farming population may be far more effec- tive as disseminators of new innovations than are the relatively more urban extension workers. Only to the extent that these various interactions are understood and their variability taken into account can extension services be improved. The methodological problems are not discussed in this chapter. It is importanl. to stress, however, that, given the complexity of the interactions and the overwhelming diversity in the constraints, regular but modest studies of the type used in the analyses presented below seem far more desirable for improving performance of the extension services than do methodologically sophisticated quantitative assessments, which require a substantial amount of data and are demanding of scarce trained man- power. Assessment of the extension systems in the various programs reviewed suggests that merely intensifying the extension service may often be futile, unless conscious simultaneous effort is made to: 1. Impart a technological package that is sufficiently profitable at the farm level to provide an incentive for the farmer to adopt innovations; 2. Train the extension staff to solve the specific but diverse farm-level constraints faced by the cultivator; 3. Develop an incentive system to encourage the extension service to perform its task efficiently, meaning not only rapid growth in produc- tion but also broad participation in the adoption of innovations; 4. Relieve extension of the heavy burden of delivering inputs, writing credit applications, chasing credit defaulters, and so forth; and 5. Enlist the active support and participation of the farmers themselves. and D.G.R. Belshaw, "Technical Innovation in Two Systems of Peasant Agriculture in Bukedi District: Uganda," paper presented at the East African Institute of Social Research Con- ference (Kampala: Makerere University College, 1965), R.G. Saylor, "The Administration of Innovations," Economic Research Bureau Paper (Dar es Salaam: University College, 1969); R.G. Saylor, "An Economic Evaluation of Agricultural Extension in Tanzania," Economic Research Bureau Paper (Dar es Salaam: University of Dar es Salaam, n.d.); R.G. Saylor, "A Social/Benefit Analysis of the Agricultural Extension and Research Services in Selected Cot- ton Growing Areas of Western Tanzania," Economic Research Bureau Service Paper 70.2 (Dar es Salaam: University of Dar es Salaam, 1970); R.G. Saylor, "Studies on the Cost/Benefit Analysis of the Agricultural Research Services in Selected Cotton and Coffee Growing Areas of Tanzania," Economic Research Bureau Paper (Dar es Salaam: University of Dar es Salaam, n.d.); R.G. Saylor, "Variations in Sukumaland Cotton Yields and the Extension Service," Eco- nomic Research Bureau Paper 70.5 (Dar es Salaam: University of Dar es Salaam, 1970): E.R. Watts, "Agricultural Extension in Embu District of Kenya," in East African Journal oJ'Rural Development, vol. 2, no. 1 (1969), pp. 63-77; S.K. Taiwo Williams, "The Confluence of Exten- sion Education, Agricultural Extension and Community Development," in Bulletin of Rural Economics and Sociology, vol. 2, no. 3 (1967), pp. 184-93, M. Crawford Young, "Agricultural Policy in Uganda: Capability and Choice." in The State o,f the Nations: Constraints on Develop- ment in Independent Afirica, ed. M.F. Lofchie (Berkeley: University of California Press, 1971), pp. 141-64. 64 DESIGN OF RURAL DEVELOPMENT Different Organizational Approaches to Extension Services The extension services in the rural development projects reviewed under ARDS may be classified under two different types. Ruthenberg has described these as the "take it or leave it" approach and the contract farm- ing method.2 In the former, peasants are brought innovations and informa- tion which they are free to accept or reject. In the latter, farmers who volun- teer to receive innovations are granted a license; those who fail to follow project guidelines may have their licenses revoked. Take it or leave it exten- sion has been most common in programs involving development of both subsistence and cash crop-e.g., CADU, WADU, and LLDP. Contract farming is usually found in projects specializing in export crop produc- tion-e.g., KTDA and the Urambo and Tumbi tobacco schemes. KTDA even has legal authority-granted under the Tea Cultivation Order-to prosecute negligent growers. With take it or leave it extension, it is evident that innovations must be highly profitable if they are to gain wide acceptance (see Section 4 of this chapter). Profitability is also a prerequisite for successful contract farm- ing-one of the "carrots" that project agencies use to enroll farmer partici- pation. However, the contract farming extension services have the addi- tional benefit of a "stick" to ensure that farmers follow all the rules of prop- er husbandry. At times, recommended practices may be unpopular, not because of their ultimate unprofitability, but because to the cultivator they seem to involve hardship or sacrifice as, for example, required culling of cattle or the necessity to destroy a crop which has been infected with dis- ease. Recommended practices may also not have an obvious immediate payoff as, for instance, measures to prevent soil erosion. In these situations voluntary cooperation of cultivators may be difficult to obtain without a very substantial effort to convince them of the return to the practice. Such effective communication may be lacking even in contract farming, thus fre- quently necessitating use of the "stick." Why have projects adopted such diverse approaches to extension? The export crops have traditionally yielded the high profit margins that are nec- essary to cover the considerable expense of extension in closely supervised contract farming. Further, export crops frequently involve sophisticated production techniques and high production standards; hence, often they cannot be left to independent field-level control.3 By contrast, diversified farming, and especially food crop production, has frequently tended to be less profitable and has, therefore, not justified heavy investment in exten- sion personnel. Also, food crops are generally assumed to be less demand- ing of quality control requirements and, hence, of technical expertise. 2. Hans Ruthenberg, "Types of Organisation in Agricultural Production Development," in Zeitschrift fdr Auslandische Landwirtschaft, Jahragang 12, Heft 3/4 (July-December 1973), pp. 234-44. 3. For example, fine plucking-that is, the picking of only the tea bud and two leaves-is one of the most important factors in quality tea production. Less precise plucking, which in- cludes older leaves, tends to have an adverse affect on the fermentation process, resulting in poor flavor and color in the finished product. Plucking at too infrequent intervals tends to make the bud hard and the leaves leggy with too much stalk. Careless plucking can damage the plant. AGRICULTURAL EXTENSION 65 It must be stressed, however, that the differences between the sophistica- tion of technology for export crops and food crops are far less pronounced than generally believed. Nor for that matter is there reason to believe that food production is always less profitable than is the production of tea, to- bacco, or similar products. Rather, in many respects the adoption of the take it or leave it approach with regard to innovations for subsistence crops is merely another reflection of the common tendency to neglect the impor- tance of food crop production for rural development (see Chapters II and III). While a coercive extension service may not be warranted, emphasis on suitable technology and effective extension of farm-level practices, the two elements frequently observed in contract farming of export crops, would seem critical in all cases for augmenting production. Similar to the advice/coerce dilemma in agricultural extension is the question of whether extension workers should be trained as generalists or specialists. Specialization has obvious benefits, particularly given the gener- ally low level of knowledge displayed by many extension workers. It may well be easier to imbue field agents with an effective package of specific rec- ommendations for a particular crop than to train them in a wide variety of farming techniques. By the same token, a specialist may be more effective in communicating his narrow but concrete expertise to the farming popula- tion. A survey in Kenya indicates that extension specialists tend to have more command over their technical areas than do agents with more general training.4 However, as pointedi out earlier, even a smallholder producing an export crop needs assistance with respect to food and livestock activities. Conse- quently, the farmer tends to demand wide-ranging advice from the exten- sion staff that he is able to see.5 Therefore, a relatively general extension service is frequently essential if it is to be relevant to the needs of the farm- er. Further, specialization may lead to bureaucratic skirmishes in the field with various "experts" pushing contradictory advice at the expense of an overall development effort. McLoughlin observed that frequently "two men give the farmer, not slightly, but radically different advice, plant your cotton early in the year versus plant your cotton in October."6 Particularly given the frequent need to develop the overall farming system outlined in Chapter II, a generalized extension designed to meet the needs of the farm- er seems warranted from the outset, although effective delivery of such a generalized service is by no means easy to introduce for the reasons dis- cussed later in this chapter. Tobacco production requires similar critical and painstaking harvesting techniques. The leaves of a tobacco plant differ chemically according to their position on the plant. Since the lower leaves ripen earlier than the higher ones, several harvests may be required. Complicat- ing matters further, the signs of ripeness are quite different in leaves of different stalk posi- tion. 4. The differential between specialists and generalists is largely accounted for by the greater intensity of rebriefing given to the former. See Leonard et al., "The Work Performance of Junior Agricultural Extension Staff," Table 6.E, 15.E, and 20.E; and Leonard, "Organizational Structures for Productivity in Kenyan Agricultural Extension," in Rural Administration in Kenya, ed. David K. Leonard (Nairobi: East African Literature Bureau, 1973), p. 141. 5. Ibid. 6. Peter F.M. McLoughlin, "The Farmer, the Politician and the Bureaucrat: Local Govern- ment and Agricultural Development in Independent Africa," notes and papers in Development no. 4 (Fredericton, New Brunswick: Peter McLoughlin Associates, n.d.). Other problems may also arise from multiplicity of extension services. In his survey in the Vihiga district, Kenya, 66 DESIGN OF RURAL DEVELOPMENT Another important question concerns whether or not extension services should be organized within the existing government bureaucracy or alter- natively as an autonomous or semi-autonomous body. Departments of agri- culture are frequently ill-equipped to run extension services because of their lack of entrepreneurial ability, their lethargic attitude towards decision making, and their inability to respond to incoming information. Establish- ment of the extension service as a separate body, run by technical experts and farmer representatives, is, therefore, frequently considered neces- sary.' However, as will be illustrated in Chapters VIII and IX, there is no organizational panacea.8 The effective organization of agricultural exten- sion has to be viewed as part of the broader and unquestionably urgent problem of improving overall administrative performance. The final choice of specific policy in this matter must, however, depend on the time frame involved and the nature and sophistication of the government bureaucracy. There are several other considerations that have entered organization and delivery of extension services and affected effectiveness of the pro- grams reviewed. Of these the greatest emphasis has been on increasing the intensity of extension services. Intensity of Extension Services in Rural Development Programs Investment in extension services has usually been assessed in terms of extension worker/farmer ratios. In LLDP in Malawi during the first two and a half years after development of each unit, there was to be one exten- sion worker to every 200 farm families. Over the next two and a half years, this ratio was to be reduced to one worker to every 400 families.9 By con- trast in the non-program area, the regional agricultural officer's jurisdiction has had one extension agent per 1,200 to 1,300 holdings. However, even in the limited program area, the planned ratio of exten- sion workers to farm families had not been reached during Phase I of the program. According to the program review, this was primarily due to the Leonard reported that three agents from the Ministry of Agriculture together spent four days visiting tea growers. Unfortunately, tea growing was the responsibility of the Kenya Tea De- velopment Authority. To compound matters, in recent years KTDA has also taken on the responsibility for the dissemination of information on hybrid maize. Such overlap in exten- sion service suggests "that the pressures in favor of a generalized extension service in peasant areas are quire strong." See D.K. Leonard, "Some Hypotheses Concerning the Organization of Communication in Agricultural Extension," Institute for Development Studies Staff Paper no. 72 (Nairobi: University of Nairobi. 1970), p. 21. 7. Hans Ruthenberg, "Adaption of Extension Projects to Changing Circumstances," paper delivered at the International Seminar on Extension and Other Services Supporting the Small Farmer in Asia by the German Foundation for Developing Countries, Berlin, October 31-November 21, 1972. 8. See John C. de Wilde, Experiences with Agricultural Development in Tropical Africa, vol. I (Baltimore: Johns Hopkins Press, 1967), p. 180. 9. It has been proposed that, subsequent to the program's development period, the regional extension staff be augmented by the output of the training section included in the program until the ratio of extension agents to farm families in the program area is ultimately stabilized at about one extension agent to 600 to 800 farm families. See World Bank, Lilongwe Land De- velopment Project, Malawi, appraisal report no. TO-610a, restricted circulation (Washington, D.C., January 1968). AGRICULTURAL EXTENSION 67 shortage of trained extension staff, although the initial concentration of ad- ministrative effort on infrastructural development may also have affected the attention that could be devoted to the development of extension ser- vices. It was anticipated that the desired ratios would be attained during Phase 11. In this period extension coverage was also to be extended to an ad- ditional 24,000 farming families-i.e., a total of 52,000 program families. In the Urambo tobacco scheme, the extension worker/farmer ratio declined from 1:323 to 1:802 between 1965 and 1970, while in the Tumbi scheme the ratio dropped from 1:116 to 1:575. Even then, the two schemes had far greater intensity of extension than average. The average extension worker/farmer ratio lor Tanzania as a whole has been 1:1,500; while in the Sukumaland cotton-growing areas, it has been 1:1,020. In Kenya KTDA has had an extension agent/farmer ratio of 1:120, compared to 1:500 in Kenya as a whole. These ratios, of course, provide only rough estimates of the magnitude of intensity. They do not indicate whether a greater number of agents in pro- portion to farmer population means that more farmers are being reached. They also fail to indicate what reaching a farmer signifies. Such difficulties in gauging intensity are illustrated by comparison of in- novations related to agro-industrial enterprises with those related to food crops. In the former, as, for instance, in KTDA, adoption of innovations is frequently tied to use of other services provided by the program. Participa- tion is, therefore, fairly easy to measure. In the case of food and some ex- port crops, on the other hand, farmers may adopt innovations, such as early planting or close weeding, without necessarily accepting the services pro- vided by the program. Estimation of the number of adopters or of the in- crease in their productivity is, therefore, difficult in the absence of detailed information on the spread of various types of innovations. These data have been lacking in many of the programs reviewed. Lacking such information, Tecle estimated effective field staff/farmer ratio for CADU and WADU in Ethiopia by assuming that only those farm- ers who borrowed credit from the programs were reached by the extension services. Calculated in this manner, in the 1972/73 period CADU had an ex- tension agent/farmer ratio of 1:470 compared to 1:335 in WADU. These are particularly high ratios for Ethiopia, given that prior to CADU or WADU there had been hardly any development of agricultural extension in Ethiopia, compared to Kenya or Tanzania.'0 In considering intensity of extension, another factor which must be taken into account is the investment in training of extension agents. There are significant variations between countries. In Kenya, for example, most of the junior agricultural assistants have undergone only a one-week course at a farmer training center." Junior staff in the LLDP areas in Malawi, on 10. In 1968 Kenya, with a rural population of 10 million, employed 5,277 extension workers. For Tanzania the figures were 11.5 million and 2,455, respectively. Ethiopia, with a rural population of 22 million,by contrast had only 124 extension agents. See Robert Evenson and Yoav Kislev, Agricultural Research Productivity (New Haven, Conn.: Yale University Press, 1975), Appendix Table 2A, Chapter 2. 11. David K. Leonard, "Organizational Structures for Productivity in Kenyan Agricultural Extension," p. 133. 68 DESIGN OF RURAL DEVELOPMENT the other hand, are given a period of pre-training and practical experience as well as an intensive course in technical agriculture at Colby College, located within the program area at Chitedze Research Station."2 The services available in terms of means of transportation to extension agents may also vary considerably between programs. These have an im- portant bearing on intensity, since significant constraints are often imposed on extension services by transportation problems. For example, in ZAPI in Cameroon, although agents do have bicycles, the brush country in which they work can often be covered only on foot. Consequently, a great deal of time is expended in traveling. A similar situation exists in the Vihiga dis- trict of Kenya, where merely attending a staff meeting may involve a four- to six-hour bicycle ride.13 In contrast, the LLDP field staff interviewed in Malawi proudly empha- sized that they had easier access to vehicles and better roads than did their counterparts in the regional agricultural office. Of course, improving transportation facilities increases costs. However, it also increases the mobility and the initiative of the extension staff. An agent is able to reach more farmers, or reach the same farmers more often, and can become more motivated to get his message across. Costs Implicit in Intensity of Extension Services Particularly because costs of extension vary significantly, depending on training, facilities, salaries, and numbers of farmers reached, it is essential to examine intensity of extension in terms of its implications for recurrent budgetary expenditures relative to the return likely to be derived from the extension service and compared to other alternative expenditures. This is necessary even though assessment of effectiveness of extension as com- pared to that of other investments may be based only on qualitative judg- ments and substantially influenced by social preferences.'4 In LLDP in Malawi the cost of extension services is relatively low. Esti- mates called for $0.90 to be spent per family per year over a period of thir- teen years of program development. Over $1.2 million, or about 8.5 percent of the LLDP budget, will be spent on extension by 1975. From other indica- tions of progress on adoption of innovations in LLDP (discussed below), to date the rate of dissemination of innovations implicit in these per family costs appears to have been overestimated. The above figures also exclude the cost of training. However, if LLDP's rather low cost estimates are ac- 12. For a more detailed discussion of the training of extension agents, see Chapter X. 13. David K. Leonard, "Some Hypotheses Concerning the Organization of Communication in Agricultural Extension," p. 20. 14. In Kenya, for example, the expenditure on extension services in 1968 was $4.5 million (see Evanson and Kislev, Agricultural Research Productivity). In comparison, the central gov- ernment's current expenditure on education was only $2.2 million. See USAID, Statistics and Reports Division, "Central Government Finances: Kenya," in AID Economic Data Book for Africa, revision no. 286 (Washington, D.C.: U.S. Government Printing Office, October 1972). As will be illustrated in Chapters Vil and IX, these proportions are almost the reverse in Tanzania. AGRICULTURAL EXTENSION 69 cepted, for the intensity of its extension program to be replicated on a na- tionwide scale, they imply an annual expenditure of nearly $0.9 million or about 11 percent of the government's estimated expenditure on agricultural development for the 1972/73 period.15 Actual expenditures on extension and training, excluding commercial export crops such as tea and cotton, amounted to approximately 6 percent of the total expenditure on agricul- tural development. Smallholder tea development in Kenya, on the other hand, involved a much higher investment. KTDA's extension effort has cost a total of more than $18 per farmer per year, including the cost of the executive and field staff plus the cost of the input package. If this investment were to be made on a nationwide basis, it would cost approximately $35 million, or more than the current annual expenditures on agriculture of the central govern- ment in the late 1960s.16 It is apparent that, without a very major commit- ment of resources by national governments to rural development and, equally important, to suitable training of the manpower necessary for ex- pansion of these services, intensity of most of these programs cannot be replicated on a large-enough scale to reach a mass of the rural population in the foreseeable future. Particularly, because of the resources implicit in these programs, three questions concerning their efficacy need consideration: To what extent are the extension services equipped with profitable technology? Have they de- veloped ways of tackling the numerous specific problems faced by in- dividual small farmers in adopting innovations? Have they acquired effec- tive means of disseminating new innovations and ideas to the mass of the rural people rather than to a privileged few? Profitability of Technology and Effectiveness of Extension The lower yields of tobacco in Urambo in comparison with those of Tumbi in Tanzania are often attributed to a shortage of extension staff at Urambo. Reports by officials in Urambo argue that the staffing constraint made it difficult to supervise new farmers and to continue to encourage and help the old farmers in maintaining the improved methods of production. However, Agarwal and Linsenmeyer observe that inadequate attention to the technological problems faced by farmers may be the major cause of the poorer yield performance of the Urambo scheme. The depletion of the field staff had been occurring at both places, but productivity at Tumbi had been increasing while that at Urambo had been decreasing. The reviews of ZAPI and SODENKAM in Cameroon indicate that, although closer contact with the population is usually considered to be es- sential for a successful extension program, it is by no means sufficient if the 15. Assuming one million farm families in Malawi. According to estimates made by the World Bank, expenditures by the Malawian government for agricultural development were $8.3 million in the 1972/73 period. 16. USAID, Statistics and Reports Division, "Central Government Finances: Kenya." 70 DESIGN OF RURAL DEVELOPMENT extension system is not equipped to tackle specific farm-level constraints faced by the cultivators. These programs use close-up teams,17 founded on the implicit assumptions that: (a) the main cause of resistance to new tech- niques is psychological; and (b) the peasant must be convinced of the value of the new techniques, which can occur only through personal contact with the vulgarisateur (extension worker). Thus, by emphasizing subjective reasons almost exclusively, planners have neglected to analyze the objec- tive factors which frequently militate against the adoption of new tech- niques. This point is illustrated by the research carried by Societe d'Etudes du Developpement Economique et Social (SEDES) in ZAPI, which dis- covered that individual farmers faced a variety of obstacles in the adoption of copper spray to combat brown rot in cacao buds, even after they were convinced of the importance of using the spray. One planter in four did not have treatment equipment, two planters in four who had the equipment met operating problems, one planter in three had difficulties in finding the water necessary, one planter in three had manpower difficulties, and one planter in two received the treatment too late for the time he would have liked to begin the treatments. Thus, supposedly irrational farmer resistence to new techniques may be far less a constraint than is the unavailability of the necessary inputs. Other objective factors may induce cultivators to resist recommended practices. For example, despite considerable intensity of extension, the BDPA groundnut operation in Mali has been less successful than antici- pated in promoting the use of a fungicide-insecticide for treating millet and groundnut seeds. The area actually treated for the two crops has consis- tently fallen short of projections by as much as 50 or 60 percent. The slow rate of adoption is explained by the low net return to the insecticide. Early sowing of groundnuts, another BDPA-sponsored innovation, has also met with resistance. In the period 1970/71 (three years after the pro- gram got under way), only 23 percent of the project area had been sown before June 15. The percentage increased to 50 percent the next year. Sig- nificantly, about 37 percent of the farmers had not planted their groundnuts as late as July 15 in the period 1970/71. By the period 1971/72 this figure had dropped to only 18 percent. The relatively slow progress in promoting early planting of groundnuts in BDPA may have been a result of a labor con- straint similar to that encountered by cotton farmers in Sukumaland.'8 LLDP's extension program has also had mixed results. The program has promoted the use of fertilizer and the increase of plant population per hec- tare, both of which have a substantial positive effect on yields of maize and groundnuts. Kinsey's analysis of LLDP Evaluation Unit's survey data for the period 1970/71 indicates that extension visits are significantly, positively associated with higher maize yields. Extension has been less 17. The head of the close-up team is a moniteur (team head) or vulgarisateur (extension worker), who is generally 20 to 30 years old with an elementary education and trained in the project itself over a period of several months at the outset and in short review sessions thereafter. Team numbers may vary, often according to financing possibilities; but it seems to be accepted that the denser the staffing, the better the results. Mediocre results are usually blamed on not enough staff. Staff density varies from one vulgarisateur to every 200 planters in the case of ZAPI to one for 50 planters in SODENKAM. 18. See Chapter 11 of this study. AGRICULTURAL EXTENSION 71 effective in increasing groundnut production, however, reflecting the futility of the extension effort when the technology is not profitable. Farm- ers receiving no extension visits during the year had mean yields of 4.93 quintals per hectare, while those receiving five visits averaged only 4.09 quintals. Farmers visited between eleven and twenty times experienced even lower yields, 3.50 quintals per hectare. Only when more than twenty visits were made was there a positive correlation between intensity of ex- * tension and groundnut yield; these farmers averaged 5.94 quintals per hec- tare. Though these findings appear to be somewhat anomalous, a partial ex- planation for the curious relationship between groundnut yields and exten- sion visits may lie in the various constraints related to groundnut produc- tion discussed in Chapter 111.19 Incentives and Performance of Extension Services The preceding discussion implies that the absolute number of extension workers in a prograrm area is less important than their individual perfor- mance. Programs cannot be very successful without energetic, motivated staff. Staff incentives are, thus, a crucial factor in the success of extension programs. Yet, while there is little disagreement on this point, there seems to be considerable difference of opinion in the project reviews as to what specific type of incentives are most effective. The two incentives conventionally considered cruicial to staff perfor- mance are salaries and promotion. Leonard's study of SRDP's extension workers in the Vihiga district in Kenya points out the effect of the absence of these incentives. Htis interviews indicated that dissatisfaction among ex- tension agents was widespread. Virtually all of the junior staff complained about their pay. About half were unhappy about the poor prospects for pro- motion-only a quarter of them had received any promotions in the pre- vious four years; more than half had been in the same position for ten years or more. Compounding the problem, advancement did not even appear to be geared to performance.20 Chambers and Belshaw point out that unpleas- ant working conditions and erratic administrative supervision also con- tribute to the sense of frustration among extension workers.21 Increasing pay, rationalizing promotional opportunities, and restructur- ing the supervisory system would, thus, seem to be an obvious means of improving staff morale and performance in national administrative struc- tures. However, when such simple steps as increasing pay scales and open- 19. On the whole, the follow-up study of LLDP, involving much more detailed analysis, in- dicates a strong correlation between extension visits and adoption of various practices and in- puts. 20. Leonard, "Some Hypotheses Concerning the Organization of Communication in Agri- cultural Extension," pp. 23-24. 21. Robert Chambers and Deryke Belshaw, Managing Rural Development: Lessons and Methods from Eastern Africa, Institute for Development Studies Discussion Paper no. 15 (Brighton, U.K.: University of Sussex, June 1973). Chambers and Belshaw feel that improved supervision of the staff, work planning, reporting, and staff evaluation are critical for increas- ing the effectiveness of extension. Also see references to the PIM management system in Chapter X of this study. 72 DESIGN OF RURAL DEVELOPMENT ing opportunities for advancement are taken on a piecemeal basis in iso- lated programs, they frequently lure staff away from other worthwhile pro- grams.22 Favorable incentives in one program also cause resentment and envy in staff working elsewhere under less desirable conditions. These fac- tors are discussed in detail in Chapter VIlI. Improvement of the extension system, thus, has to be examined in the context of improvement of the overall administrative system rather than in terms of a few small programs. In improving staff salaries and fringe benefits, another real danger ex- ists-namely, the development of a self-conscious elitism among the staff, causing friction with the local population. Belloncle and Gentil have repeatedly observed this phenomenon in their review of the French techni- cal assistance projects in Cameroon. According to them, to improve staff- farmer relations, it may, in fact, be desirable to keep salaries low, or at least at a point where living standards of the staff do not differ substantially from those of the average farmer. The extent to which low salaries will im- prove rapport of the extension worker with the rural people may, however, also depend greatly on whether the workers came from rural or urban back- grounds, an issue which will be discussed at the end of this chapter. A related point is made by Leonard, who observed that formal education of extension workers does not necessarily increase their effectiveness. In Kenya agents with a secondary education had a lower degree of agricultural knowledge than those with an upper primary education. More important, the more educated agents were less persuasive in promoting innova- tionS.23 There appear to be two explanations for this phenomenon: (a) the secondary school finishers tend to come from urban backgrounds and are less committed to extension work, and (b) through formal education such individuals may have developed a value system that regards farming as an inferior occupation. However, Leonard also found that the performance of secondary school finishers can be improved through practical training courses.14 It is obviously important that field workers be able to work closely with the rural people. Indeed, the development of a personal involvement with the people of a particular community is in itself a strong incentive. As noted earlier, in Tanzania the extension effort has been more effective in the Tumbi region than in Urambo. Linsenmeyer and Agarwal argue that this is largely attributable to the fact that in Urambo the staff live at headquarters and make periodic trips to supervise and advise the farmers; whereas in Tumbi, since 1960, extension staff live in the areas where they work. This puts them in closer contact with the farmers and reduces their travel time. This closer contact seems to have a favorable effect, as there has been a fairly steady increase in productivity per hectare in Tumbi since 1964. 22. The Department of Agriculture in Kenva has lost much of its most competent staff to KTDA, as has the Department of Agriculture in Tanzania lost its best staff to BAT, affecting the development of the regular administrative structure in the two countries. 23. Leonard, "Organizational Structures for Productivity in Kenyan Agricultural Exten- sion," pp. 135-38. Leonard notes that R.K. Harrison has observed a similar phenomenon in western Nigeria. 24. It should not be inferred, however, that extension workers require no formal education. The degree of fluency in Swahili was an important factor in determining the agents' effective- ness. Ibid. See also Chapter X for a discussion of the training of extension workers. AGRICULTURAL EXTENSION 73 Oddly enough, Leonard's analysis of extension in the Vihiga district ar- rives at an opposite conclusion. His investigations indicated that nonresi- dent agents averaged more farm visits and saw more individuals than did those who worked near their home.25 This may be explained by the fact that an agent working in his home area finds it easier to shirk his duties. Also, it seems likely that when an agent's constituency consists of his friends and relatives, they are either less persistent in asking for his assistance or know him too well to want his advice. Family and tribal duties may play an addi- tional role in diverting a resident extension agent from his work. Local Participation in Extension Activities The contradictory observations of Linsenmeyer and Agarwal on the one hand and of Leonard on the other lead to perhaps the most crucial factor in the success of an extensiotn program-the extent of popular interest. Whether an agent lives in a community or not is of little importance if the residents of the community genuinely want his services. As pointed out earlier, the profitability of the innovations is of paramount importance in this regard. However, many programs reviewed have also taken other steps to mobilize popular support. In Kenya KTDA has been remarkably successful in spreading technical innovations. By 1972, 61,000 small farmers were growing tea. Several fac- tors explain this success: As pointed out in Chapter 11, (a) KTDA has had an exceptionally profitable technology at its disposal; (b) KTDA has a rather authoritarian extension system which enforces correct management practices strictly; and (c) the project has identified the most important farm-level constraint posed in tea cultivation-namely, the labor require- ments of tea production-and has limited the area under tea production to 0.4 hectares per farim. All of these factors have obviously worked in KTDA's favor and contributed to the rapid increase in the number of bene- ficiaries of the program. In addition, however, KTDA has also enlisted con- siderable local participation through tea growers' committees, which assist in the administration of the program. The growers' committees exist at the divisional, district, and provincial levels and consist primarily of representatives chosen by the growers, although there are some ex officio members such as agricultural officers and tea company officials. Serving a dual function, the committees are both a forum for making recommendations from below and a conduit for dis- seminating policy from above. The tea committees usually attract the more successful, respected, and articulate members of the local communities. Therefore, these representatives exert considerable influence over the other growers. Tanzania's ujcmaa movement has also used a local committee system to enlist popular support for its extension effort. In most of the ujimaa vil- 25. Leonard, "Some Hypotheses Concerning the Organization of Communication in Agri- cultural Extension," pp. 22-23. 74 DESIGN OF RURAL DEVELOPMENT lages, an agricultural committee is responsible for ensuring that villagers follow the advice of the bwana shamba (agricultural field agent). However, because of the low level of training of these field agents and the deficiency in research support, the innovations have often been of dubious quality and relevance. These problems are compounded by the national extension strategy, which is directed towards sociopolitical as well as productive goals.26 Ideological considerations impinge on extension effectiveness by leaning towards exclusive concentration on cooperative agricultural produc- tion and by causing extension agents to spend more time raising political consciousness and less time spreading technical advice. How have the wajamaa (ujcmaa villagers) responded to these efforts? The review carried. out under ARDS noted that many individuals voiced agreement with cooperative production and generally supported the goals of the national leadership. The villagers' behavior, on the other hand, told a somewhat different story. Many of them continued to give priority to their private plots, making only minimal contributions to community projects. Where farmers were permitted to grow their own cash crops, they devoted as little as eight hours a week to communal farming.27 Even where mecha- nization has been used as an incentive for villagization, the popular attitude has been to neglect the communal plot. As the villagers in one village put it: "the tractors do the work for us; the tractor gives us freedom to work on our own shambas."28 Such attitudes, of course, lead to disparities in the returns from private versus communal farming, thereby reinforcing preferences for the former. In response the central government has granted village commit- tees increasing latitude to enforce the cooperation of village resi- dents.29 While it is unclear how effective such sanctions will be, forced cooperation obviously conflicts with the democratic principles underlying the ujamaa movement and casts doubt on the government's ability to enlist widespread voluntary support for its program. (For a further discussion of some of the problems faced by uj6maa, see Chapter IX). While KTDA and the uj6maa movement have attempted to involve local people through direct participation in project administration, CADU in Ethiopia attempts to encourage participation through use of model farmers. Under this approach, certain selected farmers work in close conjunction with extension agents to adopt a wide range of new farming techniques. The increased yields produced by these model farmers are expected to pop- ularize the program and encourage other farmers to adopt innovations. ZAPI in Cameroon has followed a similar course. 26. For a generally sympathetic discussion of this strategy, see William L. Luttrell, "Vil- lagization, Cooperative Production, and Rural Cadres: Strategies and Tactics in Tanzanian Socialist Rural Development," Economic Research Bureau Paper 71.11 (Dar es Salaam: Uni- versity of Dar es Salaam, 1971). 27. See P.B. Ngeze, "Some Aspects of Agricultural Development in Ujamaa Villages," paper presented to the East African Agricultural Economics Society Conference, Dar es Salaam, 1973, pp. 10-11. 28. Comments recorded by Vail at an u.jamaa village in Arusha. See David J. Vail, "Tech- nology for Socialist Development in Rural Tanzania," mimeographed, n.d., background paper for the Seventeenth Annual Meeting of the African Studies Association, Chicago, November 1974. 29. Clyde R. Ingle, "Compulsion and Rural Development in Tanzania," in Canadian7 Jour- nal of Affican Studies, vol. 4, no. 1, Winter 1970, pp. 77-100. The anti-communal attitudes of many ufamaa villagers is further revealed by the widespread theft from communal plots. AGRICULTURAL EXTENSION 75 The use of opinion leaders and progressive farmers as the vehicle for gen- erating mass support seems to result in undue concentration of extension effort on the relatively better-off members of the community. It should be emphasized that the concern is not so much that the innovations spread to the better-off farmers first but that, through such an approach, the neediest members of the community may be frequently overlooked in many pro- grams. A study made by CADU's Evaluation Unit suggests that there is a marked difference between adopters and non-adopters.30 The non-adopters have smaller cultivated areas than do the adopters. They have fewer num- bers of cows and oxen, they are less often members of groups and associa- tions; they have a lesser ability to read and write; and most of them are ten- ants (only 10 percent of adopters were tenants). Even among the adopters, the model farmers were younger, more literate, and more likely to be the members of groups, according to the study. A similar situation exists in ZAPI, where a study observed that farmer leaders differed greatly from the average planter. The leaders were young-almost half of them between 30 and 40 years of age. All of them were literate, as compared with a literacy rate of 57 percent for the male population at large. Forty-eight percent of the leaders qualified as large- scale planters, whereas only 17 percent of other farmers fell under the same category. Leonard's survey in Kenya's Western province shows similar results. The average extension agent makes 57 percent of his visits to progressive farmers (who are 10 percent of the total) and 6 percent of his visits to non- innovative ones (47 percent of the total).3' Not surprisingly, these progres- sive farmers tend to be relatively well-off, often owning cattle and fre- quently growing cash crops in what is largely a subsistence economy.32 The follow-up of LLDP also indicates a similar tendency in that program area. A related problem has been that the selected farmers receiving intensive assistance may not be those best suited to popularize innovations. As the 30. Johan Toborn, The Innovation-Diff/ision Process in the Gonde Area, CADU Special Study no. 3 (Asella, Eth.: Chilalo Agricultural Development Unit, March 1971). The findings of this study have been questioned by some administrators of CADU. For instance, Bengt Negby, the former director of CAD)U, in his discussions with the author, has pointed out that, in CADU and MPP, model farmers have been elected by their neighbors and receive no special privileges or rewards. Onlv a small part of their land is used for field demonstrations. Though the farmers do not receive privileges from the program, their initial privileged position is also pointed out by another study carried out in CADU of the cooperatives, which points out the dominance of larger farmers in the programs' cooperative effort (see Chapter VI of this study for further details). Most other studies carried out by CADU's Evaluation Unit and reported in this book also bear out the low participation of tenants. As discussed in detail later in this chapter, these outcomes of the project were, of course, greatly influenced by the sociopolitical structure of the Chilalo society over which the alien management could exercise little control. 31. David K. Leonard, "Why Do Kenya's Agricultural Extension Services Favor the Rich Farmers'" Paper read at the Sixteenth Annual Meeting of the African Studies Association, Syracuse, N.Y.. October-November 1973, p. 5. 32. Ibid., p.3 and passim. See Chapter X of this study for discussion of the Tetu Extension Program as part of SR DP which attempts to avoid this problem through a different training program for the extension staff. Leonard indicates a number of reasons why his investigations show that Kenya's extension service favors the more wealthy farmers. Extension workers tend to come from the uppper social strata and, therefore, naturally gravitate towards farmers from a similar background. Also, it is the more well-off farmers who are the most likely to complain to a senior officer if extension is not provided to them. In addition, richer farmers tend to be more receptive to innovation; hence, they are easier for the staff to work with. 76 DESIGN OF RURAL DEVELOPMENT experience in Cameroon suggests, it is quite likely that many farmer leaders in ZAPI were selected not so much for their status or authority within the traditional society but for their ability to speak French or their habit of keeping in contact with the ZAPI administration. Administrators often have preconceptions of what a progressive farmer should be and tend to select leaders based upon these preconceptions. Such isolated model farmers or farmer leaders may frequently arouse the suspicion and apprehension of their neighbors. The innovative farmer may seem to be trying to rise above his social group and may be seen as a de- viant, endangering group solidarity, or even be viewed as a traitor who has agreed to act as an accomplice of the outside world. In the ZAPIs not long ago, isolated innovators were accused of witchcraft. Belloncle and Gentil observe that for this reason the first innovators are rarely leading citizens but usually people who are already considered to be more or less marginal. The use of a model farmer approach, however, is only one of the many reasons that may explain the concentration of extension effort on a few privileged farmers. In BDPA's Op6ration Arachide, although the top per- sonnel indicate that all farmers benefit in an area in which the extension service operates, Anderson observes that in practice extension agents seem to concentrate on the farmers who are making the most progress in adopt- ing project innovations. And when extension agents begin to introduce animal traction in their secteurs de base (the lowest operational level, con- sisting of 200-250 farmers), their efforts inevitably become more restricted (to the farmers adopting animal traction), since it is critical that these in- dividuals succeed. That extension workers spend much of their time with a few farmers, and that these few farmers tend to be an elite group would not be too great a problem if close relations existed between the progressive and conservative elements in the rural society. Unfortunately, personal and class rivalries may often thwart cooperation and the sharing of technical innovations. In KTDA, for instance, tea committees have provided preferential treatment to certain growers, particularly prior to 1969 when planting material often fell short of demand. The committees ensured that the more established planters received the supplies they needed, while newer, more tenuous farmers were sometimes cut short. Such interclass conflict is also a factor in Tanzania, where there has been strong resistence to ujamaa in the more productive rural areas (see Chapter IX). Neglect of Women in Agricultural Extension Associated with the failure of many programs to reach the majority of smallholders is the tendency for agricultural extension services to focus their attention on male farmers. As noted in Chapter 11, women often con- tribute a major proportion of the family farm labor, usually to the produc- tion of food crops and to specific tasks such as weeding. In many traditional societies it is the custom for the women to provide the major support for themselves and their children, either by supplying the family with home- grown food or by obtaining cash through the sale of their own pro- AGRICULTURAL EXTENSION 77 duce.33 Perhaps an even more important indication of the women's role in agriculture is the fact that a large percentage of rural households is headed by women.34 Agricultural extension programs have frequently overlooked the impor- tance of women, both as major contributors to the farm labor supply and as significant family breadwinners. This oversight can be attributed most readily to a tendency arnong project planners and authorities to see African women in Western terms-i.e., essentially as domestic workers whose pri- mary responsibility should be in the home and not in the fields.35 Thus, the goal of extension services has frequently been not the increase in farm-level productivity of women but rather finding ways to reduce their participation in agriculture through promotion of more homebound activities. Often, such efforts have the opposite results. As pointed out in Chapter 11. some farming innovations may result in increasing the burden of labor on women. Further, if emphasis is on developing male-oriented cash crops-the income from which the men often monopolize-this can divert female labor from food production with subsequent effects on the welfare of the family. Alternatively, other innovations, especially mechanization of agriculture, may eliminate female agricultural jobs and thereby also reduce female income. In all instances neglect of women's role in agriculture may act as a drag on economic growth and contribute to imbalances in the dis- tribution of the benefits of the growth that does occur. Too often women's extension programs have been exclusively oriented toward domestic science and home economics. Of course, improving nutri- tional value in food preparation, fostering hygienic practices, introducing means to conserve labor in the home, and so forth are of significant social and economic value and should not be abandoned in most cases (see Chapter VII for a discussion of the implications of social services for rural welfare and Chapter X for an analysis of some women's training programs). However, there is a need to supplement beneficial home-related programs with efforts to preserve or to improve upon women's more strictly econom- ic functions. The most effective course of action in this regard would generally appear to be fostering improvements in the production of food crops. At the re- quest of local women's organizations, ZAPIs in south-central Cameroon have recently initiated a program to improve the cultivation of groundnuts. They have developed a package of recommendations (treatment of seeds, dense sowing, planting dates) that can result in groundnut yields that are six 33. Dobart and Shields report that, in many traditional African societies, there is no com- munity of property between married couples. Men and women often live and eat separately and keep their individual earnings. See Margarita Dobart and Nwanganga Shields, "African Women: Security in Tradition, Challenge in Change," in Africa Report, vol. 17, no. 7 (July-August 1972), p. 18. 34. According to the ECA, 45 percent and 35 percent of the rural households in Kenya and Malawi, respectively, are headed by women. See United Nations Economic Commission for Africa, "Women and National Development in African Countries," position paper prepared by the Human Resources Development Division, Women's Programme Unit, mimeographed, February 1973. 35. For a revealing example of this perception of women's role in agriculture, see the dis- cussion of the "Problem of Peasant Women," in Rural Planning in Developing Countries, ed. Raanan Weitz, Report on the Second Rehovoth Conference, Rehovoth, Israel, August 1963 (Cleveland: Western Reserve University Press, 1966), pp. 373-75. 78 DESIGN OF RURAL DEVELOPMENT to nine times greater than those obtained by traditional methods. Unfor- tunately the demonstration field approach to popularize the package has not resulted in widespread adoption-evidently either because the demon- stration fields have not been distributed widely enough or because more ac- tive promotional measures are needed.36 ZAPIs have erred, first, by postponing development of food crop technology until recently and then by failing to follow through on its extension effort. Many of the other pro- grams reviewed under ARDS do not even show an explicit recognition of the women's role in the farming system. Consequently, they have been even more remiss in extending agricultural extension to the female half of the population. Implications of Past Extension Experience for Organization of Future Extension It is apparent that the attempts to encourage genuine popular participa- tion in and support for program activities have often been ineffective. A step towards correcting this situation has been made in WADU in Ethiopia. WADU does not favor using the model farmer as a major tool for demon- strating new techniques, reasoning that, since only a few farmers may bene- fit from the actual demonstration process on their farms, the whole concept could be misinterpreted by non-model farmers as giving preferential treat- ment to a selected few. Rather, WADU has chosen a demonstrator field ap- proach. It uses a large number of farmers, employing fields of all of them for demonstrating the profitability of various farming techniques. Usually different farmers' plots are used to demonstrate different innovations. This approach seems to have been quite successful in WADU. As pointed out earlier, there has also been a larger-than-planned increase in the number of farmers the program has reached. However, it is unclear as to how much potential there is to use this ap- proach successfully in other programs. As noted earlier, a similar approach has not been particularly effective in ZAPI, though this may be explained by staff shortages and insufficient scale. WADU's success seems to be largely attributable to the fact that Wolamo farms tend to be of a rather uniform size with fairly even income distribution.37 According to WADU's management, its extension service has benefited considerably from the homogeneity of the Walamo society, which has led to a greater direct in- volvement of farmers in demonstrations than would have been possible in Chilalo. This involvement in turn lends credibility to the profitability of in- novations, for the innovations are demonstrated within the realistic resource constraints faced on farmers' fields. Further, WADU's demon- strators are usually experienced farmers from within Wolamo province 36. ZAPIs are handicapped by their limited number of animatrices (female extension workers), who are able to spend only one or two days a month in program villages. This short- age of staff has adverse effects on both productive and welfare-related women's programs. 37. For a discussion of related points, see Chapter VI of this study. AGRICULTURAL EXTENSION 79 who speak the local dialects and, therefore, have greater rapport with the local people.38 The effectiveness of involving farmers directly in the extension effort was also evidenced in the French technical assistance programs in -Cameroon. On the basis of their field experience in rural French West Africa, Belloncle and Gentil devised a method of organizing a systematic group exchange of ideas and experience regarding adoption of new innova- -tions. Through learning from each other, a process of mutual help was fostered among farmers. While this approach has not yet been applied on a large scale, it has shown a considerable potential for success when at- tempted by the reviewers. This chapter, therefore, ends with a summary of the approach they recommend. Before providing any technical assistance, Belloncle and Gentil recom- mend sponsoring group discussions in which farmers can discuss their own particular needs and problems. These discussion sessions always ought to be focused on the implications of technical change and its probable effect on each member of the group. If properly conducted, through such dis- cussions the farmers can come to see their individual problems as com- munity concerns which can be dealt with most effectively through com- munity action. The thrust of these meetings is both to instill self-reliance in the farmers and to create a more equitable relationship between them and the extension workers. Along with group meetings, Belloncle and Gentil advocate actual demon- stration of new farming methods. They stress the need for research applica- tion centers where farmers can view the results of technical schemes con- sidered feasible for general use.39 These preliminary steps can spark a genuine desire for change among the participants. To transform this desire into action, it is necessary to en- courage the group to select volunteers to try out different innovations. This is much like the traditional model farmer approach but with one important difference. In this case the innovators are innovators by consensus of the group, by delegation so to speak. After an appropriate trial period, these innovators will report on their successes and failures at a self-evaluation meeting in which accomplish- ments of individual innovators as well as the actual specific constraints en- countered by farmers in achieving the potential results can be identified. The problem then becornes one of taking steps to remove these constraints and of adopting the worthwhile innovations on a larger scale. At this stage 38. At the initial stage the Minimum Package Program experienced considerable difficulty in recruiting extension agents who could speak the local dialect. Therefore, the program had a very genuine problem of communication with farmers in some areas. 39. The crucial importance of combining discussion and demonstration is also illustrated by CADU's evaluation of its own extension program. According to Toborn, in CADU both visual and oral methods were used to promote innovations. Interestingly, visual influences were found to be more effective in advancing recently introduced innovations, while oral in- fluences had their greatest success with well-established innovations. In addition, the more progressive farmers were more strongly impressed by extension agents and demonstration plots than were their less progressive neighbors. These non-adopters tended to rely on more familiar sources (i.e., on friends, relatives, and, only to a limited extent, on model farmers) for information on new farming techniques. Toborn. The Innovation-Diffusion Process in the Gonde Area. 80 DESIGN OF RURAL DEVELOPMENT the extension agent's expertise is required. But once again it is his relation- ship to the farmers which is crucial. He should act as an advisor from whom the farmers seek information which they deem important rather than as a haughty or an inexperienced stranger handing down unsolicited advice to subordinates. The review of the French technical assistance programs indi- - cates that such self-appraisal meetings can be exceedingly successful in ex- pediting change, in particular where there exists a spirit of rivalry between individuals within a group or between various groups. With this approach, after a point the groups can become more or less self- perpetuating. Former innovators can then be trained to take over many of the functions of extension agents and to act as conduits for further techni- cal information. Belloncle and Gentil argue that ideally these vulgarisateur peasants should draw their salary from the group itself rather than from the government. Thus, the continued grass root nature of the service and its high performance would be ensured. It is important to emphasize, however, that such a group approach can work only when there is relative equality in the socioeconomic status of the target population. CHAPTER V Agricultural Credit Some Issues Related to Organization of Agricultural Credit In many of the countries reviewed under ARDS, little of the institutional agricultural credit has been directed towards development of smallholder subsistence farmers. Even as late as 1971, credit records of the Agricultural Finance Corporation in Kenya showed that 88 percent of the gross loans outstanding had gone to large-scale farmers, defined as those whose annual income exceeds KShI5.,000. Of this, 61 percent went for the purchase of lands from white settlers and 20 percent for dairy development and cattle. Less than one percent had been distributed as working capital. Despite the increasing proportion of funds being earmarked for small farmers, in the same year over 70 percent of AFC's disbursements went to land purchases, commercial farm development, ranching, and similar large-scale uses. It is evident that AFC's loans are missing the bulk of the farmers. In the Kericho district, for example, per capita loans averaged KSh4,294-i.e., they were almost three and half times larger than the average annual farmer income in the area. A similar situation exists in Malawi, where until re- cently institutional agricultural credit consisted mainly of the Master Farm- er Program directed towards development of the relatively larger commer- cial African farmers. Until the establishment of the Lilongwe Land Devel- opment Program in 1967, traditional subsistence agriculture received vir- tually no institutional credit. In Ethiopia one of the main objectives in the establishment of the Agri- . cultural and Industrial Development Bank (AIDB) was to meet the pressing needs of the agricultural sector.' Because of its minimum loan, collateral, -and other security requirements, over 80 percent of the farming population in the country has not been eligible for credit from AIDB. Even though about 63 percent of the loans distributed during the first two years of full -operation were distributed to activities in the agricultural sector, big com- mercial farmers and a few cooperatives were the major beneficiaries. Of about Eth$19 million distributed to the agricultural sector in 1972, three large-scale farms and 107 individual farmers accounted for nearly 79 per- cent of the loans. Even in Tanzania one of the major concerns of the policymakers has been w;o devise ways of reaching farmers who do not pro- 1. AIDB was established on August 28, 1970, by merging the former Development Bank of Ethiopia and the Ethiopian Investment Corporation. 82 DESIGN OF RURAL DEVELOPMENT duce traditional export crops and, therefore, have not benefited from in- stitutional credit. Thus, despite its magnitude in terms of the proportion of the population and the percentage of income generated, until recently the traditional sub- sistence sector has had to rely almost exclusively on informal sources of credit generated from within the rural areas. A strong argument is, therefore, often made for changing the composition of agricultural credit distribution through the introduction of rural development programs specially designed for small-scale subsistence farmers. On the basis of evidence in the past small-farmer credit programs, this chapter discusses the following points: 1. The potential for mobilizing savings from the traditional rural sector and, hence, the strength of the capital constraint argument as the basis for credit programs. 2. The effect of the limited trained manpower and of the conventional approaches to the institutional credit administration on costs of credit distribution; costs of recovery; and, hence, on reaching a mass of the rural poor in a relatively short time span. 3. The factors that affect the profitability of innovations and the variance in the profitability as well as the effect of these factors on credit repayment. 4. The implications of the past experience for organization of future rural credit programs. It is assumed that in the case of traditional high-value export crops, which require centralized procurement for further processing, delivery and recovery of institutional credit through parastatals or cooperatives is relatively easy; and the knowledge as to the organization of such credit is relatively more widespread. This chapter, therefore, deals with credit ser- vices for those crops for which the technology has usually involved high risk, the value of the crops has been relatively low, and the alternative chan- nels for disposal of the crops have been easily available. All these factors hold in the case of most food crops and several non-food crops. They sig- nificantly increase the difficulty in organizing credit programs. Potential for Saving in the Subsistence Sector Few baseline surveys have been conducted in designing rural develop- ment programs prior to establishment of a credit service. Consequently, lit- tle hard information is available to program planners on the target popula- tion's saving propensities or the sources from which it obtains credit.2 Since incomes are very low and the saving potential appears to be very small, the credit components in almost all the rural development programs reviewed 2. There is, however, considerable evidence on rural savings capacity in Africa. See, for in- stance, 0. Amogu, "Some Notes on Savings in an African Economy," in Social and Economic Studies, vol. 5, no. 2 (1959), pp. 202-09; Giordano Dell'Amore, The Mobilization of Savings in African Countries (Milan: Cassa di Risparmio delle Provincie Lombardi, 1971); R. Galetti, K.D.S. Baldwin, and l.O. Dina, Nigerian Cocoa Farmers: An Economic Survey of Yoruba Cocoa Farming Families (London: Nigerian Cocoa Market Board, 1956); Polly Hill, The Gold Coast Cocoa Farmers: A Preliminary Survey (London: Oxford University Press, 1956); Polly Hill, The Migrant Cocoa Farmers of Southern Ghana: A Study in Rural Capitalism (Cambridge: Cambridge University Press, 1963); Polly Hill, Rural Hausa: A Village and a Setting (Cambridge: Cambridge University Press, 1972); and Polly Hill, Studies in Rural Capitalism in West Africa (Cambridge: Cambridge University Press, 1970); G. Hubner, "Private Savings in Uganda," in AGRICULTURAL CREDIT 83 have been based on the assumption that seasonal credit is one of the major constraints to rural development. However, von Pischke's review of smallholder credit in Kenya suggests that smallholders have a greater capacity to generate savings than is commonly assumed. For example, deposits totaling KShl' million by 1973 (ust three years after the rural banks had initiated the Cooperative Thrift Scheme) were totally unantici- pated by officials in Nairobi. Also indicative of rural savings capacity is the existence of indigenous mutual savings societies. Von Pischke's survey of nineteen farmers in the Murang'a district revealed that in one such group each member would contribute KShlS at regularly held monthly meetings, with the total amount thus accumulated then distributed to one member of the group on a rotating basis. This process would continue until everyone had received the monthly pool, whereupon the cycle would be repeated. Similar saving groups., or tontines, exist among Ewondo women in Cameroon and Chilalo women in Ethiopia. The survey conducted by CADU's Evaluation Unit on the traditional sources of credit in two areas (Digela and Yeloma) within Chilalo Awraja, points out that agricultural investment-such as buying a new plow, hiring an extra day of labor for weeding, and so forth-has been an important reason for borrowing. Less was borrowed to fulfill consumption needs.3 Fifty-one percent of the farmers interviewed whose average annual family income was estitmated at about Eth$300 (US$130) had some debt. Out of the 109 farmers sampled for the study, 41 percent were owner-culti- vators and 59 percent tenants. Overall, 54 percent of the loans were used for improvements on the farms and 46 percent for consumption, including feast expenses, tax payments, and court case expenses. The average size of the loan was estimated to be Eth$64 for the productive loans and Eth$51 for the consumption loans. The study also indicated that only 36 percent of the total number of loans were repaid upon the agreed due dates. The survey further showed that the major sources of credit were rela- tives, close friends, and other farmers. Only 6 percent of the credit was bor- rowed from traders and moneylenders. About 64 percent of the number of loans from relatives and friends were found to carry no interest charge, compared to only 10 percent of the number of loans from other sources; the average annual interest rate ranged from 34 percent for the loans from rela- tives and friends to about 110 percent for the loans from other sources. In the case of cash loans, the most common interest charged was found to be 10 percent per month.4 Financial Aspects of Development in East Africa, ed. Peter von Marlin (Munich: Weltforum Verlag, 1970), pp. 93-174; William 0. Jones, "Economic Man in Africa," in FoodResearch In- stituteStudies, vol. 1, no. I (May 1960), pp. 126-33; Benton F. Massell, "Consistent Estimation of Expenditure Elasticities from Cross-Section Data on Households Producing Partly for Sub- sistence," in Review of Economics and Statistics, vol. 51, no. 2 (May 1969), pp. 136-42; R.A.J. Roberts, "The Role of Money in the Development of Farming in the Mumbawa and Katete Area of Zambia," Ph.D. dissertation, University of Nottingham, 1972, and Hans Seibei and Andreas Massing, Traditional Organizations and Economic Development: Studies oJfIndigenous Cooperatives in Liberia (New York: Praeger, 1974). 3. G. Bergman and H. Lindqvist, CreditSituation in Chilalo Awrala, CADU Minor Research Task no. 3 (Asella, Eth.: Chilalo Agricultural Development Unit, July 1969). 4. In personal communication with the author, John Cohen pointed out that loans in kind are prevalent and also carry high interest rates. According to Cohen, it is not uncommon for farmers to borrow one quintal of wheat in February and have to return three quintals during the harvest period. 84 DESIGN OF RURAL DEVELOPMENT Thus, although interest rates charged by moneylenders may have been prohibitive, it is by no means certain that the average effective interest rate was very high in the two Chilalo areas. In regard to the security on such loans, the survey indicates that usually some type of security was required for the loans. However, only 46 percent of the relatives and friends, com- pared to 95 percent of the other lenders, required security for their loans. Of the loans analyzed in the sample survey, 67 percent had some kind of writ- ten agreement, 60 percent had between one to three witnesses, 44 percent had guarantors, 9 percent of the lenders required land as a collateral, and 73 percent of the loans had a collateral of some sort. Only 27 percent of the loans were given without any kind of collateral or security. Though the results of the Digelu-Yolema areas credit study are too limited to base conclusions for the entire CADU project area, interviews and discussions with the various CADU staff, who had been making simi- lar observations in other parts of Chilalo, indicated that the Digelu-Yolema areas situation may, in fact, have been quite representative of Chilalo. The CADU survey indicates that the traditional rural sector may have greater potential for mobilizing savings than is generally presumed. The survey carried out by WADU's Planning Unit is similar but covered a wider area. All areas within Wolamo Awraja, except those within the WADU project area, were included in the sample survey. The findings show a remarkable similarity with those in the Digelu and Yolema areas. For example, neighboring farmers were found to be the major lenders. Six- ty-six percent of the 33,000 households surveyed borrowed from other farmers, compared to only 10 percent from traders. Twenty-eight percent of the surveyed households paid no interest at all, about 26 percent paid in- terest ranging from one to 50 percent per annum, and almost 40 percent of the households paid interest rates as high as 100 percent per annum. Almost all of the loans were given in cash with less than 5 percent being given in kind. This seems rather unusual; but most of the extension, credit, and marketing staff at WADU felt that it was because people in Wolamo borrowed mainly to finance feast celebrations. A separate sample survey, conducted within the WADU project area in 1971, arrived at conclusions similar to those above.5 Institutional Smallholder Credit and Formal Administrative Requirements The Case of Ethiopia Despite this evidence in all the three Ethiopian projects, distribution of institutional agricultural credit constitutes one of the central foci of project activities. Table 5.1 shows the amount of credit distributed in these proj- ects, while Table 5.2 shows the amounts of inputs distributed. In CADU the number of loans increased to nearly 14,000 in four years. The decline in the number of loans in the 1971/72 and the 1972/73 periods is partly related to 5. It is instructive to compare the interest rates charged by traditional creditors with those charged by the programs. On seasonal credit AFC charges annual interest of 7.5 to 8.0 per- cent. In LLDP the rate is 10 percent, while in CADU and WADU it is 12 percent. AGRICULTURAL CREDIT 85 its marketing problems discussed later in Chapter VI. The eviction of ten- ant borrowers also contributed to the drop in CADU's loans.6 In WADU over 7,000 loans were advanced in the third year, as compared to the 4,000 anticipated. In MPP areas 4,500 farms were reached in the second year. The approach to credit administration has been generally similar in the three programs. For example, all provide their loans in the form of farm in- puts. However, there are some distinct differences. Unlike CADU and MPP, WADU gives short-term cash loans for consumption purposes. Wolamo farmers traditionally borrow credit from private sources for holi- day celebrations which occur just prior to the harvest time. WADU's justification for the cash loans is that, if it does not advance consumption credit, the farmers' ability to repay WADU's productive credit will be jeopardized by their need to pay off costly private loans. There are also variations in the eligibility criteria for credit. After two years' experience with the unrestricted participation of all classes of farm- ers in CADU's credit program, it became apparent that the big landlords were the major beneficiaries of credit in the Chilalo area (see Table 5.3). Furthermore, a combination of easy availability of credit, demonstrated profitability of innovations, and absence of taxes on tractors and fuel con- tributed to the eviction of over 500 tenant families in Chilalo during 1969 and 1970. Consequently, land-owning cultivators with over 25 hectares and tenants cultivating more than 40 hectares were excluded from CADU's credit program in 1970. By 1972 CADU had lowered the limit to 20 hectares for owner-cultivators and 30 hectares for tenants.7 MPP had imposed simi- lar restrictions since its inception. Still, landowners, who constituted less than half the population, received the bulk of the loans (see Table 5.4). WADU, on the other hand, had allowed all classes of farmers to partici- pate in its credit program as long as they were full-time farmers. However, landholdings in Wolamo have been small, averaging only 0.5 hectares, com- pared to 3.0 hectares in CADU. One rarely noted cultivators with over 2.0 hectares of cultivable land participating in WADU's credit program, despite the fact that the program has been open to all categories of farmers. Thus, although landowners received the lion's share of benefits, distribution was actually more equitable than appears at first (see Table 5.5). WADU's security requirements have been considerably more liberal than in either CADU or MPP. Borrowers were asked only to present two guarantors acceptable to the screening committee. In contrast, both CADU and MPP required down payments, signed lease agreements between ten- ants and landlords, and two acceptable guarantors-one of whom had to be the landlord if the borrower was a tenant. Apart from their use as a security measure in absence of a land tenure reform, the one-year lease agreements were developed by projects principally as a stopgap arrangement to avoid eviction of tenants. According to Tecle, as a further security measure all three projects adopted the policy of holding farmers jointly responsible for all of the credit extended in their areas. 6. See Chapter It of this study and John M. Cohen, "Rural Change in Ethiopia: The Chilalo Agricultural Development Unit," in Economic Development and Cultural Change, vol. 22, no. 4 (July 1974), pp. 608-14, esp). 70n. 7. Nonetheless, tenant evictions appear to have continued. Cohen estimates that 6,000 ten- ants may have been evicted by 1974. Ibid, 64n. Table 5.1: Amount of Credit and Number of Loans Extended in Three Rural Development Programs, Ethiopia, 1967/68-1972/73a Chilalo Agricultural Wolamo Agricultural Minimum Package Program Development Unit Development Unit Amount Amount Amount Number of loan A verage Number of loan A verage Number of loan A verage Crop year of loans (Eth$) loan (US$) of loans (Eth$) loan (US$) of loans (Eth$) loan (US$) 1967/68 189 15,700 83 _ b _ b _ b _b bb b 1968/69 868 158,500 182 _ _ b _ b _ b b _ b 1969/70 4,769 502,900 105 727 24,000 33 _c -c -c 1970/71 14,146 1,437,500 102 3,923 80,200 20 3,613 240,000 67 1971/72 12,624 1,063,100 84 4,791 159,300 33 4,609 370,000 80 1972/73 13,301 961,950 72 7,400 304,000 41 _ d d _ d a. Figures for the total amount loaned and average size loan have been rounded. b. Program not implemented. c. Figures werc insignificant. d. Figures were not available at the time the review was conducted. Source: Chilalo Agricultural Development Unit figures were taken from D. Yirgou, G. Hunter, S. Bekure, and Ht. Ryden, Final Report on the Appraisal of CADU and EPID (Addis Ababa: Government of Ethiopia and the Swedish International Development Authority, May 1974), p. I1, Table 2; Wolamo Agricultural Development Unit and Minimum Package Program figures were obtained from Credit Section reports from the two projects. AGRICULTURAL CREDIT 87 Table 5.2: Seed and Fertilizer Distributed on Credit in Three Rural Development Programs, in Quintals, Ethiopia, 1967/68-1972/73 Chilalo Agricultural Wolamo Agricultural Minimum Development Unit Development Unit Package Program Crop year Fertilizer Seed Fertilizer Seed Fertilizer Seed 1967/68 42 1,470 _a _ a _ a _ a 1968/69 2,820 4,540 a _a _ a _a 1969/70 15,380 8,202 144 117 6,160 154 1970/71 41,461 13,434 1,420 367 9,280 _ b 1971/72 35,309 2,698 _ b _b _b _b 1972/73 32,051 1,640 _ b _b _b _b a. Program not implemented. b. Figures were not available at the time the review was conducted. Source: Government of Ethiopia, Chilalo Agricultural Development Unit, Marketing Division, "Preliminary Market Data for 1972/73 Annual Report," mimeographed, January 1974; and in- formation supplied by the Credit Sections of the Wolamo Agricultural Development Unit and the Minimum Package Program. What effect has the use of these criteria of creditworthiness had on par- ticipation of the lowest income groups? Though not definitive, interviews with some field agents and project staff have indicated that, due to the rather stringent security requirements of CADU and MPP, the poorest seg- ments of the target population have been excluded, either because they could not afford the required down payment or because their landlords were not willing to sign the lease clause. Even in the case of WADU, where the only requirement has been that a borrower have two reputable guaran- tors, interviews have indicated that in some cases the poorest tenants found it difficult to find guarantors acceptable to the screening committees. Besides, under sharecropping arrangements, where all costs-including those for major improvements on the land- were borne by tenants without any compensation, teinants often did not find credit for yield-increasing in- puts very rewarding. Assuming that a tenant received a favorable sharecropping arrangement, which required paying the landlord one-third of his harvest as rent plus a tithe, over a 60 percent increase in yield was necessary to make the use of improved seeds and fertilizer profitable to the tenant. By contrast, any net increase in yield was profitable to the landlord. If costs were shared proportionately, a yield increase of about 30 percent made the venture profitable to the tenant under the above assumptions.8 The three programs also vary in the degree of local participation in credit delivery. In CADU credit applications received preliminary approval from an extension agent and final approval from the Commerce and Industry Department. Plans had been made to employ a screening committee com- posed of local farmers and extension workers, but this approach had been handicapped by organizational difficulties.9 WADU has also used a screen- 8. An evaluation team report points out that rents were often much higher, ranging up to 60 percent of the tenant's produce. The tithe or asrat was illegal, but many landowners con- tinued to demand it. D. Yirgou et al., FinalReport on theAppraisaloJlCADU andEPID (Addis Ababa: Government of Ethiopia and the Swedish International Development Authority, May 1974), p. 42. 9. For a discussion of some of the difficulties in building farmer committees, see Cohen, "Rural Change in Ethiopia: The Chilalo Agricuitural Development Unit," pp. 607-08. Table 5.3 Number and Percentage of Credit Takers and Distribution of Gross Benefits, by Land Tenure and Size of Holding, CADU, Ethiopia, 1968-71 Landowners (holdings in hectares) Item Tenantsa 1-10 11-20 21-40 Above 40 Unknownb Total 1968 Credit takers Number 16 81 36 21 18 17 189 Percentage 8.5 42.8 19.0 11.1 9.5 8.9 100.0 Percentage of benefits 3.5 14.8 19.3 29.3 31.1 2.1 100.0 1969 Credit takers Number 134 438 144 66 54 32 868 Percentage 15.4 50.5 16.6 7.6 6.2 3.7 100.0 Percentage of benefits 5.5 29.1 21.9 12.8 26.9 4.9 100.0 1970 Credit takers Number 1,540 1,903 441 179 54 652 4,769 Percentage 32.3 39.9 9.3 3.8 1.1 13.7 100.0 Percentage of benefits 22.7 37.2 17.9 8.9 3.3 9.9 100.0 1971 Credit takers Number 5,392 8,212 387 80 0 75 14,146 Percentage 38.1 58.1 2.7 0.6 0 0.5 100.0 Percentage of benefits 35.8 58.2 4.5 1.5 0 0 100.0 a. Approximately 90 percent of the tenant credit takers in the 1969170 period cultivated less than 6 hectares. See Henock Kifle, An Analysis of the CADU Credit Programme 1968/69-1970/71 and Its Impact on Income Distribution, CADU Publication no. 66 (Assella, Ethiopia: Chilalo Agricultural Development Unit, August 1971). b. Unknowns are farmers for whom status of land tenure or area owned was not obtained trom credit applications. Source: Johan Ilolmberg, ''The Credit Programme of the Chilalo Agricultural Development Unit (CADU) in Ethiopia," in Aid Small Farmer Credit, Spring Review, vol. 8, no. 180 (February 1973), p. 55. AGRICULTURAL CREDIT 89 Table 5.4: Distribution of Inputs on Credit by Tenancy and Sizes of Farms in Nine Minimum Package Areas, Ethiopia, 1971/72 Quantity of fertilizer Beneficiaries distributed Classification Number Percent Quintals Percent By tenancy Tenants 507 21.8 1,627 32.6 Landowners 1,819 78.2 3,373 67.4 Total 2,326 100.0 5,000 100.0 By area cultivated (hectares) 0-4.9 1,325 57.0 1,573 31.5 5-9.9 600 25.8 1,604 32.0 10-14.9 233 10.0 830 16.6 Above 15 168 7.2 933 19.9 Total 2,326 100.0 5,000 100.0 Source: Compiled from Extension and Project Implementation Department credit files and Minimum Package Program loan application to the World Bank, July 1972. Table 5.5: Distribution of Inputs on Credit by Tenancy of Farms in the Wolamo Highlands, WADU, Ethiopia, 1971/72 Quantity of fertilizer Beneficiaries distributed Classification Number Percent Quintals Percent Tenants 199 11 128 11 Tenant-owners 112 7 75 6 Landowners 1,460 82 988 83 Total 1,771 100 1,191 100 Source: Calculated from information on the Wolamo Agricultural Development Unit Year 2 Credit Pro- gram. ing committee, though in this case it has been composed only of farmers. The WADU committee makes the final decision as to who is eligible for credit before bulk-orcler dispatches are sent to the Marketing, Credit, and Cooperatives (MCC) D)ivision. The emphasis on grass root participation has been the lowest in the Minimum Package Program. Credit applications have been processed by supervisors and their extension staff with some help from model farmers. How effective have these programs been in fostering farmer participation in the administration of credit? CADU had difficulty even to bring commit- tee members together, whereas WADU reports that the farmers' commit- tees have been meeting regularly and have exerted considerable effort in trying to do a good job. Since neither CADU nor WADU has built-in incen- tives for committee members, it appears that the success in WADU is due to the higher level of awareness of the Wolamo people with respect to the project's functions and to the facts that the Wolamos have historically been socially cohesive and have traditionally been familiar with group actions and organizations. As pointed out earlier in Chapter IV, the social cohesion in Wolamo seems to be closely related to the rather equitable distribution of land rights 90 DESIGN OF RURAL DEVELOPMENT and is reflected in the broader participation of Wolamo farmers in delivery of extension. As will be discussed in Chapter VI, this may also explain the rather more encouraging start in the development of cooperatives in WADU. WADU's superior performance in repayment of credit, discussed in Section 4 of this Chapter, may also be explained by these social fac- tors.'0 However, often the response of the Wolamo has also been attributed to respect for authority and to a lack of any past assistance by the govern- ment rather than to the social organizational factors mentioned above. Tak- ing these various considerations into account, to what extent is the group responsibility approach in WADU replicable elsewhere? An indication of this may be obtained by comparing the three Ethiopian programs with LLDP in Malawi. Credit in the Lilongwe Land Development Program There was little information available to the planners of LLDP on the nature of the rural credit market in Lilongwe. In the appraisal of LLDP, it was anticipated that, although inputs may initially be distributed on credit, cash sales should acquire an increasing importance over time. However, for a variety of reasons that will be apparent in this discussion, credit rather than cash sales became the central focus for expanding the use of fertilizer and improved seeds. Table 5.6 shows, for years of operation, the sharp in- crease in the number of borrowers and in the total amount loaned, the grad- ual decline in the repayment rate in terms of the proportion of value loaned recovered, and the increasing number of defaulters. It is noteworthy that in one year the number of credit recipients in LLDP increased sharply from about 4,400 (1970171) to almost 20,000 (1971/72). There has been a particularly dramatic increase in the amount of credit issued for both fertilizer and groundnut seeds since the 1969/70 period (see Table 5.7). Credit has obviously been the major vehicle for the rapid promo- tion of purchased inputs. Kinsey's detailed reappraisal of LLDP indicates that cash sales are lag- ging considerably behind projections, apparently because the credit pro- gram is serving a larger-than-anticipated number of repeat customers. For example, in the 1970/71 period it was estimated that farmers would purchase almost 3,000 metric tons of fertilizer, 1,765 of which would be paid for in cash, the remainder being purchased with credit advanced by LLDP. However, when the actual records came in, it was discovered that almost 4,000 metric tons were purchased, of which only 58 metric tons were purchased with cash. The lack of progress on cash sales, is, however, by no means an indica- tion of their poor potential. Farmers do, on occasion, spontaneously band 10. Some observers of the Ethiopian programs, such as Guy Hunter, have argued that, in addition to the more favorable social environment in Wolamo, the management of WADU has also made a more conscious effort to utilize the traditional social groupings than has been made in CADU. An evaluation of WADU, carried out earlier by Paulos Abraham, the former director of CADU, and reported in Chapters VI and VIII of this study, holds a much more skeptical view of WADU's management in this regard. This is despite the fact that Abraham in his personal conversations with the author emphasized that even CADU should have made greater use of traditional groupings of low-income target farmers. These conflicting assessments emphasize once again the inadequacy of systematic comparative analysis and the predominance of subjective elements in evaluation of the important institutional questions. AGRICULTURAL CREDIT 91 together and act as a group when it is advantageous to do so. For example, to take advantage of a 7 percent discount and to obtain fertilizer before pro- gram staff were posted to the locale, farmers in the southwestern Modified Input Area (MIA) grouped together to purchase on a bulk basis 505 metric tons of sulphate of ammonia fertilizer from the Chileka ADMARC market. Such group sales reportedly have been a fairly common method of obtain- ing fertilizer for farmers outside the program area where there is no credit program. These purchases are on a cash basis. If the southwestern MIA is considered the equivalent of two standard program units, fertilizer inputs on a cash basis here are comparable with inputs on a credit basis elsewhere in the program area. The implication is that, if credit is not available, at least some farmers are willing to pay cash for nonfarm inputs. By contrast, with- in the program area where credit issues received heavy emphasis, only 34 metric tons of fertilizer (less than one percent of the total issue) were sold for cash as recently as the 1972/73 period. Table 5.6: Seasonal Credit: Amount Loaned, Number of Borrowers, and Default Rates, LLDP, Malawi, 1968/69-1973/74 Percentage Percentage qf credit Percentage of credit Total recipients of credit repaid amount Number defaulfing by not repaid as of loaned of August 31 by August 31 April 1, Year (kwacha) borrowers of each year of each year 1974 1968/69 4,830 660 54 34 100 1969/70 7,040 560 7 4 100 1970/71 83,310 4,410 44 14 99 1971/72 328,660 19,800 44 21 95 1972/73 392,400 21,470 41 25 88 1973/74 495,920 25,620 _ a - a _ a Total 1,312,160 72,520 42 22 92 a. Figures were not available at the time the review was conducted. Source: Data supplied by the credit section of the government of Malawi, Lilongwe Land Development Program and by the program's Quarterly Report (n.p., January-March 1974). Table 5.7: Total Inputs Supplied on Credit and Hectarage Covered, LLDP, Malawi, 1968/69-1972/73 Fertilizer Maize seed Groundnut seed Credit Metric Metric Metric year tons Hectares tons Hectares tons Hectares 1968/69 151.3 611.3 41.3 1,672.1 12.3 101.2 1969/70 125.5 510.1 162.6 6,583.0 11.7 93.1 1970/71 814.4 3,297.2 213.6 8,647.8 102.3 830.0 1971/72 3,877.6 15,700.4 223.2 9,036.4 206.3 1,672.1 1972/73 3,820.1 15,465.6 129.7 5,251.0 227.1 1,838.1 Source: Government of Malawi, Lilongwe Land Development Program. Credit Section, Repou jlb the Year 1970 (n.p., nd.): and Monthly Report, various issues. Other data supplied by the program's Credit and Marketittg Sections. 92 DESIGN OF RURAL DEVELOPMENT In a subsequent field survey in the program area, conducted as part of the follow-up of the first evaluation, fifty-nine of the one hundred farmers in- terviewed at random showed interest in group cash sales. The evidence, thus, indicates that there may be far greater potential to promote either cash sales or some form of local group financing of inputs than is usually presumed in rural development programs. This form of credit mobilization may, however, require far greater flexibility in the rate of project expansion and far greater willingness to innovate and undertake the risk of not reaching ambitious targets than is usually allowed for in de- signing rural development programs. Since the knowledge as to the precise potential of these methods of credit mobilization is so limited, there is an uncertainty as to their results. In addition, greater time may be required for these methods to come to fruition. However, such new approaches seem to be necessary, for, as the evidence cited below illustrates, small-scale credit distribution on a mass scale, using traditional institutional approaches, seems neither administratively practical nor financially feasible. LLDP's procedures for assessing creditworthiness are even more elabor- ate than those in the Ethiopian projects. The credit system has been predi- cated on the assumption that a credit recipient must possess a certain stan- dard of managerial ability in order to achieve the potential benefits and high returns from packages of new farm inputs. Managerial ability is assessed by a combination of two factors: 1. A personal evaluation of the individual's ability that is made by qualified persons who are in close contact with the applicant. (Farmers are registered and placed into different credit groups according to cer- tain personal characteristics used as indicators of trustworthiness.) 2. A cumulative rating of all growers, based on certain husbandry aspects, that is maintained by extension agents." The farmer identification list is submitted to the local village planning com- mittee, which, with the assistance of the village headman, goes through the list and evaluates the character and reliability of each grower. This list is then forwarded to the development officer, who combines the personal evaluations with the managerial rating. The officer then forwards a list of approved applicants to the credit section, where the authorization forms are prepared. The use of this mix of objective and subjective criteria, fairly widely ap- plied, seems to deserve much of the credit for the high repayment rates that have been achieved in LLDP in the first few years. The system, however, imposes a very formidable work load for each extension worker (nearly 300 growers per worker). This may result in shortcuts being taken in the rating system by the arbitrary assignment of the previous year's, or any other year's, rating. The rating system has also been criticized frequently for 11. In the past this rating was expressed in terms of a percentage scale; but, as this system resulted in a very wide range of ratings and made cut-off points difficult to interpret meaningfully, it has been modified and only four ratings are now given: poor, average, good, and excellent. One of these ratings is assigned by the extension agent to each of the seven aspects of crop husbandry-ridging, planting, spacing, fertilizer use, weeding, banking (the sec- ond or third weeding), and harvesting. The evaluation of the ratings is done by the develop- ment officer in charge of the unit where the applicant lives. If the applicant resides in one of the longer established units, his rating is compared with that of the previous year in order to see if he has improved his standards of husbandry. A farmer with more than three poor rat- ings out of seven items is considered ineligible for credit in any event. AGRICULTURAL CREDIT 93 being applied very rigidly in the first few years of the credit operation, resulting in a spotty and selective pattern of adoption. The criticism is to some extent justifiecl. The detailed statistical analysis of socioeconomic data carried out in thie follow-up study of LLDP revealed that the credit- worthiness rating was rarely carried out as it was meant to be. Both in the 1970/71 period and the 1971/72 period, some 55 percent of the project farm- ers in the sample had not been rated. In some of the project's administrative units, no farmers had been rated. The analysis further indicated that farm- ers receiving program credit have a higher-than-average socioeconomic status, cultivate one-third more land, and have incomes that are 84 percent higher than those who do not receive program credit. Nevertheless, LLDP's concentration of credit on the relatively more wealthy farmers is lesser when compared to the government credit program which operated in the area prior to the project. Factors Affecting Repayment of Credit The issue of credit repayment is closely related to that of administrative procedures and raises a number of complex questions. For example, to what extent have the elaborate administrative procedures used for credit distribution assured high credit repayments? Has rapid expansion of the programs affected repayments? How feasible is it to expand credit adminis- tration on a massive scale beyond the first few thousand farmers? In this section these questions are examined on the basis of experience in Ethiopia, Malawi, an.d Kenya. In addition to the interaction between ad- ministrative factors and repayment outlined above, attention is focused on the implications of overall crop profitability for repayment rates. Repayment in CADU, WADU, and MPP The experience of the Ethiopian projects with regard to credit repay- ments provides useful insights into the dynamics of social change generated by the credit prograrns. Analysis of the repayment figures in the three projects shows that small- scale farmers are by n1o means worse defaulters than large-scale farmers. In fact, a higher percent of the defaulters were found to be the larger farmers, possibly because, being politically powerful, they feel that they can renege on their debts with impunity. The refusal of these larger farmers to settle their dues has probably led to a more far-reaching problem in the case of CADU. The due-date repayment rates have been declining gradually as the number of borrowers increased. CADU's failure to take the first defaulters to court was very likely taken as a sign of weakness by other farmers, in- dicating that no action would be taken against them if they failed to repay. Actually, CADU had plans to take defaulters to court, and that fact is made clear to all farmers before they are given credit. But it could not effectively litigate for a number of reasons. First, the court system has been so ineffi- cient that prompt action, as CADU would require, was impossible under the setup that existed then.'2 Second, close cooperation by the local admin- 12. In personal communications Cohen indicated that court corruption rather than court inefficiency was the major reason that CADU had difficulty prosecuting defaulters. 94 DESIGN OF RURAL DEVELOPMENT istration is necessary in order to be effective in enforcing repayments; this has not been forthcoming. Third, local interest groups were taking advan- tage of the situation by complaining that CADU did not come to help farm- ers but to imprison them. Without the full support and cooperation of local institutions and the local administration, the projects had found it difficult to enforce their policies. The Ethiopian evidence raises doubts about the general contention that, because small farmers tend to use their increased income or to divert pro- duction credit more for consumption rather than for repayment of their debts, they are a greater credit risk than large farmers. Nevertheless, CADU and MPP have required down payments. In WADU no down payment was required until recently. WADU was able to achieve repayment rates of 98 percent and 95 percent in the first two years, compared to about 90 percent, 85 percent, and 92 percent during the first three years in CADU'3 and about 90 percent in the first year in MPP.'4 There are a number of reasons why WADU may have achieved a relatively higher repayment rate. These reasons emphasize that the profit- ability of investment interpreted in its broadest sense may be much more important than stringent security. In addition to grains, farmers in the WADU project area produce cash crops like coffee, whose prices have recently remained at higher levels than those of cereals, giving the farmers a better chance of making timely repay- ment. In contrast, farmers in CADU and MPP have grown mainly cereals, whose prices have been declining, jeopardizing the benefit of the credit pro- gram to the borrowers and consequently reducing their potential to repay the dues. The most important reason seems to be related to the Wolamos' traditional habit of borrowing for consumption purposes, noted earlier. Since WADU decided to provide consumption credit at low interest rates, the farmers have wanted to maintain good faith with WADU by showing a willingness to repay on time. This desire was reinforced by WADU's policy, followed from the outset, of excluding all farmers from a given area from future credit programs if repayment for the area fell below 95 percent. CADU had just recently adopted such a policy. However, it is unclear as to what extent the group responsibility for repayment has been enforced rigorously in practice in CADU and further whether the Ethiopian experi- ence with credit organization and repayment can be generalized to the rest of sub-Saharan Africa.'5 Ethiopia's unique feudalistic heritage may have fa- cilitated interventions which would be ineffective in areas with different sociopolitical traditions. 13. Cohen has argued that CADU's final repayment rate is somewhat higher than these figures indicate, averaging around 95 percent. See Cohen, "Rural Change in Ethiopia: The Chilalo Agricultural Development Unit," 41n. 14. By contrast, AFC in Kenya has had a collection rate on both its large- and small-scale loans of only 75 percent. This is despite the fact that, according to the existing estimates, the financial rate of return to small-scale AFC borrowers is 40 percent and above. Von Pischke at- tributes AFC's low repayment rate primarily to poor administrative procedures in loan super- vision and follow-up. Tighter discipline would increase the repayment rate dramatically. However, because the agency is a publicly supported institution, there is no pressure to make a profit in order to survive. Further, given the widespread arrears on AFC loans, there might well be political opposition to tighter loan discipline. 15. The evaluation of CADU carried out recently has urged that CADU eliminate group punishment for credit defaults, arguing that this practice unfairly penalizes small farmers. See Yirgou et al., Final Report on the Appraisal ot CADU and EPID, p. 23. AGRICULTURAL CREDIT 95 Repayment in LLDP Though LLDP has been quite effective in ensuring final repayment of credit, punctual repayment has been a consistent problem (see Table 5.6). Defaulters are handled in two ways: (a) granting of moratoria until the opening of the markets the following season to farmers who agree to certain conditions; or (b) taking defaulters to court. However, as in CADU, the courts' ability or willingness to enforce repayment laws has deteriorated. In certain areas defaulters appear to have read the courts' inaction as a sign of weakness, and court rulings are often held in contempt with the knowledge that no action will be taken if a defaulter fails to repay. What explains defaults? Analysis of the data collected by LLDP's Evaluation Unit provides conflicting explanations. Default rates were found to be significantly higher for farmers growing groundnuts. Of all the borrowers in the sample, 14.7 percent defaulted, while the figure for farm- ers with groundnut packages alone was 18.5 percent and for those with groundnuts plus other packages, 34.6 percent. The lowest default rate, 7.0 percent, was found among those growers who had only tobacco packages. For a variety of reasons discussed earlier in Chapter III, gross returns per man day have been lower for groundnuts than for maize or tobacco, though gross returns per hectare have been higher for groundnuts than for maize. The follow-up study of LLDP indicated further that farmers receiving groundnut packages have holdings 24 percent smaller than the mean. So the high default rate of groundnut loans may have been related to the fact that groundnut farmers are poorer than average. However, the detailed statisti- cal analysis also indicates that the farmers paying less than 100 percent of the credit were significantly better off than the average credit recipients. It is important to note, however, that the analysis also suggests that mean ag- ricultural income is significantly greater when a member of the family is working outside the country and sending remittances to the other members remaining behind to cultivate the land. So it is difficult to establish an ac- curate relationship between farm income and tendency to repay credit. The lack of clarity concerning this issue underscores again the need for more ap- propriate data to evaluate realistically the performance of program inter- ventions. Both credit and defaults were distributed quite unevenly among the unit areas in Lilongwe. Again, the considerable amount of data collected in LLDP are not adequate to explain what causes the geographical variation in defaults. However, the attitude of village headmen may be playing an im- portant role in the repayment performance and may partly account for the geographical variation. Land registration introduced under the project has been eroding their traditional power base."6 The village headmen, therefore, often perceive the project as a threat which, in their view, weakens their position. This may be setting up groups of individuals (the village commit- tee) that are willing to act in concert against the traditional authority. Kinsey points out that, when members of the project staff identify a village as good or bad, they have in mind the degree of cooperation obtained from 16. The chiefs have also had their traditional power altered. However, their new role as salaried institutionalized officials has enhanced their prestige. 96 DESIGN OF RURAL DEVELOPMENT the village headman. The attitude of the local leaders thus appears to have a substantial bearing on the effectiveness of LLDP's activities in a particular village. The Vihiga Maize Credit (Kenya) The experience of SRDP's Vihiga maize scheme emphasizes the impor- tance of identifying the precise constraint to adoption before instituting a credit program. The Vihiga Maize Credit Program is far less ambitious com- pared to the rural development programs in Ethiopia and Malawi. Of the sample of 600 farmers selected at random to participate in the Vihiga Maize Credit Program in 1971, only 54 qualified for credit under the Program's standards of creditworthiness. Adding another 22 eligible farmers from out- side resulted in 76 potential borrowers. Only 63 farmers finally utilized loans. The repayment in Vihiga was much lower than in the Ethiopian pro- grams or in LLDP. Only 80.7 percent of the credit was repaid. In 1972 the expanded program included 383 farmers. The evaluation of SRDP carried out by IDS contends that farmers who were qualified to receive credit were not necessarily the ones who needed credit, since a considerable subjective element entered the definition of creditworthiness. The poor repayment rate may be explained by the lower-than-expected profitability of the maize enterprise. The lower profitability seems to have resulted from the fact that credit was needed less for fertilizer, as stipulated in the program, and more for purchase of labor. Although there were rumors regarding isolated attempts to resell the material inputs made available through a credit program, the extension agents' visits to the farms appear to have been useful in ensuring the use of the inputs for the intended purpose. However, out of the twenty- two farmers who did not reapply for loans in 1972, ten indicated that an im- portant reason was their inability to use the loan to hire labor for land prep- aration. Implications of Past Credit Components for Design of Future Programs Whatever the specifics of individual programs, the reviews illustrate with a remarkable consistency that the credit procedures adopted in the pro- grams are tedious, cumbersome, and highly demanding of the very limited trained manpower. Once the crop of the thoroughly screened, early credit takers is harvested and as the number of credit applications increases, the repayment rate slides and the costs of distribution and recovery rise. There is a genuine question as to whether and how the less eager adopters, who may be less able to bear the risk or may be less creditworthy, will be handled by the current approaches to credit distribution. The creditworthi- ness problem may be compounded if project innovations cause landlords to increase rents, as in CADU. To assure repayments through individual litigation or by goading creditors is expensive, cumbersome, time-consum- ing, and, judged by the experience in CADU and LLDP, by no means helpful in fostering amicable relations between the project authorities and the rural people. The long-term financial and administrative viability of AGRICULTURAL CRE-DIT 97 such an approach to credit distribution is even more questionable if the credit is really to reach all segments of the rural population. Some projects, such as the Salima project financed by the West German government in Malawii, consciously attempt to foster an element of private enterprise through a complex ladder rating, which rewards better managers and has a credit incentive system. However, many fear that, by neglecting to tackle the apathy of the nonparticipants, the Salima project is accentuat- ing the differences in the economic status between participants and nonpar- ticipants. What explains the poor repayment of credit and how could it be allevi- ated? As illustrated above, a number of hypotheses have been put forward in the project reviews about the repayment rate. For example: 1. Poor extension explains low repayment. One of CADU's evaluation studies shows a high positive correlation between an independent assessment of the quality of extension supervision and the repayment rates of credit in the same area. 2. Inadequate yields from new technological packages cause low repay- ments, as in the case of groundnuts in Lilongwe or maize in Vihiga. 3. Low harvest prices result in low profitability and, hence, low repay- ment. Such is the case in CADU. 4. Repayments are inversely related to social status, as in CADU and perhaps in LLDI'. 5. Overly ambitious expansion of credit facilities with inadequate follow- up by the government or project authority responsible for collection leads to low repayment. This appears to have occurred with AFC and to some extent in CADU and LLDP. Most projects, however, provide little data to permit a more systematic analysis than presented above. It is critical that these various hypotheses be tested so as to be able to ameliorate deficiencies which may cause a reluc- tance to borrow or to repay. On the basis of available information, it cannot even be stated how cru- cial the supply of credit is for adoption of innovations. On the one hand, the project reviews indicate that the existing potential for savings in the rural sector is not being exploited. On the other hand, LLDP's rate of credit ex- pansion raises a question as to whether dramatic short-run increases in maize production in the Lilongwe district would have been possible without an emphasis on credit distribution. Few programs seem to want to take the risk to answer this question. Nevertheless, it is apparent that traditional, individually oriented institu- tional credit distribution is not efficient in reaching large numbers. As in the case of delivery of extension, new group approaches may be an econom- ic necessity in organizing credit systems. Wherever social systems will per- mit, group responsibility and group benefit will have to be emphasized (a) to reduce cost of credit distribution, (b) to assure repayments of credit, and (c) to mobilize new savings into the credit system. This may require a far greater encouragement of spontaneous group in- put purchases of the type that have been occurring in Lilongwe. However, even if credit is not a constraint to adoption of innovations, as demon- strated by the cash purchases in Lilongwe, there may still be a reluctance among farmers to take up new practices due to two additional obstacles: (a) 98 DESIGN OF RURAL DEVELOPMENT if the innovation does not seem sufficiently profitable in relation to its costs, or (b) if the risk of adoption seems high. Although introduction of in- stitutional credit is frequently justified on the basis of a credit constraint, in practice credit may be more effective in alleviating these latter two obsta- cles to adoption. The low interest rates usually associated with institutional credit reduce the cost of adoption. And since credit is frequently provided under the same program that promotes innovations, the farmers seem to get an impression that, if the adoption fails, the credit may not have to be repaid. Such an impression, although unconscious, seems to reduce the risk of adoption. These characteristics of institutional credit raise a question as to whether, in the intial stages of development, input subsidies may not increase rela- tive profitability more easily than credit. During this early period when the number of adopters is low, the financial cost of subsidies is not very high. Of course, to make subsidies less demanding administratively, it is neces- sary that there be a far greater effort to exploit the traditional channels for input marketing. In most rural development programs reviewed, credit is given in kind; and credit programs only reinforce the need for formally ad- ministered input distribution systems. Although input subsidies are attractive due to administrative ease, it is frequently argued that, because of political pressures, once instituted they may be difficult to remove. This problem would appear to be no more serious in the case of input subsidies than in the case of the subsidized cred- it provided in most programs. The other argument against input subsidies, however, is that, if subsidies are withdrawn in the later stages of develop- ment when innovations have become popular, the risk-averting, slow adop- ters may be bypassed by the benefit of subsidies. The late adopters may be even greater losers, since the prices of their produce may decline as a result of adoption of new innovations. There is, therefore, a question as to whether a combination of general subsidies and specific credit programs directed towards low-income groups may not be used to minimize the use of scarce administrative resources and to ensure the distribution of benefits to the lower income groups. However, such important questions as to how to use alternate means to reach the lowest income groups cannot be answered satisfactorily, since so few programs have deviated from the stan- dard credit approach. More attention also needs to be paid to institutionalization of rural sav- ings through traditional organizations. Evidence discussed earlier indicates that there probably exists a considerable potential for savings in the rural sector which remains untapped. A survey carried out by LLDP's Evalua- tion Unit indicates that in the LLDP area people bury considerable amounts of savings under mattresses and in tin cans. The mobile savings banks being tried by the project management have not yet met with much success, apparently because the people do not trust banks which go away with their savings. Certainly some ways must and can be found to mobilize these savings and to use the local energies to carry out the mobilization. At the same time, the sense of responsibility and participation in the rural people can be increased through their understanding of the benefits of sharing a bigger pie over time. Involving rural people in saving as well as in receiving credit will lead to more direct interest in assuring repayments. An AGRICULTURAL CREDIT 99 alternative is to introduce matching credit grants in proportion to the amount of resources mobilized and effectively utilized from within the rural areas. This approach to rural credit implies that limited, formal admin- istrative manpower would be best utilized initially for institutionalizing ways of mobilizing and distributing rural credit rather than for the rapid ex- pansion of an institutionalized credit program. The approach suggested here is very similar to the suggestion made by Belloncle and Gentil for fostering group involvement in extension (see Chapter IV). As with extension, a group approach to credit distribution requires considerable oral exchange with the rural people. They must be able to identify for themselves their own need for credit. They should be involved in administering credit and should understand the importance of repayments to keep the pie growing. This approach is different from using local committees to distribute ex- ternally injected credit and to bring pressure on the rural people to repay. Too often such commtittees are not delegated genuine responsibility at the local level. As Belloncle and Gentil have noted, planners often feel that the peasant is incapable of adopting recommended changes on his own initia- tive and look on adult peasants as only big children who need constant watching. The dangers of such paternalism are manifest. For example, farmers' associations set up under the Local Progress Enterprise (EPL) in Cameroon are reduced to requesting financial assistance from EPL in order to carry out projects that it has elaborated. This assistance is granted only insofar as EPL considers it justified. Associations have been forbidden by EPL to procure their own resources to finance their own investments. The inevitable result, in the opinion of Belloncle and Gentil, is a deep sense of frustration which may be transformed into aggressiveness against EPL and may lead to a belief that the enterprise is out to rob the farmers. LLDP's experience in introducing group credit illustrates the same point. LLDP attempted to rnake the former village chiefs responsible for credit distribution for the village as a whole. This did not prove to be very effec- tive. Because the chiefs were illiterate, they were poor administrators of credit. Using them led to an increase in administrative costs, as all records had to be maintained in duplicate. However, according to the former credit manager at LLDP, the lack of effectiveness of the village chiefs as credit ad- ministrators was a result of their eroding authority, caused by the land set- tlement program. There is probably an element of validity in both these arguments. Clearly, ways have to be devised, on the one hand, to impart basic accounting and writing skills and, on the other, to exploit the changing leadership structure in a village. Both these factors mean that it will take a lot of time and effort to build grass root institutions, a task which is neither easy nor without risks of failure. It cannot be overemphasized, however, that at initial stages project ex- pansion must take place at a slower pace than has been the case in the proj- ects reviewed. Thus, considerable attention can be devoted to understand- ing the rural social structure and the role that it can play and gradually to delegating genuine responsibility of administration to local organizations. Such a participatory, as distinct from a paternalistic, approach to develop- ment seems critical for the long-term viability of rural development pro- grams beyond the stage of donor involvement. CHAPTER VI Marketing of Agricultural Output Inadequate marketing and transport facilities, fragmented markets, and uneconomic and unpredictable prices are generally considered to be some of the major bottlenecks to rural development. Consequently, in assessing the essential minimum components of a rural development package, mar- keting attracts a great deal of attention.' The need for and the exact nature of a marketing component are, however, very difficult to assess. Marketing involves a complex and, at times, an obscure interaction of traditional and modern institutions with broader economic policy questions, such as the determination of relative prices among crops and between sectors. Because of the fragmented nature of rural markets, the tendency in most programs is to establish elaborate formal marketing structures. Such an approach to marketing is not only demanding of the limited administrative capacity but also overlooks the potential to utilize the informal market channels already prevalent in many rural areas. There appears to be a wide scope for adopting a pluralistic approach in the development of marketing institutions so as to relieve the staffing/administrative constraint faced by many programs and to increase marketing efficiency through competition between formal and informal channels. Frequently, however, the potential benefits of a pluralistic approach are neglected; and policies are instituted which, in fact, undermine the building of viable local market systems. This chapter draws on the evidence related to output marketing to sup- port these arguments. There may also be considerable scope for a pluralistic approach to the marketing of agricultural inputs, once demand is created for such inputs through promotional programs and once standardization is in- troduced in the servicing of inputs through regulatory means related to quality control and weights and measures. Unfortunately, this avenue re- mains unexplored, as most projects rely almost completely on organized in- put channels. Promotion of informal market channels may also generate greater growth linkages within the rural economy, since the demand for a wide range of goods and services from farm tools to radios is substantial in many project areas once incomes increase. In the initial stages of implementation, the projects have, by necessity, focused on marketing of a limited range of 1. For instance, this concern was strongly present in the preparation of the Minimum Package Program financed by IDA in 1973. It is reflected in the reports of the members of the preappraisal missions and in the correspondence between the World Bank's Washington office and the Regional Mission in East Africa regarding the various marketing components proposed in the Minimum Package Program. MARKETING OF AGRICULTURAL OUTPUT 101 agricultural commodities. However, frequently the traditional markets have already begun to respond to the rural demand.2 Encouraging the growth of such informal markets through provision of infrastructure, train- ing programs, and small-merchant loans can stimulate investment in ser- - vice activities and can be a major source of employment generation. To illustrate these points, the discussion in this chapter is focused on three questions: 1. What considerations have entered into the design of marketing com- ponents? 2. How effective have the marketing components in the past rural devel- opment programs been in meeting the objectives that prompt them? 3. What are the lessons of the past experience for designing marketing components in future rural development programs? Factors Influencing Marketing Components The reviews show that the substantial marketing component in package programs is usually prompted by a concern about price incentives. As the Ethiopian Minimum Package Program illustrates, production increases resulting from the introduction of new technologies may lead to a pre- cipitous drop in harvest prices. The problem of price decline is particularly severe when production surpluses are concentrated in a small region and where markets are fragmented by lack of transport, communication, and storage facilities. The reviews of CADU and LLDP show how the market- ing components in these two programs evolved from a concern for assuring farmers a guaranteed rninimum price and for reducing seasonal year-to-year rice fluctuations so as to provide farmers an incentive to use new inputs. The argument for introducing a marketing component is often supported by a widespread belief that traditional trade channels are inefficient, ex- ploitative, and generally antisocial. According to this view, the substantial regional and seasonal disparities in agricultural prices are partially ex- plained by the exploitative nature of traditional trade.3 Further arguments 2. In the LLDP area, for instance, a number of shops have emerged, established by resi- dents who have returned with savings after serving in mines in neighboring countries. These shops carry a range of consumer goods, including shoes, hurricane lamps, locks, tinned food, umbrellas, and ready-made clothing. 3. To the contrary, substantial literature has amassed which confirms the potential for com- petitiveness and efficiency in the traditional trading channels. Much of this evidence relates to West African countries. See, for example, Q.B.O. Anthonio, "The Marketing of Staple Foodstuffs in Nigeria: A Study of Pricing Efficiency," Ph.D. dissertation, University of Lon- don, 1968; P.T. Bauer, West African Trade: A Study of Competition, Oligopoly and Monopoly in a ChangingEconomy, rev. ed. (London: Routledge and Kegan Paul, 1963); Polly Hill, "Markets in Africa," in Journal of Modern African Studies, vol. 1, no. I (December 1963), pp. 441-54; H.C. Kreisal, Cocoa Marketing in Nigeria, Consortium for the Study of Nigerian Rural Develop- ment Paper no. 21 (East Lansing: Michigan State University Press, May 1969). For a state- ment of the issues related to marketing, see Uma J. Lele, "The Role of Credit and Marketing in Agricultural Development," in Agricultural Policy in Developing Countries, ed. Nural Islam (London: Macmillan, 1974), pp. 413-41. Although the evidence is far more scanty, a similar potential for competitiveness has been observed in some of the East African countries. See the Kenyan case in William 0. Jones, Marketing Staple Foodstuffs in Tropical Africa (Ithaca, N.Y.: Cornell University Press, 1972). Thoday's study makes similar observations in Ethiopia. See A.R. Thoday, Marketing of Grains and Pulses in Ethiopia (Stanford, Calif.: Stanford Re- search Institute, April 1969). On the other hand, the limitations of the alternatives frequently followed are presented forthrightly in government of Tanzania, Report of the Presidential Special Committee of Enquiry into the Cooperative Movement and Marketing Boards (Dar es Salaam: Government Printers Office, 1966). 102 DESIGN OF RURAL DEVELOPMENT in support of a marketing component relate to the ease of production credit collection. According to this view, cultivators may sell their produce to tra- ders at low prices and, thus, will be unable to repay credit unless marketing is combined with credit. Gross return to the farmer is dependent on the cost of inputs, the physi- cal return to inputs, and the output price. As was documented earlier in the discussion of extension, in many projects reviewed adequate attention has not been paid to the farm-level constraints which affect the farmer's yield. Yields are often affected by lack of availability of inputs at the right time, in the right place, in the right form.4 A considerable portion of the yield differences may still be explained by vagaries of weather. As long as yields vary substantially from year to year, a price stabilization program may lead to substantial year-to-year fluctuations in monetary incomes. The fluctua- tions in monetary incomes may be even more substantial in the case of small farmers, since the elasticity of marketing with respect to changes in production is likely to be greater in their case. However, in the absence of concrete empirical evidence on these questions, the extent to which in- comes fluctuate with or without price stabilization programs and whether farmers respond to fluctuating incomes rather than to fluctuating prices is largely a matter of conjecture.5 Preliminary evidence from the review of LLDP and the Ethiopian projects suggests, however, that it may well be the uncertainty about yields and the consequent risk of not maintaining a mini- mum level of subsistence production, combined with an unwillingness to incur unpayable debts on new inputs, which dissuade small farmers and tenants from adopting new technology (see Chapters II, 111, and V). Thus, frequently the logic of the price-support program is, at best, based on shaky grounds. However, even if the efficacy of support prices is accepted, an im- portant question is whether they are administratively feasible, given the complexity of the task, the limited administrative capability in most African countries, and the alternatives that frequently exist. To answer this question, the relative benefits and costs of the various price-support pro- grams in the projects reviewed are examined below. Effectiveness of Marketing Components in the Development Projects A variety of marketing approaches has been used in the various projects reviewed. These various approaches have, however, consistently failed to show potential for a viable low-cost approach that will also provide an in- centive for increased agricultural production. As will be evident from the experience reviewed below, the causes are as numerous as the various ap- proaches tried. 4. In interviews the management of the German-financed Saiima project in Malawi indi- cated that the project's effort in improving the Agricultural Development and Marketing Cor- poration's administrative capacity to distribute inputs has centered around ensuring that AD- MARC can estimate and distribute on time different types of fertilizers in the combinations in which they are recommended by the extension service. 5. For a discussion of the elasticity of consumption and marketing with respect to change in production, see John W. Mellor, "Agricultural Price Policy and Income Distribution," un- published paper prepared for the World Bank, December 1974. MARKETING OF AGRICULTURAL OUTPUT 103 Formally Administered Marketing Components One of the importantL factors which explains the limitations of the mar- ket interventions is that the national grain boards or their equivalents are often not interested in handling food crops. Most marketing boards have traditionally handled the export crops of commercial farmers. Commercial crops are easier to procure; their areas of production are usually endowed with relatively good marketing infrastructure of roads, storage, and process- ing facilities. Most marketing boards also lack the administrative capacity and the manpower required to purchase small, scattered supplies of food crops. Further, the marketing boards are frequently not interested in dis- tributing procured food supplies much beyond the capital city.6 Table 6.1: Grain Purchases by the Marketing Division of CADU, Ethiopia, and Additional Benefits Received by Farmers, 1967/68-1971/72 Wheat Barley Flax Additional Additional Additional Farming Quantity benefitsa Quantity benefits" Quantity benefitsa season (quintals) (Eth$) (quintals) (Eth$) (quintals) (Eth$) 1967/68 480 1,440 _b _ b _ b _ b 1968/69 2,300 6,900 _b _ b 710 2,840 1969/70 6,310 18,940 _b _b 800 3,190 1970/71 23,980 71,940 520 1,560 6,100 24,400 1971/72c 93,890 281,660 12,150 36,450 3,330 13,310 a. Additional benefit is the difference between the net price paid by CADU and local market price. Figures have been rounded. b. No purchases made. c. During the 1971/72 season CADU also bought 1,070 quintals of peas, 900 quintals of rapeseed, and 1,900 quintals of maize. Source: Government of Ethiopia, Chilalo Agricultural Development U nit, Cost/Benefit Analysis of CADU .for the Period 1967/68-1973/74 (Asella: CADU, 1971), pp. 4-5. In Ethiopia CADU and WADU undertook the purchase of major crops grown in their respective program areas at specified prices. Tables 6.1 and 6.2 show the quantities of various crops handled by the marketing divisions of CADU and WADU. As can be seen from Table 6.1, CADU's marketing activities have centered mainly around wheat. However, between 1966 and 1971 the area under wheat in CADU increased from 23,700 hectares to 51,000 hectares; total output jumped from 32,000 metric tons to about 102,000 metric tons during the same period.7 Nevertheless, in Chilalo hec- tarage under barley is much larger than that under wheat.' 6. Kinsey points out, for instance, that ADMARC has requested the Ministry of Agricul- ture in Malawi to encourage maize production in deficit areas in an attempt to reduce its own transportation costs. 7. G.Bergman and H.Lindqvist,CreditSituationinChilaloAwraja,CADU Minor Research Task no. 3 (Asella, Eth.: Chilalo Agricultural Development Unit, July 1969); and Chilalo Agri- cultural Development Unit crop sampling surveys. More recent figures indicate that the area in wheat was 27,000 hectares in the 1967/68 period, 47,000 hectares in the 1971/72 period, and 150,000 hectares in the 1972/73 period. See D. Yirgou et. al., Final Report on the Appraisal of CADU and EPID (Addis Ababa: Government of Ethiopia and the Swedish International De- velopment Authority, May 1974), p. 11. 8. This may be because barley has traditionally been used for home consumption while wheat is mostly marketed. CADU could easily sell wheat to big flour mills, whereas most of the other crops have to be sold in small quantities in retail markets, which CADU found to be very costly. Table 6.2: Commodities Handled by the Marketing, Credit and Cooperatives Division of WADU, Ethiopia, 1971/72 Purchasesa Sales Stocks Value Total Value Total per metric ton value per metric ton value Product Metric tons (Eth$)P (Eth$) Metric tons (Eth$)b (Eth$) Metric tons Coffee cherries 215.5 198.5 42,776.8 215.5 264.3 56,965.6 -' Cotton 146.1 730.0 106,653.0 146.1 746.3 109,028.1 _c Ginger 579.8d 146.0 84,650.8 _ e _ e _ e 131.5f Maize 134.8 120.0 16,176.0 100.2 255.8 25,629.8 34.6 Chilies 29.4 500.0 14,700.0 29.4 680.5 20,005.1 -c Teff 252.3 190.0 47,937.0 17.6 355.5 6,256.0 234.7 Wheat 479.0 200.0 95,800.0 13.4 362.4 4,851.4 465.6 Hides and skins 9.6 832.0 7,987.2 9.6 933.8 8,964.7 _C Tobacco 19.5 42.2 822.9 19.5 60.3 1,175.5 _c Total 1,866.0 417,503.7 551.3 232,876.2 866.4 a. First payments. b. Average figures, not necessarily reflective of total value column. c. No sales handled. d. Wet weight. e. No stocks handled. f. Dry weight. Note: According to Tecle, the reliability of some of the information given in the table is questionable. Sale price of wheat, teff, and maize seems to be too high. Source: Government of Ethiopia, Wolamo Agricultural Development Unit, Annual Repor thr Year 2 Programs(Soddo: Ministry of Agriculture, May 1972) and files of the program's Marketing, Credit, and Cooperative Division. a S * S~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ MARKETING OF ACGRICULTURAL OUTPUT 105 In the Lilongwe Land Development Program of Malawi, the marketing of inputs and agricultural output has been entrusted to the Agricultural De- velopment and Ma]rketing Corporation. ADMARC, by virtue of its monopoly position as a purchaser and exporter, is responsible for formulat- ing and implementing Malawi's price policies with respect to all the major agricultural commodities. However, in practice LLDP's program manage- ment has played an important role in the purchase of maize in the program area. For example, as a way of promoting the sale of maize through pro- gram channels, LLDP has introduced a maize-shelling program. According to Kinsey, the LLDP mobile maize-shelling program has, for the moment, solved the transportation problem for many growers. The maize shellers are pulled from village to village by a team of hired oxen; and, after the maize has been shelled, a private lorry hired by LLDP comes to the village to collect it. All that remains for the farmer to do is to go to the ADMARC market to pick up his payment. ADMARC has strongly resisted the maize- shelling program and has even refused to allow LLDP to use ADMARC cashiers at the unit markets to make payments for maize purchased through the shelling program. Consequently, LLDP has had to establish mobile teams that circulate among the units, paying out for maize purchases. Despite such efforts on the part of LLDP's program managers, it is difficult to state how important formal marketing channels have been in handling the total marketed surplus of maize in the program area. No infor- mation is available on the quantities handled by informal channels. Their role must, therefore, be surmised indirectly through estimates of produc- tion increases, domestic consumption, and so forth. As noted in Chapter III, detailed reappraisal of the program shows that estimates of maize pro- duction in the program area vary substantially, depending on which of the existing estimates of size of holding, area under maize, and yields per hec- tare are accepted. The estimate of marketed surplus handled by the program as a proportion of production (or of the total marketed surplus) depends on which one of these various estimates of production is accepted. Existing data, therefore, cannot resolve the question of whether to accept a substan- tial success in achieving an increase in the production of maize (with only a moderate success in purchasing the marketable surpluses of maize through the program's marketing channels) or, alternatively, to accept a consider- able success in acquiring a major portion of the increased production through organized channels (with only a limited success on the production front).9 Whatever its relative importance in the total marketed surplus, the con- siderable initiative on the part of the program management in purchasing maize seems to be accompanied by a reluctance on the part of ADMARC to involve itself in maize purchases in the program area, at least for the time being. 9. ADMARC handles a very small portion of the marketed surplus of maize, although one of ADMARC's stated goals is to assure that the internal demand for maize in Malawi is satisfied and that there is an annual carry-over of some 7 percent of a normal season's produc- tion as a safety margin. For example, in the 1971/72 period only 7,710 metric tons of maize were sold to ADMARC in the project area, though the LLDP Evaluation Section estimated the marketable surplus to be 19,090 metric tons. According to Kinsey, the amount of maize marketed through ADMARC in any given season represents only the tip of the iceberg (about 4 percent of estimated production in the 1971/72 period) and as little as a 10 percent drop in production results in the complete curtailment of sales to ADMARC. 106 DESIGN OF RURAL DEVELOPMENT With respect to LLDP, ADMARC has maintained a relatively low profile in maize-marketing activities. ADMARC's management would prefer to come in on a large scale only when maize becomes a commercial crop. In the opinion of ADMARC's director, LLDP would not supply maize in commercial quantities at least until 1976. On the other hand, ac- cording to LLDP marketing staff, the price of maize cannot be reduced without a drastic drop in production and a shift by farmers away from maize growing into tobacco and groundnuts. The current national purchase price for maize in Malawi is well below the prices established for maize elsewhere in Central Africa.10 The price policy towards maize has been a major bone of contention be- tween the program management and ADMARC. LLDP's experience in this regard illustrates the complexities and difficulties involved in imple- menting a price policy when the individual program's objectives are in con- flict with those of the marketing board. For example, until the 1972/73 period the guaranteed farm-level price of maize was established by ADMARC at K2.50 per bag. Through its maize-shelling program, however, LLDP was able to pay farmers K2.70 per bag in the 1971/72 period while receiving K3.30 when the maize was delivered to ADMARC. There has been considerable dissatisfaction, both within the Ministry of Agriculture and ADMARC, over the differential price structure. In general, ADMARC's objections appear to stem from the fact that farmers outside LLDP have become aware of the LLDP price and have been insisting upon being paid this price in other areas. ADMARC, of course, will pay only the stated price of K2.50. On occasion, the policies of ADMARC and LLDP are in diametric opposition. The best example of this occurs in a crop season when maize export prices are at relatively low levels. ADMARC, not wishing to buy and store large quantities of maize above the target surplus, would like to close the markets when it has ac- quired enough maize. LLDP management, on the other hand, is particularly concerned about credit repayment rates and would like to see the markets remain open so as to improve sales. Both sets of objectives are valid, but the conflict between them is very real. To fulfill its other objectives, once a marketing component is introduced in a project, the project management often ends up providing a marketing service at some guaranteed price. The problems of the project authority may, however, only begin with such an intervention. First, there is the problem of providing a low-cost service. With intensive marketing facilities even the most efficient organized marketing schemes involve much higher costs of operations than do the traditional alterna- tives. According to data collected by Tecle, in CADU the unit costs (exclud- ing overheads to the project management) of marketing wheat between Asella and Addis Ababa amounted to more than twice those of the private traders. The program's relatively higher handling costs raise serious ques- tions as to how the Marketing, Credit, and Cooperative Division will be able to function as a viable autonomous unit in the future. Starting from the 1973/74 season, the MCC Division was expected to cover all of its market- ing costs frorn the deductions that have to be made before making second 10. The 1971/72 pre-planting price in Zambia, for example, was K4.30 per 90-kilogram bag, compared to K2.70 per bag in Malawi. MARKETING OF AGRICULTURAL OUTPUT 107 payments to farmers. So far, salaries of the MCC Division staff and a few other costs such as depreciation of equipment have not been charged to farmers as marketing costs. LLDP's maize-shelling program suffers from similar problems of high costs of operation. Second, there are numerous problems related to the determination of the levels of guaranteed minimum prices in the context of an overall price policy and the implementation of the minimum price. The experiences of CADU and WADU cited below show that, given the fragmented nature of the national market and the extreme unpredictability with which national grain boards often export, import, and distribute grain, even the best mar- keting experts can do little more than resort to their astrological abilities in predicting prices and determining the project's price policy. The management may become a financial loser at this highly speculative game, as illustrated by the losses incurred by CADU in the 1970/71 and 1971/72 periods. For the 1972/73 season, it was decided to let farmers bear the marketing risks involved. A two-payment system with 90 percent of the prevailing local market price as the first payment was proposed. However, as local traders expressed their willingness to buy all the marketed produce at prevailing local market prices, farmers refused to sell their grains to CADU at 90 percent of the market price. This forced CADU to reverse its policy and buy farmers' grains at prevailing local prices. The decline in the number of borrowers in Chilalo in the 1972/73 period compared to the 1971/72 period has been explained by the project management as a conse- quence of the farmers' dissatisfaction with the wheat price, which, in its view, was exploited by the local trading-interest groups." Yet another variant of the marketing problem faced by the project au- thorities occurred in LLDP. It arose because ADMARC has been reluctant to pass gains from rising world maize prices on to farmers. The export price of maize in 1972 was K6.60 per bag, compared with the price of K2.50 paid by ADMARC to farmers. There has been a similar problem in the pricing policies in Mali, as indicated earlier in Chapter 111. LLDP management pointed out during interviews that farmers become discontented when they realize that the prices received by them in the project markets are lower than those offered elsewhere by private traders. In the 1972 harvest season, this price differential resulted in farmers throwing stones at some of the LLDP marketing centers. Under these circumstances, the project authority may either tell the farmer that it cannot pay as high a price as offered elsewhere or suggest that the farmer sell his produce elsewhere. WADU encounters problems similar to those faced in CADU. It handles too insignificant a proportion of the products marketed in Wolamo to be able to influence the marketing system (see Table 6.2). Since Wolamo is a net importer of grain (teff, barley, wheat, maize), WADU prefers not to buy these products. Farmers have continually been pressuring WADU to buy their grain, believing that WADU is able to offer them a higher price than I1. According to Cohen. the decline in the number of borrowers may also be explained by the fact that (a) many tenants, who were previous borrowers, have been evicted from their holdings: (b) many farmers have acquired their own seed from previous purchases and felt that they no longer needed CADU's credit to purchase CADU's seed; (c) some farmers purchased seed sold with a CADU label at the local market; and (d) in the view of the farm- ers, the price of CADU seed seemed high relative to the local price for wheat. See Cohen, "Rural Change in Ethiopia," pp. 595-601. 108 DESIGN OF RURAL DEVELOPMENT the local traders. WADU stands to lose if it buys grain, as its handling costs are much higher than those of the local traders. Nevertheless, it has decided to buy some grain in order to preserve its good relationship with the farm- ers. WADU has also encountered problems in implementing an ad hoc price policy, although it has allowed farmers to participate in the risk in- volved in unexpected price declines. WADU's policy has been to pay farm- ers 60 percent of the anticipated sale price of any product as a first payment and, when the product is sold, to make a second payment after deducting an amount equal to 20 percent of the first payment to cover the marketing costs. A second payment is made only if and when the realized sale price more than covers the first payment plus the marketing costs. The main obstacle to this policy seems to be WADU's inability to forecast product prices with sufficient certainty to ensure second payments. For example, WADU was unable to make second payments for ginger, wheat, teff, and maize for the 1971/72 season because large amounts of these crops were in stock as late as October 1973. Failure in making second payments generates mistrust among farmers, as indicated by interviews with some farmers, because WADU's first payments, particularly those for grains, are usually smaller than the prices offered by local merchants. How effective is a price-support program when only small proportions of the marketable surpluses are handled by the project authorities? As they have in CADU, farmers usually develop an expectation of receiving from the project authority not only a guaranteed minimum price but also a share in the profit. Ironically, in the 1972 season farmers in the CADU area ac- cused management of making a profit at their cost. CADU's management observed that the subsequent drop in credit applications may have been partly due to the farmers' expressed desire to boycott credit the following year, though tenant eviction was likely to have been an equally important factor in explaining the decline in the number of borrowers. How useful is the marketing component for assuring high credit repay- ments? The reviews of CADU, WADU, and LLDP show that the provision of collection centers certainly aided collection of credit in the first few years when the number of credit takers was small and credit administration was relatively easy. However, those farmers who do not want to repay credit find many ingenious ways of avoiding this responsibility. In LLDP, for ex- ample, farmers are required to reveal their credit record when selling their produce at the project market. Those owing money are required to pay off their debts immediately with the cash gained from the sales. Many farmers get around this requirement by borrowing the credit records of persons whose debts are paid up. Another frequently used route is to market pro- duce through private channels. The performance of the marketing systems in the package programs raises doubts as to their viability in the long run. What is disturbing about the marketing systems reviewed is their inability to get at the more basic factors-e.g., if the marketing boards are not interested in purchasing the food crops produced by small farmers, what is the guarantee that they will do so after the project authority phases out?12 Would it not be easier to im- 12. Kinsey reports that ADMARC officials are quite confident that they could cope with any amount of maize that might be marketed. However, such confidence might well be tem- pered by the fact that storage and transportation facilities at the Lilongwe depot already ap- pear to be operating close to capacity. Certainly a significant increase in marketed maize over two consecutive good seasons would severely strain local facilities. MARKETING OF AGRICULTURAL OUTPUT 109 prove the traditional marketing organization that obviously already handles a substantial portion cof the marketed surplus, even in such program areas as CADU, WADU, and LLDP? In any case, if the long-term viability of the project is a consideration, the question remains whether it makes sense to expand distribution of credit rapidly without developing a genuine sense of responsibility and farmer participation in the project activities and without building a more Viable marketing and price policy system. Performance of Cooperatives As an alternative--or a complementary approach-to other marketing efforts, a number of the programs reviewed have established cooperative societies. However, an effective cooperative development requires organi- zation, leadership, and entrepreneurial capability. Even if this is recognized, as indeed it is by many of the program managers interviewed, developing such a capability is a difficult and long-term task. Moreover, an examination of the steps taken to introduce cooperatives causes one to wonder whether they will be successful, even in the long run. The cooperative train-ing courses usually consist of principles of coopera- tion, borrowed from textbooks not dissimilar to those used in nineteenth century Europe. Courses also include training in accounting. However, they fail to cover the trading and speculative aspects of marketing food crops and the entrepreneurial skills required to deal with these problems. An evalua- tion of CADU's experience in cooperative promotion observed that the graduate is handicapped when he goes to the field, due to the fact that the training so acquired is not job oriented. Nor does it cover essential courses in cooperative organization and management. The lack of practical training can become a source of frustration and embarrassment to the trainees.'3 The need for trading skills is by far the most important difference between the export crop cooperatives and the cooperatives being formed for marketing food crops. The export crop cooperatives have performed the relatively straightfor- ward administrative function of procuring for further processing or export standard, scientifically graded crops at prices usually fixed by the marketing boards. Collinson attributes the success of the Sukuma cooperative move- ment for cotton in Tanzania-apart from its unique marriage with the broader political movement-to the relative simplicity of the task. For in- stance: 1. The cooperative movement inherited a history of government inter- vention in the rnarketing of the crop. From 1949 onwards, there were no alternative channels for selling the crop overseas. The government was the sole buyer. 2. It dealt with a crop with no domestic use. All the production was ex- ported. The board was never forced into the position of being a buyer 13. Government of Ethiopia, Chilalo Agricultural Development Unit, "An Appraisal of CADU's Experience in Cooperative Promotion," unpublished report, July 1972. According to Cohen, to date CADU's involvement in cooperatives is rather limited, perhaps as few as 1,000 farmers belonged to such groups in 1974. However, Nekby, the first manager of CADU, has indicated in personal discussions that the CADU cooperative movement has about 10,000 members and is very active. Conflicting data of this nature highlight the need for institu- tionalizing systems for maintenance of regular and reliable data on numbers as well as on the degree of involvement to be able to assess more objectively the progress of institutional devel- opmen t. 110 DESIGN OF RURAL DEVELOPMENT of last resort with its inherent uncertainties, nor did it have to face il- legal local and inter-farm sales; so, vis-a-vis the local producers, the board was the sole consumer. 3. The board was formed at a period when a booming world market allowed it to consolidate its position with no financial problems, creat- ing a security never threatened until perhaps the near exhaustion of its Price Assistance Fund in 1966. 4. The Lint and Seed Marketing Board (LSMB) took the trading role out of the cooperative movement in cotton. It removed the commercial sting which has often been the downfall of cooperatives operating in competition with private traders. It left the cooperative as purely a col- lection channel for the crop. If the grain cooperatives are to survive, unlike the export crop coopera- tives, they have to be able to compete with the private trading channels in their ability to speculate and to operate at low costs in highly fragmented domestic food markets.14 In the case of ZAPI in Cameroon, trading prob- lems are particularly difficult for inexperienced cooperatives to handle when the latter are confronted with illiteracy, poor transport and com- munication facilities, and the virtual absence of financial institutions in many rural areas. Furthermore, there is some evidence in the reviews to in- dicate that fraud is an extensive problem. Costs of cooperative marketing tend to be high, even in Tanzania, which has had a relatively well-devel- oped cooperative network.15 Another problem is the tendency to develop a paternalistic rather than participatory approach to cooperative develop- ment. The reviews indicate with remarkable consistency how this has ad- versely affected the rural people's sense of involvement in project activities and their willingness to be involved financially and organizationally to make project-initiated institutions viable. For instance, Belloncle and Gentil observed that the ZAPI project in Cameroon was seen by the farmers as a series of operations decided upon far from them in the capital and perhaps even in France, set up by foreign- ers and which, undoubtedly, would last only a very short time as in the case of the neighboring projects. Abraham, et al., who carried out an indepen- dent evaluation of WADU's Phase 1, attribute the poor performance of the WADU Group Marketing Organization (WGMO) to its emphasis on work- ing for the farmers rather than with them. During interviews the team had with farmers in Wolamo, both members and nonmembers believed that WGMO was really WADU; and the idea of joint conduct of the organiza- tion's program seemed foreign and, perhaps, suspect to them.'6 The review 14. As pointed out earlier, although some of the East African countries have a less devel- oped private trading system compared to several other countries in West Africa, there is a considerable informal exchange of produce in rural areas. 15. For problems of cooperatives in Tanzania, see Elliot J. Berg, "Socialist Ideology and Marketing Policy in Africa," in Markets and Marketing in Developing Economics, eds. R. Moyer and S. Hollander (Homewood, Ill.: Richard D. Irwin, 1968), pp. 24-47; Stanley Dryden, Local Administration in Tanzania (Nairobi: East African Publishing House, 1968); and government of Tanzania, Report of the Presidential Special Committee of Enquiry into Cooperative Movement and Marketing Boards. 16. In personal discussions Abraham, the first Ethiopian manager of CADU, indicated that one of the lessons of CADU is the need to pay far greater attention to farmers' involvement in the project activities from the outset so as to reduce their sense of dependence on project activities. However, given the social and political realities in Chilalo, increased farmer in- volvement was difficult to institute. MARKETING OF AGRICULTURAL OUTPUT 111 of SODENKAM in Cameroon makes a similar point, as do evaluation stud- ies conducted in CAI)U. However, it must be emphasized that this pater- nalistic attitude is by no means confined to projects financed by donors and designed and implemented by expatriates. It seems to apply equally well to the ujamaa movement in Tanzania and may, in fact, explain some of the problems encountered in its implementation (see Chapter IX below). To some extent paternalism results in subsidization of local institutions, jeopardizing the chances of developing in them a sense of financial respon- sibility.17 CADU's Evaluation Unit, for example, observed that one of the ways in which Chilalo's cooperatives can become independent of CADU's benign but suffocating influence is by their involvement in financial mat- ters. There is yet another prerequisite for success of cooperatives which must be taken into account. Evidence from the project reviews and other sources strongly indicated that, where traditional authority and power are unequally distributed, cooperatives become an instrument in the hands of the few relatively large farmers and provide little assistance to subsistence farmers.18 There may also be a conflict between the objectives of the finan- cial self-reliance of cooperatives and participation of the lowest income groups, since the latter may not have the ability to make financial contribu- tions. However, all too often it is the relatively better-off farmers who control the cooperatives and who are also subsidized through programs. The evaluation of CADU's cooperatives, cited earlier, noted that, whereas the proportion of landlords in the overall population was estimated to be about 55 percent in the project areas, landlords constituted between 60 and 95 per- cent of the members in the cooperative societies of Asella, Bilalo, Kechema, and Sagure. The size distribution of the landownership was not known. However, these shoulld have been small landowners with farms of 25 hec- tares or less, since the large ones had been excluded from the program. The study observed that tenants who had not joined were afraid of losing their tenure, had less cash available, and had considerably less land at their dis- posal. Thus, though the requirements and procedures for becoming a mem- ber did not discrimate against them, tenants were very much underrepre- sented. It must be emphasized that it was Chilalo's social structure rather than the program's procedures that seem to be at fault in the pattern of par- ticipation. However, the question remains as to whether the relatively bet- ter-off farmers should not have been made to contribute somewhat more to the cooperative organization than they had been. Even in Tanzania, cooperatives have been embroiled in a controversy over whose interests they serve. The dominant role of the rural elite growing export crops in 17. The question of subsidization applies equally well to the uojmaa movement, as will be il- lustrated later. To the extent that subsidization involves transfer of resources to poor regions, classes, or sectors which would not have occurred otherwise, this may be genuinely desirable. However, if such a resource transfer does not generate a long-term productive potential through augmenting technical and organizational capacity, its scope is inevitably very restricted in the long run (see Chapter IX for further discussion of this question). 18. Raymond J. Apthorpe, "Some Problems of Evaluation," in Cooperatives and Rural De- velopment in East Africa, ed. C. G. Widstrand (Uppsala: Scandinavian Institute of African Stud- ies, 1970). Also see Goran Hyden, "Government and Cooperatives," in Development A dminis- tration: The Kenyan Experience, eds. Goran Hyden, Robert Jackson, and John Okumu (Nairobi: Oxford University Press, 1970), pp. 296-318. 112 DESIGN OF RURAL DEVELOPMENT these organizations has been the basis of a conflict between the cooperative and the ujimaa movements.'9 The performance of both the formally administered marketing compo- nents and the cooperatives suggests that a simple transplant of modern marketing organizations into the traditional sector is unlikely to be viable, not only because of the shortage of skills, financial capital, and infrastruc- ture and communication facilities, but also because of the existence of in- terest groups-e.g., grain boards and the commercial farmers- that stand in the way of a viable market development in the subsistence sector.20 Implications for the Design of Marketing Components What lessons do the past programs offer for development of the market- ing systems? First, to avoid serious marketing bottlenecks, emphasis on in- creased food production often needs to be gradual, geographically dis- persed, and commensurate with a simultaneous development of markets rather than on highly integrated, ambitious package programs which aim at increasing marketed surpluses rapidly without devoting adequate attention at the outset to institutional development. A second lesson of the projects reviewed is that, as soon as large rural de- velopment programs with a substantial potential food production compo- nent are identified, examination of the overall price policy and resource allocation questions should be undertaken with particular emphasis on the various marketing institutions, their performance, potential, and their rela- tionship to the organization of the program. Price policies adopted in pro- grams should become a part of an overall price policy rather than ad hoc measures adopted to meet successive unanticipated crises, as has been the case in the projects reviewed. Third, the foremost shortcoming of the approach adopted in many of the programs is that it has paid little attention to making the traditional forms of marketing organization work. The existing documentation, mostly from West African countries,2' indicates that the traditional traders (men and women) possess a highly developed trading skill and carry on an efficient marketing system. Relatively little is known about the traditional trading channels in Eastern Africa.22 19. As was pointed out in Chapter 1, the relatively more well-to-do export crop producers who constitute the rural elite in Tanzania are, nevertheless, small farmers. 20. An extremely good illustration of the conflict between the interest of commercial and subsistence farmers was provided in an unpublished restricted comment prepared by the Kenya National Farmers' Union on the International Labour Organisation's Employment, In- comes and Equality: A Strategy for Increasing Productive Employment in Kenya, 1973. 21. K.R. Anschel, "Agricultural Marketing in the Former British West Africa," in Agricul- tural Cooperatives and Markets in Developing Countries, eds. K.R. Anschel, R.H. Brannon, and E.D. Smith (New York: Praeger, 1969), pp. 140-62; B.W. Hodder and N.I. Ukwu, Markets in West Africa: Studies of Markets and Trade Among the Yoruba and Ibo (Ibadan: Ibadan Univer- sity Press, 1969); Marvin P. Miracle, "Market Structure in Commodity Trade and Capital Ac- cumulation in West Africa," in Agricultural Cooperatives and Markets in Developing Countries, eds. R. Moyer and S. Hollander (New York: Praeger, 1969); Marvin P. Miracle, "Market Struc- ture in the Tribal Economies of West Africa," in Agricultural Cooperatives and Markets in De- veloping Countries, eds. K.R. Anschel, R.H. Brannon, and E.D. Smith (New York: Praeger, 1969), pp. 120-39. 22. C.M. Good, Rural Markets and Trade in East Africa: A Study of the Functions and Devel- opment of Exchange Institutions in Ankole, Uganda, Research Paper no. 128 (Chicago: Univer- sity of Chicago Department of Geography, 1970). MARKETING OF AGRICULTURAL OUTPUT 113 The extent to which the disappearing Asian trader is being replaced by the African trader is not known. However, existing studies show that in Ethiopia there is a well-developed network of traditional trading channels that handles substantial interregional trade and possesses its own market intelligence system.23 Where means of communication and transportation are well developed, regional price movements are well coordinated. In Malawi LLDP project management observed that, although traditional unlicensed traders are barred in the project area, Malawian (as opposed to Asian) traders are quite active and procure a substantial portion of the mar- keted surplus of maize and groundnuts. The project management further pointed out in the interviews that, had it not been for these unauthorized traders, ADMARC's attitude toward purchases in the program area would have led to a substantial drop in maize prices and discouraged farmers from growing it. In LLDP, disposal of grain through nonformal channels is facilitated by the close-knit network of feeder roads that is being develQped in the program area; nearly a third of the project expenditure is allocated to the development of the infrastruc- ture. In the Ethiopian Minimum Package Program, however, despite provi- sions for feeder roads, the progress in their construction has been slow. This has led to a major bottleneck, since increased surpluses in far-off and ill-connected locations do not seem to attract private traders. The Ethiopian case illustrates the need first to remove the basic market- ing constraints of poor institutional development and inadequate means of transport and commnunication. Only then can the more apparent con- straints, such as lack of fertilizer and nonincentive prices, be effectively tackled. LLDP in Malawi, by contrast, has undertaken a very ambitious road development program, though its local institutional capability in terms of trained manpower and ability to mobilize resources necessary for road maintenance is still very limited. Both these cases indicate the need to reach a proper balance between institutional development and measurable targets, whether the latter be in terms of amount of fertilizer distributed or miles of roads constructed. Improving the means of transport and com- munications is critical for improving performance of traditional trade chan- nels. The removal of these and other constraints receives relatively little emphasis in the marketing components in comparison to the attention paid to replacing traditional marketing systems. The approach to rnarketing that is suggested involves attacking the con- straints faced by the farmer (and/or the trader), rather than spoon-feeding the farmer through tailor-made marketing programs. This will require several simple but explicit steps to improve the farmers' knowledge of the process of price formation, of the standard weights and measures, and of the role that these factors play in causing high market margins. The studies conducted by the Evaluation Unit of CADU, the survey by Manig of the German AID Mission in Bako, and the study in marketing by the Stanford 23. Government of Ethiopia, Economic Research Division, Planning and Program Depart- ment, Findings of a Market Structure Survey and an Analysis of Those Grains That Provide the Basic Subsistencefor the People of Ethiopia (Soddo: Ministry of Agriculture, July 1973); also see W. Manig, Marketing of Selected Agricultural Commodities in Bako Area, Ethiopia, Occasional Paper no. 66 (Ithaca, 1N.Y.: Cornell University Department of Agricultural Economics, December 1973); and Thoday, Marketing of Grains and Pulses in Ethiopia. 114 DESIGN OF RURAL DEVELOPMENT Research Institute24 note that most price exploitation that has been ob- served in Ethiopian markets is covert, through false weights and measures, rather than overt. This implies that the farmer does have the potential to enjoy real bargaining power. Once the cultivators are made aware of the covert practices, the demand for correct weights and measures can be sub- stantial. There probably could not be a better testimonial to the importance of introducing correct weights and measures than the resistance received from traders to CADU's program of making available standard weights and measures at its trade centers. Cultivators in the program area could use the facilities even if they were selling their produce to the private traders and not to the trade centers. The plan was tried on an experimental basis in four village markets and the primary market at Asella. In almost all of the areas where the program was operative, local merchants resented it vehemently; and at times they were violent. In Bekoji and Asella the municipality chiefs opposed the whole idea of the program and exhibited strong opposition to the scheme. lntervention of the enderassie (governor) was required to imple- ment the program in Asella. In spite of these setbacks, during the two and a half months that the program was operative, the Planning and Evaluation section considered the experiment as a relatively successful venture and suggested that it be run in the future on an expanded basis. Farmers also need a market intelligence program to inform them of prices prevailing in the major market centers and of the importance of these prices for the prices they should receive. A survey conducted early in 1972 in Lilongwe indicated that half of all the male growers surveyed had no idea of the current maize price and were unwilling even to hazard a guess. Among those farmers who gave a maize price, only half were within the range of the correct ADMARC price-K2.50 per bag-while a quarter knew the price offered by the program's maize shellers. Thus, of all the farmers queried, approximately three-quarters did not know the price of maize that could be obtained through the legal outlets. It is interesting to note that, of the 60 percent of male farmers stating a price that could be ob- tained from local traders, almost two-thirds gave a figure higher than the ADMARC price.25 Tecle observes that, generally, the lack of adequate and reliable information at all levels of the marketing system is a primary cause of unequal bargaining power, poor producer prices, and the low level of in- ter-market price consonance. Manig, on the basis of his microlevel market- ing survey conducted over a three-year period near Bako in Ethiopia, pointed out in personal discussions that informal exchange of price infor- mation is substantial among farmers and that farmers are generally well in- formed of the prices prevailing in the nearby areas. However, the extent of their knowledge about prices is closely related to their physical access to markets and to whether the crop is commercial, such as chilies, or subsis- tence, such as sorghum. The Bako farmers sell very little of the latter.26 24. Manig, Marketing of Selected Agricultural Commodities in Bako Area, Ethiopia; and Tho- day, Marketing of Grains and Pulses in Ethiopia. 25. According to LLDP's management, the validity of these surveys may be questionable, since farmers are not used to thinking of prices in standard European measures. More re- cently in disseminating prices, LLDP has made an attempt to translate prices into units more familiar to the farmers. 26. See also Manig, Marketing of Selected Agricultural Commodities in Bako Area, Ethiopia. MARKETING OF AGRICULTURAL OUTPUT 115 Price information could, of course, be facilitated through improvement of feeder roads and increased farmer mobility or through extension agents. In Malawi LLDP has undertaken such a price dissemination program through extension agents and through displays of prices in marketing cen- ters. The program has been somewhat restricted in scope, as it attempted to inform farmers only of the price paid by LLDP for purchases on behalf of ADMARC, whereas a considerable portion of the produce was sold elsewhere. The marketing service provided by programs also needs to con- vey to the farmer the risks involved in seasonal storage so that the cultiva- tor will have the choice of storing for the off season if he is willing to under- take the risk. In essence, in designing and implementing marketing components, there needs to be a shift frorn viewing trading as an unproductive and anti-social activity to viewing it as a productive and socially desirable endeavor which requires considerable skill and involves a great deal of risk. In developing the marketing system, it is also essential to convey to the farmers the im- portance of cooperation among them and to impart to them the necessary accounting and trading skills. It is possible, however, that the cultivators may decide to undertake intermediate forms of cooperation, as, for in- stance, organizing themselves to ensure the implementation of standard weights and measures or to build market yards or storage facilities for their common use. As another intermediate form of intervention, storage facil- ities may be combined with seasonal advances for grain storage by cultiva- tors in the off season. This will allow farmers to repay their cash obligations without forcing them to sell their crops in the immediate post-harvest period, while at the same time encouraging them to share the risk of seasonal storage. Initially, at least, farmers may decide to leave the trading function to the experts. But, in any case, the important principle of sharing benefits and costs (including risks) must be conveyed to the farmers from the very initial stages in any program of market modernization.27 27. This very point was also emphasized in the report of the ILO to the government of Tanzania. See International Labour Organisation, Report to the Government of the United Re- public of Tanzania on Consumers' Cooperatives (Geneva: ILO, 1970). p. 58. CHAPTER VII Social Services The provision of social services has been of vital importance in many of the projects reviewed under ARDS. The humanitarian argument for such services is obvious. Improvement in health, water supply, diet, and educa- tion can improve the welfare of the rural population. There are also sound economic reasons for introducing social welfare services, since better health, reduced domestic chores, and higher levels of education can con- tribute to increases in agricultural productivity through increases in the quantity and quality of labor. Planning of social services, however, poses by far the most difficult problem of resolution between economic, social, and technical choices. This is because demand for a wide range of social services is substantial in many rural areas. However, the financial, manpower, and organizational resources for effective delivery of social services are often extremely limited. Consequently, three sets of interrelated questions are faced in planning rural development programs. First, given that resources are limited, how should additional fiscal, organizational, and manpower resources be raised from within the rural areas? Second, are such resources likely to be mobilized more easily if the services to be provided lead to a measurable impact on productivity or if they generate substantial enthusi- asm for development programs among rural people? In other words, are social choices regarding allocation of resources related to social willingness to pay? Third, how can low-cost delivery of services be organized? Is greater participation in planning and/or implementation of programs more con- ducive to a low-cost delivery of services? This latter question of relation- ship between cost and participation is, of course, related to the nature of ser- vices and the degree of technical input required in planning and implemen- tation. These various issues are dealt with in this chapter. The existing information on social services is, however, sparse and pro- vides little systematic evidence, either on the impact of these various social service interventions on rural welfare or on productivity. Nor does much systematic analysis exist of alternative technical and administrative choices in provision of social services or of the effect of these choices on the costs of organizing social services. What data that do exist are largely of a sugges- tive nature, underlining the need for more research in this critical area. Because of the paucity of hard information, the discussion in this chapter is necessarily tentative. The purpose is more to raise the important issues faced in designing rural development programs than to provide definitive prescriptions. SOCIAL SERVICES 117 Because of the multiple causes of low welfare and productivity, social ser- vices are defined broadly in this chapter. They include the commonly in- troduced measures such as broad-based immunization against disease and provision of rural health clinics. Improved water supplies, labor-saving measures in the hornes, and improvement of local diets through new varieties of food crops are no less important for rural welfare, although perhaps less evident. And, of course, educational and training programs are crucial in any effort to improve the quality of life of low-income rural popu- lations. The implications of education and training for rural welfare are dis- cussed separately in Chapter X. Economic Justification for Social Services Several project reviews have drawn attention to the generally poor health of the population and the adverse implications which ill health has for labor productivity. In the squatter settlements in Kenya, the settlers have poor resistance to diseases due to heavy energy demand of farm activities under conditions of low energy supply, inadequate medical services, and insuffi- cient insulation against disease vectors. In ZAPI malaria and intestinal parasitic diseases and sores are predominant and are common at all ages. Among children numerous cases of bronchitis, pneumonia, and epidemic illiness are recorded. The farmers are especially susceptible to diseases because of the lack of preventative medication. It is not rare to see children and even adults die of malaria.' Apart from its effect on mortality rates, the extremely debilitating effect of malaria has been well documented.2 Chronic parasitic infection can have a similar effect on overall physical well-being and performance.3 The reviews both of ZAPI and of the squatter settlements point out that such health problems could be reduced by simple protective measures, such as the provision of quinine to prevent malaria or the introduction of water filters to reduce the incidence of intestinal parasites. Local food shortages, inadequate nutrition, and susceptibility to debilitat- ing diseases form a vicious cycle in many rural areas. In Kenya, for exam- ple, WHO surveys show that throughout the year about two-fifths of the rural families consume less than 80 percent of the average protein require- ments.4 Vitamin and mineral deficiencies are also widespread throughout Africa. Insufficient dietary intake is, of course, compounded by disease and parasites, which make a higher demand for nutrients to preserve integrity. 1. In an analysis of health programs in Niger. Belloncle cited a WHO survey, according to which 93 percent of the population is affected by malaria. See Guy Belloncle, "Total Participa- tion in Public Health Programs: Some Reflections on the Niger Experience," mimeographed (Paris: Institut de Recherches et d'Applications des Methodes de Developpement, 1973). 2. See Robin Barlow, The Economic Effects of Malaria Eradication (Ann Arbor: University of Michigan School of Public Health, 1968). 3. For descriptions of parasitic and other common tropical diseases, see G.W. Hunter, W.H. Frye, and J.C. Swarzwelder, A Manual of Tropical Medicine, 4th ed. (Philadelphia: W.B. Saun- ders, 1966). 4. See M. Bohadal, N.E. Gibbs, and W.K. Simmons, "Nutritional Survey and Campaign Against Malnutrition in Kenya, 1964-1968," Report to the Ministry of Health of Kenya on the WHO/FAO/UNICEF-assisted project, mimeographed, n.d. 118 DESIGN OF RURAL DEVELOPMENT To the extent that their health is affected by poor diets, rural people are able to devote fewer and less-productive hours to agriculture. In addition, there seems to be a relationship between malnutrition and mental deficiencies, which poses a further constraint to increased productivity. It is obvious, therefore, that better health facilities and preventative - medicine can provide only a partial solution to the rural health problem. Efforts must also be made to improve diet through increasing or altering food crop production. This is desirable, even in many cash crop schemes, - where shift from food to cash crops may lower nutritional standards if in- - creased income is not devoted to food purchases.5 As pointed out in Chapter II, in some cases the introduction of new food crops-which replace traditional subsistence crops-may itself create imbalances in the local diet. The replacement of the more nutritious and drought-resistent cereals like millet and sorghum by maize, noted in Sukumaland in Tanzania, has also been observed in the marginal areas of Kenya by the World Bank's agricultural sector survey. In northeastern Tanzania, surveys have found that the increasing prestige of cabbage over wild spinach has limited vitamin A intake.6 New cropping patterns may also counterbalance these developments, e.g., increased cereal/legume production can increase the availability of protein, calories, and calcium; greater fruit and vegetable production can augment intake of vitamins A and C. However, research is needed to find locally suited varieties of food crops that are both high yield- ing and are high in nutritional content. Another step which could improve local diets is to foster better methods of food storage and preservation. This is important in several respects. Ap- proximately 25 percent of the food produced in Africa is said to be lost through spoilage.7 If these estimates are correct, better storage techniques may increase rural food intake significantly and perhaps provide a greater marketable surplus. Further, improved methods of preservation can main- tain the nutritional value of stored food, prevent food poisoning and other ailments related to the consumption of spoiled food, and help eliminate the off-season hungry months. The technology involved in improved food storage and preservation need not be complicated. The use of raised cribs, protected by rat baffles, can frequently protect stored grain from rodent at- tack. Sun drying of fruit and vegetables is an inexpensive and widely ap- plicable means of preserving food. Thus, there seems to be a wide scope 5. One study in Kenya found that cash crop farmers are nutritionally no better off than subsistence farmers, at least in the short run. The apparent explanation for this fact is that cash farmers preferred to spend their increased income on conspicuous consumer dura- - bles-e.g., bicycles, radios, and on education for their children. Improved diet was given a - relatively low priority. It is suggested that educational programs stressing health and nutrition could help to change this situation. For details see R. Korte, "The Nutritional and Health Status of the People Living on the Mwea-Tebere Irrigation Settlement," in Investigations into - Health and Nutrition in East Africa, eds. H. Kraut and H.D. Cremer, Afrika-Studien no. 42 (Munich: Weltforum Verlag, 1969), pp. 267-334. 6. J. Kreysler and C. Schlage, "The Nutrition Situation in the Pangani Basin," in Investiga- tions into Health and Nutrition in East Africa, eds. H. Kraut and H.D. Cremer, Afrika-Studien no. 42 (Munich: Weltforum Verlag, 1969), p. 170. 7. United Nations Economic Commission on Africa, Food and Agriculture Organization of the United Nations, and the Netherlands government, Report: 5 Workshops for Trainers in Home Economics and Other Family Oriented Fields, Eastern and Southern Africa, mimeographed, 1973. SOCIAL SERVICES 119 through home economics extension to improve the health and welfare of rural people and concurrently to increase the productivity of their labor.8 In addition to low labor productivity, low labor availability is also one of the chief constraints to increasing agricultural production. Poor health and lack of health facilities can have a direct effect on labor availability. For in- stance, residents of the Kenyan squatter settlements have to go to the Trust Land Area if they require medical attention. For the Ngwata squatters this means a trip to Makindu, a place over 25 miles away. The amount of time and scarce cash spent traveling to and from these places is considerable. Further, the number of illness periods reported by householders in the set- tlements is significantly greater than has been reported in the sixteen con- trol sites. The cost of these periods of illness in the form of reduced labor availability is difficult to quantify but is higher than in other areas. In view of the substantial effect of ill health on the allocation of time for productive activities as well as for family welfare activities such as child care, it would appear that the provision of a set of fairly simple health and welfare facilities would contribute significantly to the efficiency of time use. However, caution must be exercised before instituting measures hastily. The provision of water supply is a case in point. It has been noted fre- quently that, if water supply were within easy reach of homesteads, it would significantly reduce the amount of time spent by women fetching water. However, in order to provide such facilities on an economic basis, a fairly high degree of residential concentration is frequently necessary. Par- ticularly in unproductive areas where extensive farming is practiced, this often necessitates walking long distances to fields. The result may be little or no increase in the amount of labor time available to agriculture. The in- terrelations between social services and the quality and quantity of labor are, thus, extremely complex; highly variable; and, hence, need to be ex- amined carefully. Where absence of particular social services leads to ex- treme adverse effecl: on welfare and, hence, on labor productivity, the allocation of scarce resources to such services can, of course, be justified both on productive and welfare grounds. It is where such adverse welfare and productive effects are not extreme that it becomes difficult to establish economic priorities without a substantial improvement in the knowledge of the relative effect of various services on productivity and welfare. 8. From the viewpoint of agricultural production, improvements in home economics can provide additional benefits. More efficient use of domestic labor can result in significant sav- ings in time and energy, which can be transferred to other productive activities. However, far too often the identification of constraints and the interventions introduced are not very rele- vant to the low-income situations. For instance, an FAO study notes that in a poorly planned home a woman may have to walk more than 180 meters to prepare a pot of porridge. Accord- ing to the study, rearranging the location of the stove, the food storage area, and so forth can significantly reduce the walking required in housework. The efficiency of such high-in- come-dominated home economics studies in the case of very small homes, frequently noted in the case of low-income rural households, seems highly questionable. However, there can be more useful, inexpensive, intermediate labor-saving technologies available for home use. For instance, in Zambia the Intermediate Technology Research Station is said to have designed a $1 manual maize huller that can reduce the time for hulling a bag of corn cobs by one-third as well as reducing wear on the hands. For a further discussion of home economics innovations and techniques, see ibid. Also see Chapter 11 of this study for additional points on the implica- tions of labor-saving technology for labor allocation and productivity. 120 DESIGN OF RURAL DEVELOPMENT Discrepancies Between Demand for Social Services and Resource Availability Establishment of priorities is necessitated, particularly, in view of the substantial demand for social services in many rural areas. Such demand is reflected in the eagerness with which rural people often contribute their labor to construction of schools and clinics. It would appear, therefore, that the allocation of resources to specific services may quite effectively be linked to the additional resources generated for those services. However, despite the widespread evidence of demand for services and for the poten- tial for local resource mobilization, in practice social service programs have often been implemented primarily with central or donor resources and with little regard for utilizing grass root support. Such policies call into question the long-term viability of many social service programs, as is illustrated by the evidence from the projects reviewed. The problems faced by LLDP management in raising recurrent resources for health services illustrate the point. Provision of health services was not included in the original design of LLDP. Further, it is the declared policy of the Malawi government that improved health remain a low priority in the immediate future. However, because of the very substantial local demand for health clinics, the project management subsequently decided to provide such facilities. Villagers contributed their labor enthusiastically for con- struction of clinics. However, the management has had to make ad hoc temporary arrangements through missionaries and other voluntary organi- zations to provide medical personnel. As of mid-1974 the project authority had not received a commitment from the government, either for the alloca- tion of fiscal resources or for the provision of trained manpower to main- tain the activities of the health centers. The LLDP example, of course, raises the question as to whether, in initially planning the health services, additional recurrent resources could not have been raised locally, given the substantial local demand and given the substantial allocation of resources to the program area, which has already had a considerable favorable effect on rural incomes. Social Services Administration, Fiscal Responsibility, Regional Equity, and Self-Reliance How social services are organized, of course, should have substantial im- plications for the extent to which resources for their provision can be gener- ated locally. It is, therefore, useful to examine the extent to which the top down and the bottom up strategies of development provide different lessons in the organization of social services. The top down approach is an apt description of the earlier externally financed integrated programs such as LLDP, CADU, and WADU, which were in essence planned and imple- mented by expatriate staff (see Chapter VIII). In contrast, the bottom up ap- proach is exemplified by the ujtmaa movement in Tanzania. It reflects a national political philosophy and emphasizes local participation and de- centralization of the indigenous administration (see Chapter IX). In- terestingly, both approaches seem to encounter substantially analogous problems in providing social services. SOCIAL SERVICES 121 Difficulties related to administration, financing, and regional imbalances encountered in the top down approach are best illustrated by SODENKAM in Cameroon. In order to meet the substantial demand for health services, SODENKAM provided facilities which include six dispensaries; fourteen medical attendants; a hospital; a testing laboratory; and an advanced health center that is administered by a registered chief medical attendant, two assistant medical attendants, and three nurses. Mobile clinics are arranged weekly in those villages that have no permanent clinics. The medical sup- plies for the clinics are provided through SODENKAM's budget. In the 1971/72 period 110,000 consultations were provided in pioneer villages, although the total population in these areas is only 3,500. Thus, the program area has a dense health service coverage compared with neighboring areas. However, in SODENKAM, services were organized without identifica- tion of local needs and as in LLDP without exploring possibilities of local financing. This explains the emphasis on elaborate clinical procedures rather than on simple preventive measures as well as the substantial amount of financing provided by the project authority. It also highlights the extensive regional imbalances resulting from the program. The intensity of services available to the indigenous population of the Nkondjok arrondisse- ment (district), bordering on the SODENKAM settlement areas, is ex- tremely low. The estimated 16,000 inhabitants of the area are served by nine primary school teachers, one state medical attendant, and three dis- pensary attendants. The comparable figures for SODENKAM are twenty- six to forty teachers and fourteen health workers for a population that is less than a quarter of that of Nkondjok. Not surprisingly, the disparity be- tween the services provided by SODENKAM and those provided for resi- dents of Nkondjok has generated much envy and frustration. A 1968 agri- cultural survey points out the danger of creating a barrier between the poorly served local population and the pioneer villages, thus providing "an ideal opportunity for what might well develop into serious friction." 9 It could, perhaps, be argued that the SODENKAM example is somewhat extreme. Nevertheless, it does epitomize very clearly the dangers inherent in a misguidedly elaborate program which, despite it size, fails to meet the real needs of the rural people and which might, in fact, create problems which far outweigh any benefits. The ujamaa villages in Tanzania illustrate the difficulties which may be encountered in a bottom up administration. In recent years Tanzania's public expenditure devoted to the provision of social service programs has been substantially greater than that devoted to productive activities. Ac- cording to the information gathered by the World Bank's rural sector survey of Tanzania, education, health, and water programs accounted for about 60 percent of the regional development budget during the 1972/73 9. P. Capot-Rey, D. Audebert, and R. Owerra, Operation Yabassi-Bafang: Enquete Agricole (Yaounde, Cam.: Ministere de l'Agiculture, Direction des Statistiques, May-June 1968), p. 41. 10. The sector survey points to similar trends on a national scale-i.e., including ministerial spending. According to the information gathered, the share of directly productive investment declined from 27 percent of all development expenditures in the 1969/70 period to 13 percent in the 1973/74 period. In the three-year period ending in 1972/73, recurrent expenditures on social services increased more rapidly (11.2 percent per annum) than recurrent expenditures on economic services (7.7 percent per annum). To some extent as a result of this spending pat- tern, agricultural production appears to be stagnating and perhaps even declining, a question discussed later in Chapter IX of this study. 122 DESIGN OF RURAL DEVELOPMENT and 1974/75 periods. In the same period the share of agriculture ranged from 16 percent to 21 percent. These social service-oriented development allocations have led to a similar imbalance in regional recurrent expen- ditures. Health, education, and water services have absorbed 50-60 percent of the total regional recurrent expenditures during this period, while the share of agriculture has varied from 4 percent to 12 percent. In the 1973/74 period the regional budget accounted for about 22 percent of the public ex- penditure in Tanzania.'0 The increased allocations to social service programs seem to have been the result of a need to elicit popular support for ujcmaa as well as of the gov- ernment's ideological commitment to improve overall rural welfare. The re- cent decentralization of the rural development administration, which has assured that the needs of the poorer farmers will receive priority in the allocation of funds, has reinforced the trend towards increased social ser- vice spending. However, despite Tanzania's pronounced support for self- help, there has been little local fiscal participation in this process. Quite to the contrary, local taxing authority has been reduced; and many taxes have been eliminated as colonialist and oppressive. Consequently, funds for local development must come increasingly from the center. Because the govern- ment is able to channel resources into the country's poorer areas, it helps to avoid regional imbalances of the type exhibited by SODENKAM's social service program. However, in its emphasis on bringing about a regional bal- ance through social services, Tanzania has basically relied on foreign assis- tance rather than on mobilizing additional domestic resources. Substantial assistance from bilateral donors, especially the Scandinavian countries, is being accepted without due regard to the recurrent cost implications, which are generally financed nationally. The amount of resources devoted to non- aided development programs has fallen, both in absolute amounts and rela- tive to the total development expenditure. Given the long gestation period and high recurrent cost implications of investment in many social services, emphasis on a massive provision of social service programs seems misplaced, not only from the point of view of its impact on the domestic economy, but also in terms of Tanzania's declared policy of self-reliance. The costs implicit in social services and the difficulties in generating ad- ditional central resources imply that, in rapidly expanding investments in the rural sector, the initial emphasis may frequently have to be restricted to those services which will increase productivity. The provision of the other services, including many social services, may be feasible only when addi- tional local resources can be raised for this purpose. This seems necessary for the following reasons: 1. The rural sector is very large in most African countries. 2. The average income is very low, and with a few exceptions it is relatively equitably distributed in rural areas in Africa." 11. Although the estimates of income distribution are the poorest for Africa, whatever evi- dence exists shows that, by and large, the distribution of income in Africa is more equitable. See Irma Adelman and Cynthia Taft Morris, "Who Benefits from Economic Development?" International Meeting of Directors of Development Research and Training, Belgrade, August 28-30, 1973. Also see A.M. Kamarck, "The Structure of the African Economies," in Economic Change and Development: TropicalAfrica, ed. J. Oliver Hall (East Lansing: Michigan State Uni- versity Press, 1969), pp. 59-80. SOCIAL SERVICES 123 3. Consequently, the:re is not a significant scope for income redistribu- tion as in many countries in Latin America. The low tax base origi- nates in the relatively small commercial agricultural and modern in- dustrial sectors.'2 4. Finally, the administrative capability and the political will required to mobilize additional resources or to divert existing resources from ur- ban to rural areas is frequently limited. Under these circumstances the substantial allocation of central resources to social services frequently occurs at the cost of more immediately produc- tive investments in rural areas and, therefore, may prove self-defeating in the long run. Costs of Social Services The provision of social services will, of course, not have to be restricted if greater attention is devoted to developing low-cost social service systems that are in line with the generally low resource base. Alternatives do exist which can lower labor costs. For example, the use of "barefoot" doctors and paramedical personnel may result in major savings in the provision of rural health clinics. The use of harambee or self-help labor can reduce the costs of developing local water supplies. However, programs organized with local labor to reduce recurrent costs frequently suffer from organizational con- straints, thus reducing their effectiveness. The weaknesses in the Kenyan self-help program organized to provide water supply are discussed in Chapter IX. Some social services also involve considerable capital costs and technical input that can be largely inflexible and incapable of significant reduction. For instance, the construction of water facilities often involves large initial outlays for pipes, pumps, and drilling equipment. The recurrent budgetary requirements and, hence, the need for continuous resource mobilization is also limrited in the case of such services. The recurrent expenditure requirements, however, vary substantially among services, as illustrated by the Kenyan estimates for health and education on the one hand and for roads, waterways, and agriculture on the other (Table 7.1). Compared to the Kenyan estimates, Tanzanian data show even larger recurrent shares in the case of health and educational services (Table 7.2). Consequently, although the scope for direct local mobilization of resources varies between types of services, the high recurrent expen- ditures involved in somne services may also create insupportable financial - burdens. This is particularly true where large capital outlays have been made to provide physical facilities, such as school buildings or health - clinics. It is these complex relationships between the capital and recurrent 12. D. Ghai, "Towards Tax Reform in East Africa," in Readings on Economic Development and Administration in Tanzania, ed. H.E. Smith, Institute of Public Administration, University College, Dar es Salaam (Londion and Nairobi: Oxford University Press, 1966), pp. 195-208; G.K. Helleiner, "The Fiscal Role of the Marketing Boards in Nigerian Economic Develop- ment, 1947-1961," in Economic Journal, vol. 74, no. 295 (September 1964), pp. 582-605; A. Ogunsheye, "Marketing Boards and Domestic Stabilization in Nigeria," in Review of Econom- ics and Statistics, vol. 48, no. I (February 1966), pp. 69-78, USAID, Statistics and Reports Divi- sion, "Central Government Finances: Kenya," in AID Economic Data Book firAfrica, revision no. 286 (Washington, D.C.: 'U.S. Government Printing Office, October 1972): and i. Waller- stein, "The Range of Choice: Constraints on the Policies of Governments of Contemporary African Independent States," in The State o/fthe Nations: Constraints on Development in Indepen- dent Africa, ed. M.F. Lofchie (Berkeley: University of California Press, 1971), pp. 19-36. 124 DESIGN OF RURAL DEVELOPMENT costs of various social services and the associated technical input which determine the scope for provision of inexpensive welfare benefits to the rural people and for mobilizing resources directly for specific services. Table 7.1: Estimated Development/Recurrent Expenditure Ratios for Public Services,a Kenya, 1966-70 Roads 5:1 Waterways 5:1 Agriculture 2:1 Education 1:3 Health 1:3 a. Development expenditures are primarily capital investment. Source: Government of Kenya, Development Plan for rtie Period 1965/66-1969/70 (Nairobi: Government Printers Office, 1966), p. 40. Table 7.2 Estimated Capital/Recurrent Expenditure Ratios for Education and Health,a Tanzania, 1973-80 Year Education Health 1973 1:10 1:5 1974 1:8 1:5 1975 1:7 1:6 1976 1:8 1:6 1977 1:8 1:6 1978 1:7 1:6 1979 1:7 1:5 1980 1:7 1:5 a. Based on proposed capital and recurrent expenditures. Source: Government of Tanzania, An7nual Manipower Report to the Presidenit (Dar es Salaam: Ministry of Economic Affairs and Development Planning, 1971) and working figures of the Ministry of Health. Implication for Resource Mobilization and Increasing Participation Social services provide considerable potential for involving, as opposed to merely reaching, rural people. The project reviews suggest that there may be a close relation between the degree of politicization and the willingness and ability of the rural people to articulate their demands for social services. How can policy measures mobilize the maximum potential of local admin- istrative talent and financial resources? At least in some cases, local partici- pation and resource mobilization provide a means of overcoming the prob- lems of the high recurrent costs of many social services. It could be argued, however, that a self-help type of approach might contribute to widening regional disparities, since the more productive areas may be able to generate more resources for social services. If this is the case, encouraging a self-help approach might require some fairly direct policy interventions, particularly in the initial stages, to correct regional disparities, in particular through the SOCIAL SERVICES 125 allocation resources to generate a productive potential. Experience in Kenya, however, suggests that mobilization of local resources is not neces- sarily related to income and, indeed, the poorer areas may frequently be more successful in initiating self-help activities. There have been a large number of spontaneously initiated harambee projects in the squatter settlements in Kenya. These schemes have blossomed in Ngwata and the Ngoliba areas where the need for schools and -health clinics is urgently felt. In Ngwata in 1973 local self-help projects were attempting to raise KShl80,000 from the area's residents in order to fund both a secondary school and a medical center. Similarly, in the Ngoliba set- -tlements primary schools have been built with volunteer labor.'3 Mbithi and Barnes attribute these remarkable instances of local initiative to the isolation of the area from government services and infrastructure, thus forcing the people to unite in their struggle for maintaining a minimum de- cent standard of living. While it may not be possible to undertake self-help health or education programs on as massive a scale as the above in less dis- advantaged communities, it may, nevertheless, be feasible to attempt to mobilize local resources through fund-raising drives or through the charg- ing of small fees for the service.14 In addition to demonstrating the potential for resource mobilization, the Kenyan self-help projects emphasize an additional important point- namely, the need to arrive at the optimum degree of government interven- tion. Experience in the Tetu district in Kenya-an area with a long history of local initiative-provides much insight into this question. In an evalua- tion of SRDP, Almy and Mbithi contend that the number of self-help schemes has actually declined subsequent to the introduction of SRDP in 1970. Before 1968 seventy-seven such projects had been started and sixty- five others were added in the next two years. In 1970, however, the number of new projects registered dropped off dramatically to only twenty-two.'5 Almy and Mbithi attribute this to the rather heavy-handed attempts of SRDP at integrating into the on-going self-help projects. They note: "Self- help in Tetu has been covertaxed by the large number of Harambee projects pushed by aspiring politicians in the last few years ... Poor farmers can only contribute a limited amount to development, but the politicians do not recognize this local problem in their struggle for national recognition. Field workers report that there are in fact a number of new self-help groups form- ing in these areas, but they are avoiding registration so that their project is .not expanded by politicians or admihistrators beyond what they can afford."'6 Like SRDP, ZAPI bhas occasionally pushed peasant organizations to do too much too soon. Although the evidence suggests that such clumsy inter- vention as that in SRDP or ZAPI is likely to prove self-defeating, projects must also avoid the paternalism evinced by LLDP, SODENKAM, and 13. At one school built in 1970, money from local inhabitants was used even to pay teachers' salaries. 14. As indicated earlier in Chapter V of this study, tontines, or savings clubs, were ob- served in many of the project areas. 15. S.W. Almy and P. Mbithi, "Local Involvement in the SRDP," in An Overall Evaluation of the Special Rural Development Programme, 1972, Institute for Development Studies Occa- sional Paper no. 8 (Nairobi: University of Nairobi, 1973), Section G, p. 13. 16. Ibid., Section G, pp. 14-15. 126 DESIGN OF RURAL DEVELOPMENT ujemaa. The approach recommended here is to allow the people to deline- ate for themselves their critical needs and to identify the remedial steps that the people are willing to take. The programs could then provide the relevant technical, pedagogical, and other forms of assistance, making it possible to alleviate pressing social or welfare problems without overburdening the- local resource base. Implementing this approach will mean taking steps similar to those sug- gested by Belloncle and Gentil for mobilizing popular participation in ex-- tension, as outlined in Chapter IV. Belloncle indicates that there has been considerable success in Niger in organizing public health programs along such lines. Health officials begin these programs by sponsoring group meet- ings at which villagers discuss their medical problems. During the dis- cussions the formally trained medical personnel are made cognizant of the villagers' explanations of and cures for the most common diseases, for it seems much easier for the staff to explain Western techniques, once they have a knowledge of traditional medicine. If the villagers have shown an in- terest in improving their health, they are asked to choose a volunteer for training at the nearest dispensary and to collect money to purchase a village medical kit. The prospective medic is given a brief course in which hygiene, preventative medicine, and first aid are stressed. The more highly trained medical personnel working in the area make follow-up visits to check on the village health worker and to replenish his supplies. The villagers are asked to repay their medic by helping him with his farm work and by bring- ing him gifts in kind. Of the hundred such health workers trained in Marandi province in Niger in 1971, eighty were actively engaged in their villages. During the first six months of the year, they received visits from 83,600 patients, or about half as many patients as were received by the eighteen dispensaries in the province. Of course, these medics are not able to treat all diseases; but they can deal with a wide range of common ailments and do have the knowledge to recognize problems which need professional care.'7 Similar steps could be taken to organize other social services. Village women could be organized to identify their key needs in the field of home economics. One of them could be given brief training and returned to the village to advise her neighbors. Similarly, farmers could be brought together to discuss their water supply needs and alternative solutions. Though new wells may be necessary, it may be possible to delegate one farmer to carry out the job of distributing the water on a community-wide basis with the use of a cooperatively owned ox cart. In practice there are a myriad of po- tential solutions to social service needs. In most cases the appropriate choice of interventions must depend largely on what the rural people are willing to do for themselves."8 17. See Belloncle, "Total Participation in Public Health Programmes: Some Reflections on the Niger Experience." One problem encountered with this approach has been resistance on the part of traditional witch doctors or healers. The problem does not appear insurmountable, however, since conceivably these traditional practitioners could be brought into such pro- grams. Belloncle also notes success in a similar program for training village midwives. 18. See the related discussion on the education of rural people in Chapter X of this study. CHAPTER VIII Forms of Rural Development :Administration 1: A utonomous Projects Issues Related to Project Authorities Although both Ethiopia and Malawi have low levels of income, they also have a few regions with obviously high development potential. These ap- peared to offer a good opportunity to make a substantial impact on the levels of living of the low-income rural populations in selected regions. However, even by contemporary African standards, the two countries were endowed with relatively little physical or institutional infrastructure or trained manpower.' Therefore, to make a noticeable impact in a short period and to deal with the complementarities in production and to some extent in consumption, substantial simultaneous investment was deemed necessary in a number of activities, including agricultural extension, credit, marketing, roads, soil conservation, training, cooperative development, and water supply.2 Because the rural administrations in Ethiopia and Malawi have had very limited capacity to absorb and utilize finances at the district and divisional levels (or their equivalents), both countries were ill-equipped to take on large-scale development functions. For example, in Ethiopia the local-level administration had been primarily a tax collection and a law en- forcement agency and lacked developmental capacity almost com- pletely.3 With perhaps a few exceptions, the past Ethiopian leadership had been apathetic towards, if not at odds with, the concept of mass participa- tion in rural development. In these two countries the inadequacy of the ex- isting institutions to administer complex multisectoral programs and the absence of well-articulated national commitments to bring about substan- tial political and administrative improvements led expatriate planners to 1. See Colin Baker, "The Administrative Service of Malawi-A Case Study in Africanisa- lion," in Journal of Modern Atriicon Studies, vol. IO, no. 4 (1972), pp. 543-60. Also see chapter on "Education" in American University Foreign Area Studies, Area Hanidbook for Ethiopia (Washington, D.C.: U.S. Government Printing Office, 1971), pp. 185-206. 2. Improved water supply was expected to improve health, increase availability of labor for productive activities by reducing time allocated to fetching water, and to motivate people to participate in the programs. See Chapters 11 and VII of this study for further discussion of these points. 3. For a discussion of local government in Ethiopia, see John M. Cohen and Peter H. Koehn, "Local Government in Ethiopia: Prospects for Reform in the 1970s," paper presented to the Seventeenth Annual Conference of the African Studies Association, Chicago, Octo- ber-November 1974. 128 DESIGN OF RURAL DEVELOPMENT view the establishment of autonomous project authorities as a logical way of making a noticeable impact and, thus, of exercising a demonstration effect on policymakers, administrators, and, of course, on the rural people. The relatively little initial emphasis placed on improvement of the in- digenous administrative systems in the resulting rural development pro- grams, such as CADU, WADU, and LLDP, is partly a reflection of the lack of indigenous participation in planning and implementation of these pro- grams. Furthermore, Africanization of administration had not been as im- portant an issue in these countries as it has been in Kenya or Tanzania. The Malawian government's strategy did not preclude extensive use of expatri- ate manpower to maintain administrative efficiency.4 In Ethiopia, which did not have a colonial history comparable to that of many other African countries, there also was relatively little concern about the use of a substan- tial number of expatriates in managerial positions at the time the programs were conceived. Establishment of project authorities involving substantial concentration of expatriate staff was, thus, more feasible in Ethiopia and Malawi than it would have been in Kenya or Tanzania.' The project authorities in these rural development programs have carried out functions of several departments (or ministries) and local institutions involving a wide range of activities related to agriculture, transportation, health, community development, credit, and marketing. It must be noted that the new project authorities are, thus, different from the earlier export crop authorities in at least two important respects: (a) they actually or po- tentially substitute a range of activities that are already the responsibility of the normal governmental structure, and (b) they look to the fulfillment of a broad range of sociopolitical objectives. For these reasons the new project authorities sooner or later face the pressure to integrate with the existing administrative structure. The first step towards integration is, of course, Africanization of the top- level management of the project authorities. The second step is to pass on the responsibility for the various functions carried out by the project au- thorities to the appropriate departments of the central government, local governmental institutions, and commercial organizations without signifi- cantly jeopardizing the administrative efficiency achieved by the project au- thorities. To assess the progress of integrated programs in Africanization and inte- gration, it is necessary to note the characteristics which distinguish project authorities from the normal governmental structure. The managers of the 4. The controversial statement of Dr. H. Banda, President of Malawi, regarding Africaniza- tion in Malawi quoted below is obviously a contrast with the emphasis on Africanization in Tanzania and Kenya. "I do not want to dismiss any European just because he has a white skin ... Our African civil servants ...must be patient, they must be trained, they must become efficient before they can expect me to promote them into the jobs now occupied by the Europeans. I would rather be shot dead . . . than Africanize just to please anyone who wants Africanization." As quoted in Baker, "The Administrative Services of Malawi- A Case Study in Africanisation," p. 557. 5. It must be noted, however, that, despite Kenya's emphasis on Africanization in recent years, as much as 41 percent of the aid to Kenya consisted of technical assistance. In com- parison with many other countries, Kenya has been one of the largest recipients of linancial and technical assistance on a per capita basis. The effectiveness of such assistance in accelerat- ing the process of Africanization has, however, been questioned. See H. W. Singer and A. C. Doss, "Technical Assistance in Kenya: Some Thoughts on Flows and Programming," in Eastert- A/iica Economic Reiew, vol. 1, no. I (June 1969), pp. 17-27. ADMINISTRATION: AUTONOMOUS PROJECTS 129 projects have the clear responsibility and the necessary authority to coordi- nate activities and to supervise performance of a number of different departments. Compared to indigenous administrations, the project authori- ties, of course, benefit from a clearer orientation towards achieving visible and measurable results within a specified time frame and enjoy consider- ably greater independence. This situation is the result of the greater coinci- dence in formation and implementation of work programs and a shorter chain of command in autonomous projects. Further, project authorities usually have greater financial resources and considerably more technical and managerial expertise with the potential for providing the indigenous staff with more training, additional administrative incentives, higher sal- aries, more fringe benefits, and greater job satisfaction. However, there is an additional characteristic of the integrated programs central to their design and performance. Because there are no easy objective criteria by which to judge such accomplishments as the training of man- power or the development of administrative abilities, investment in these integrated programs tends to be judged primarily by the criterion of an ac- ceptable internal rate of return, calculated on the basis of quantifiable pro- duction targets. These targets are derived from standard input-output coefficients for physical inputs, such as fertilizer and seed, and do not ex- plicitly take into account some of the crucial complementarities in realizing production objectives-- namely, the supply of trained manpower to admin- ister an input delivery system, the effectiveness of administrative pro- cedures, or the existence of other physical infrastructure such as roads to fa- cilitate the movement of inputs. Thus, the larger the proportion of expen- diture in a given project on these latter types of components, the greater ap- pears to be the need for ambitious production targets to carry the burden of these indirectly productive activities, so that the project can be acceptable in terms of its internal rate of return. This may necessitate setting ambitious production targets in tlhe short run, introducing a contradiction in the proj- ect design from the very outset,6 for such targets may distract the attention of the project authorities from acquiring and training competent indigenous staff, from evolving administrative procedures that will last long past the stage of donor financing, and from developing effective working relations with the normal administrative structure. The greater administrative flex- ibility and financial resources of the autonomous project authorities and the higher salaries and adlministrative incentives that they may offer to the local staff may also become a source of annoyance and envy to the in- digenous administration. An elitist project administration may develop that is not capable of establishing rapport with the normal administrative struc- ture. These tensions will persist unless the performance and the incentives in the normal governmental administration are improved or unless the autonomous projects lose their special characteristics in the course of inte- gration. Alternatively, or to some extent, concurrently, the tendency of the 6. In LLDP over a third of the expenditure of $1.2 million during the first four years of Phase I was allocated to development of physical infrastructure (capital costs) and $925,000 to staff salaries, manpower training, and so forth. To justify this expenditure along with the ex- penditure on production inputs, 23,000 farmers had to be reached through inputs (and credit) during the first five years to bring about the necessary targeted increase in agricultural produc- tion. It is noteworthy that the number of credit recipients in LLDP increased from 4,000 to 19,000 in one year to keep up with the targets. See Chapter V of this study. 130 DESIGN OF RURAL DEVELOPMENT governmental administration may be to avoid making the modification in regulations necessary to improve the administrative environment for the indigenous staff in the project authorities, thus affecting the availability of indigenous trained manpower to project authorities. In any case, unless there is far-reaching reform of the indigenous rural administration, it seems unlikely that in the long run the objectives of rural development can continue to be realized simply through the establishment of autonomous project authorities. Thus, even if project authorities are to be established as a way of alleviating the constraint of absorptive capacity and of creating a noticeable impact in the short run, there is an important advantage in facing the problem of its integration with the existing govern- ment structure at the outset in that it facilitates the necessary adjustments in project design, thus increasing the long-run effectiveness of the project. On the basis of the experience related to project authorities, this chapter points out the need for adjustments in respect to three factors: (a) the avail- ability of trained manpower and the steps needed to increase their supply, (b) the degree of coordination between various government departments and the effort needed to increase it in the foreseeable future, and (c) the overall administrative and institutional performance and the steps neces- sary to improve it. Without these adjustments and, equally important, with- out the necessary national commitment to tackle these problems, rural de- velopment programs are unlikely to be effective in the long run. Africanization of Management in Autonomous Projects Although the importance of indigenous management is recognized and early Africanization is proposed in most donor-aided rural development programs, in practice turning a project over to African management is beset with many difficulties. For example, in the case of LLDP, IDA proposed that a qualified Malawian be appointed to the post of deputy program man- ager and that other qualified Malawians, who lacked practical experience, spend at least one year as understudies to more experienced officers. The major division chiefs and other senior positions were to have understudies, and a number of supernumerary training posts were to be created for some forty professional and technical positions in the program that were cur- rently filled by expatriates.7 Because of difficulties in hiring qualified Malawian staff, none of these proposals had been put into effect by October 1973-i.e., six years after the program had been under way. Only one candi- date had proved qualified for a top-level management position; he subse- quently transferred to the nearby German-funded Salima project. WADU in Ethiopia experienced similar difficulties in hiring both ex- patriate and Ethiopian staff to occupy high-level positions and had particu- lar difficulty in retaining Ethiopian staff. Of the nine expatriate staff posts, only four were filled at the commencement of the project; and, four years 7. The Phase I appraisal of LLDP envisaged eighty expatriate positions, though only thirty of these have been filled at any one time. The explanation for the limited use of expatriates appears partly to have been the result of a conscious decision by the project management to use indigenous personnel whenever possible. Also, the project has been unable to find enough foreign manpower to staff the expatriate positions. ADMINISTRATION: AUTONOMOUS PROJECTS 131 after establishment of WADU, none of the expatriate staff posts had been taken over by Ethiopians. Two expatriate staff positions were altogether va- cant. By the end of 1972, twelve graduate-level Ethiopian staff (including the deputy project director) had left WADU. Salaries, Administrative Incentives, and Africanization Staffing problems appear to be compounded by the terms, conditions, and administrative salary scales that are defined by the public service com- missions and applied to nationals. At the current salary levels, there does not appear to be the incentive to take on jobs of considerable responsibility involving the disadvantages of living in remote rural areas. Frequently, the moral commitment of the type noted among some Tanzanians is also lack- ing among most staff serving in development projects. The reluctance on the part of the governments to change the rules and regulations applicable to the indigenous staff in project authorities seems to arise from the far-reaching implications of such changes in salary structures for rural administrative staff generally. The close relation of regular govern- ment pay and promotion regulations with lack of progress on Africaniza- tion of staff is evident in LLDP in Malawi. The project authority must pay salaries according to the government scales. Promotions must be tied to a rigid scale of time served in any given position rather than being based on ability or experience in a given project. Unless he has the requisite number of years at the next lower level, Malawi government statutes also forbid the payment of full salary to an individual who holds a post in an acting capacity. It is not possible to pay two men simultaneously for the same posi- tion, however briefly; hence, there can be no overlap between old staff and new and no opportunity for outgoing staff to introduce new staff to either the program or the post they will fill. Statutory regulations also forbid the paying of salaries to staff in training, making it difficult to upgrade staff within LLDP. The project encounters additional problems in staffing due to the complex relations of the project's administrative structure with the nor- mal government administration and the lack of security that this creates for the government staff who might otherwise join the program. In order to permit nlexibility and limit long-term government liability, a dualistic staff structure was created especially for LLDP. This structure consists of two types of posts: established posts, which were permanent and which would continue beyond the life of the program, and nonestablished posts, which were anticipated to last only for the active life of the program. Any officer serving on an established basis cannot resign his post and re- enter government service at a later date, whatever the circumstances of his resignation. As a result, all nonestablished posts are blocked to serving es- tablished officers. To compound the problem, the program has experienced difficulty in keeping nonestablished staff once they have been trained, since many are anxious to enter established posts elsewhere in government so as to obtain job security. WADU's management explains the high turnover of its indigenous staff in terms of similar factors. In WADU the remoteness of the project area is combined with the ability of many of the degree-level staff to find attractive positions elsewhere. Frequently, such positions have prospects for further training and long-terrn careers. Interviews with Ethiopian staff indicate that 132 DESIGN OF RURAL DEVELOPMENT WADU's staffing problems may also have arisen from working conditions that seem unacceptable to the indigenous staff. According to them, WADU had unfair recruitment procedures (different fringe benefits promised for the same-level staff) and an unbalanced salary structure (different begin- ning salaries offered for staff with similar qualifications). In the view of the persons interviewed, the project also suffered from the incompetence of some of the senior staff (both Ethiopians and expatriates), a lack of dis- ciplined working conditions, and from the deliberate exclusion of some staff when making certain important decisions.8 Whether prompted by legitimate or perceived grievances, the dissatisfac- tion of WADU's staff illustrates the vicious circle created by initial staffing problems; poor working conditions; and inadequate training given to the available indigenous staff and, hence, inadequate incentive for them to re- main in project authorities. Because of staffing difficulties, project authori- ties are generally left with relatively few, if any, African employees in man- agement positions and with a considerable concentration of expatriate man- power at the project headquarters. As pointed out earlier, at the time the review of LLDP was carried out, the project had only one inexperienced Malawian in a managerial position, compared to over thirty expatriates. The minority position of indigenous staff is also not very conducive to a profitable interaction between the expatriate management and the few relatively less trained and less experienced African counterparts. Some of the Ethiopian staff in WADU pointed out that this was also one of the problems in WADU. Quite understandably, management practices of the project authorities gravitate towards the style of the expatriate expert, to which Africans often find difficulty adjusting. For instance, both LLDP and WADU, like many other agricultural projects in East Africa, suffer from a shortage of well- trained African accountants. This is one of the many reasons why these projects have poor cost records. Such shortcomings could be rectified by in- stituting a simple cost-accounting system which could be followed by in- digenous staff with a reasonable amount of practical training. CADU is one of the few projects reviewed which has managed to introduce such a simple cost-accounting system and teach African staff to implement it. How does the pace of the projects affect the building of viable managerial systems staffed by African nationals? The projects are frequently geared to meeting ambitious targets. As pointed out in Chapter 111, many of these are by necessity based on the poor initial information at the disposal of the preparation teams. Management procedures for the fulfillment of many of the targets are also poorly specified. As is to be expected, many unantici- pated problems arise at the stage of implementation. The managerial task in many of the autonomous programs is frequently comparable in complexity to that of a small corporation, involving use of funds from various sources, international procurement of items, and the establishment of effective working relationships with many government structures and with donors. Ad hoc managerial procedures to tackle these various problems evolve based on the individual manager's ability to adapt to the system within the 8. The staff made similar complaints to Paulos Abraham, the first Ethiopian director of CADU, and others during their independent evaluation of WADU undertaken for prepara- tion of WADU's Phase It application to IDA. (Based on discussions with Paulos Abraham.) ADMINISTRATION: AUTONOMOUS PROJECTS 133 very severe time constraint under which he operates. Such procedures fre- quently disappear with the departure of expatriate managers, since the man- agers often cannot develop a systematic who-will-do-what-when-and-how management organization which can be imparted to African counterparts. Not all projects, however, have been beset with manpower problems of the magnitude experienced it, LLDP and WADU. CADU had a greater task of Africanization with an initial input of forty expatriates, compared to eight in WADU. However, in the first four years, CADU made significant progress in developing Ethiopian management. The number of expatriate staff had declined to twenty-four by 1972 and to eight by 1974. CADU's progress can be attributed in part to its successful removal of some of the same constraints which LLDP has been unable to tackle. In es- tablishing CADU, provisions were made to employ staff on contract basis with a three-months' notice of termination. This system proved to be effec- tive and had been retained in the Extension and Project Implementation Department (EPID) in the Ministry of Agriculture for all package projects. This worked favorably for CADU in comparison with the procedure employed by the Central Personnel Agency (CPA), which hired all regular government employees. CADU's project director acknowledged that con- tract employment contributed significantly to the effectiveness of its work program, as it allowed promotions and salary increments to be based on performance rather than on tenure, unlike in other government agencies. CADU also seemed to have gradually gathered highly motivated Ethiopian staff who have had amiable working relations with their expatriate counter- parts. The staff attributed this to the meaningful jobs that CADU provided and to the quality of leadership at all levels of the project. In addition, CADU has provided in-service training for many of its staff and sent others abroad for further studies. Among other programs, which seem to have succeeded in attracting, training, and retaining talented indigenous employees, are the older parastatal organizations such as the Kenya Tea Development Authority and the recently formulated Tanzanian Tobacco Authority, which retains much of the staff trained by the British American Tobacco Company and the Tanganyika Agricultural Corporation. Although the parastatals may have trained their own qualified indigenous staff, the diversion of the best African manpower to parastatals may, however, itself inhibit improvement of the existing government departments. The Kenyan Ministry of Agricul- ture has experienced repeated staffing difficulties as a result of the drain of skilled employees to KTDA. This drain is likely to continue, since the Min- istry has made commitments to provide KTDA with its manpower needs through the 1977/78 period.9 Some Contradictions in Africanization of Management Trained manpower is a scarce factor and, therefore, commands high return in the form of attractive salaries in private or parastatal jobs, city life, fringe benefits, and foreign training. At the same time, the terms and condi- 9. See government of Kenya, Kenya Tea Development Authority Third Tea Project (Nairobi: Government Printing Office, 1972), p. 15. KTDA estimates that it will require thirty-nine ad- ditional agricultural assistants for its Fourth Plan. This would amount to 11 percent of the graduates of the Embu Institute of Agriculture between 1973 and 1978. 134 DESIGN OF RURAL DEVELOPMENT tions dictated by the manpower shortage in recruiting and retaining staff for rural development programs poses the danger of further widening the exist- ing differentials in salaries and leading to increased disparities between the salaried classes and the rural people. Once integrated programs are estab- lished, they, therefore, face a difficult dilemma in dealing with the conflicts arising from the realities of the labor market. Both Tecle and Abraham in their independent evaluations of WADU point out that WADU's inade- quate salary and fringe benefits have affected its progress towards Ethio- pianization. However, Abraham also attributes WADU's inability to estab- lish satisfactory rapport with the Awraja administration largely to the high salaries of the WADU staff and the envy that they invite from the local ad- ministration.' When the reviews were carried out, it was too early to judge whether LLDP would encounter difficulties similar to those encountered by WADU (if differences in incentives received by its staff and the staff of the regional agricultural office were maintained) or whether it would lose its best staff (if incentives between the project and the non-project staff were equalized). However, the effect of different working conditions on performance and at- titude of LLDP's field staff are already apparent and hint at the problems likely to be encountered in integration with the existing administration. LLDP has tried to ensure that fieldworkers do not allow their formal training to deteriorate and, moreover, that they learn to translate and apply effectively this training in actual circumstances. There is little doubt that LLDP field staff enjoy working conditions superior to those of the regional staff; this superiority appears to extend as much to the ability, enthusiasm and effectiveness of the senior staff as to the material benefits-good hous- ing, transportation, and so forth-available in the program. Junior staff voiced their strong support for an administration that gets things done and tries to minimize red tape." Politics, Management, and Africanization Apart from questions of salary structures and staff incentives, the inte- gration of autonomous projects with normal administration also raises questions of the effect of likely political influences on the long-run perfor- mance of rural development programs. To a considerable extent the project authorities are free, once the program planning is completed, from social and political pressures concerning such matters as which areas will receive priority in road construction or in soil conservation measures or which areas will receive credit. However, for the same reason, the job of the ex-, patriate project manager is often made more difficult, compared to his in- digenous counterpart, by the fact that the many sociopolitical constraints faced by local administrators are not always easily communicated to an 10. Conflicts between WADU and the local government have also been a result of political factors. Local landlords and merchants, who often found WADU's initiatives at odds with their own interests, tended to pressure local officials into opposing the program. See the later discussion in this chapter. 11. For a further discussion of staff incentives in LLDP and other projects, see Chapter IV of this study. ADMINISTRATION: AUTONOMOUS PROJECTS 135 alien project authority. As the manager of LLDP in Malawi indicated, much of the perspecti:ve he had acquired on the marketing organization (ADMARC) and the interest group orientation of its management were based on the lessons he had learned from the problems faced by the project vis-a-vis ADMARC and not through an open discussion of issues between the two parties (see Chapter V for further details on the relationship be- tween LLDP and ADMARC). Even if an alien in a project authority is somewhat more cushioned from administrative pressures than his indigenous counterpart, problems are likely to arise in the post-integration period. Expatriates may underrate the difficulties of establislhing linkages between project authorities and the ex- isting government administration and may, therefore, make inadequate efforts to establish effective links. Ethiopians have contended that WADU would probably have established better rapport with the administration of Wolamo Awraja if the top management had been Ethiopian. But, did the problems encountered by WADU's management with the local administra- tion arise because the project management was alien or because the objec- tives of WADU were at odds with the interests of the local landowners and merchants? They appear to have resulted from a combination of the two. Although Wolamo has had a less staunch feudal structure compared to Chilalo, WADU's settlement activities still led to a considerable protest from the landowners i,n the region. The review of the autonomous groundnut operation in Mali comments on some of the additional problems that are likely to arise from political and administrative pressure. The integration of the groundnut operation under the Ministry of Production presents challenges (especially for the future as French expatriates are phased out) in defining the relationship be- tween the operation and the Malian administrative structure. Prior to the establishment of the operation, indigenous agricultural staff were fre- quently subject to direct interference from officials in the local administra- tion. For example, a ci rcle commandant (roughly the equivalent of a district officer) could request the vehicle of an agricultural officer for a week's trip. The latter had no choice but to comply and, thus, was unable to carry out his own duties in supervision and training of extension agents at the local level. Holmquist (in his case study of the cattle dip program initiated by the dis- trict administration in Kisii district in Kenya) also points out how the inade- quate ability of various levels of government to resist counter pressures (ad- -ministrative and political) lead to very considerable changes in the outcome during the course of imaplementation, which were not anticipated when the program got under way.'2 CADU provides by far the most striking illustration of the conflicts aris- ing from sociopoliticall pressures exerted by the local power structure and the extent to which they can be resisted by indigenous and alien staff. CADU was one of the first comprehensive rural development programs un- dertaken by SIDA. CADU involved far more careful planning compared to 12. See Frank W. Holmquist, "Implementing Rural Development Projects," in Development A dministration: The Kenyan Experience, eds. Goran Hyden. Robert Jackson, and John Okumu (Nairobi: Oxford University Press, 1970), pp. 201-29. 136 DESIGN OF RURAL DEVELOPMENT SIDA's earlier efforts in North Africa."3 The Comilla project in East Pakistan (now Bangladesh) had a considerable influence on CADU's strategy. CADU is often described as the Comilla of Africa. However, in a recent visit to Ethiopia, Akhter Hameed Khan, the founder and former Director of Comilla, pointed out an important difference between the Com- illa Project and CADU: while CADU was implemented through a parallel administrative structure, Comilla aimed, among other things, at transform- ing the Thana police station, the only local governmental institution of any consequence, from a law enforcement agency to a development agency responsible for all the major development activities.'4 According to Khan, the level of institutional development in East Pakistan in the 1950s, when Comilla was conceived, was probably not much superior to that in Ethiopia.'5 To a considerable extent the noninvolvement of Chilalo's local govern- ment in CADU is explained by the opposition of the past landlord-domi- nated power structure to mass-oriented rural development.16 Given the lit- tle or no direct involvement of nationals, the Swedes, as outsiders, experi- enced considerable difficulty in establishing close links with the local ad- ministration and with the mass of rural people. It is also true, however, that greater national leadership in a rural development program that aimed at tilting the scale in favor of the small farmer would have been unlikely under the political structure that then existed in Ethiopia. Nor should the difficult task of developing local institutions and the time required for it, particularly in a relatively unsympathetic environment, be underrated. The confrontations with landowners encountered by CADU in introducing land leases and credit ceilings or with merchants in introducing standard weights and measures illustrate the strong conflicts of interest that arose concerning the means of rural development. There is a real question as to whether CADU would have had as many accomplishments on its balance sheet, in- cluding a considerable increase in agricultural production and a substantial initiative in getting a nationwide development effort under way, if the proj- ect had taken on the additional uphill battle of developing local institutions in an alien and unwelcome environment. Thus, the fact remains that, 13. Bengt Nekby, CADU: An Ethiopian Experiment in Developing Peasant Farming (Stockholm: Prisma Publishers, 1971). 14. Based on a series of personal discussions with Dr. Khan during Development from Below Workshop organized by the Overseas Liaison Committee of the American Council on Education and Association for the Advancement of Agricultural Sciences in Africa, Addis Ababa, October 12-20, 1973. 15. There are, however, also the interrelated questions frequently raised as to whether the Comilla experiment had been successful in providing transferable lessons for developing the overall developmental administration in the former East Pakistan and as to the extent to which Comilla's success is explained by Dr. Khan's charismatic personality. See Arthur R. Raper et al., Rural Development in Action: The Comprehensive Experiment at Comilla, East Pakistan (Ithaca, N.Y.: Cornell University Press, 1970). Also see the review of the above book by Harvey M. Choldin, in Economic Development and Cultural Change, vol. 20, no. 3 (April 1972), pp. 594-601. 16. Cohen argues, however, that the previous central government's attitude towards rural development has also been a major constraint on CADU's success. He points out that, while the central authorities supported the basic goal of rural development, they had been opposed to the local political changes that this implied. Consequently, in Cohen's view, they did not offer the national commitment necessary to ensure the realization of CADU's objectives and, in fact, may have lent their prestige to tactics that were incompatible with CADU's ultimate success. See Cohen, "Rural Change in Ethiopia: The Chilalo Agricultural Development Unit," in Economic Development and Cultural Change, vol. 22, no. 4 (July 1974), pp. 608-10. ADMINISTRATION: AUTONOMOUS PROJECTS 137 despite CADU's achievements on other scores and despite its considerable effort at the time the review was conducted, it had not yet had a significant success in augmenting the indigenous administrative capacity in Chilalo or in developing viable grass root institutions."7 Compared to that of the Swedes, the Ethiopian management of CADU was, of course, more open to sociopolitical pressures, particularly those gen- erated by local vested interests. To some extent the departure of CADU's first Ethiopian manager is explained by the demoralization that resistance to such local pressure prompted. But then how useful are the short-run achievements when in the long run the projects will be run by nationals who are more subject to domestic sociopolitical considerations? To what extent does national commitment affect the realization of the participatory and productive objectives of the projects subsequent to national takeover? To what extent can the two objec- tives be reconciled without substantial emphasis on development of an effective, indigenous administration? A comparison of the Tanzanian and Kenyan parastatals with the Ethiopian experience is instructive in this regard. In 1963 the Urambc Tobacco Scheme in Tanzania was taken over by the Tanganyika Agricultural Corporation from the Village Settlement Commis- sion (VSC). TAC was directly responsible to the Prime Minister's office.'8 The tobacco scheme in Tumbi, however, remained with BAT, a commercial organization geared to meet export demand. The yield differences between IJrambo and Tumbi and the absolute decline in to- bacco yields in Urambo subsequent to 1964 (see Chapter IV) seem to be ex- plained at least partially by the adverse effect of the takeover on Urambo's management. Though much of the staff of VSC was retained in the transfer, many of the decisions regarding project implementation were no longer made in Urambo but in Dar es Salaam. Less stringent criteria were used in the selection of tobacco farmers in Urambo, compared to Urambo's earlier criteria or compared to those of BAT. In contrast, BAT continued to use only trained and experienced farmers and weeded out those who did not perform well. Even after the Tabora Tobacco Growers Cooperative Society was registered in 1964, BAT remained largely in charge of manage- ment of the Tumbi scheme, while the company gradually trained the cooperative staff. Cornplete responsibility for the Tumbi scheme went to the cooperative staff only in 1969. In the BAT scheme there has also been stricter discipline enforced in repayment of credit than has been possible in the Urambo cooperative. 17. Nekby, the first director of CADU,has argued that more significant changes had begun to occur in the recent perioci. For instance, the appointment in 1974 of Ato Tesfa Bushen, then the Minister of State for Agriculture, as Governor-General of Arussi province, along with ad- vancements in cooperative organization were beginning to lead to an integration of CADU with the administration that existed in Chilalo at the time. In Nekby's view this indicated that autonomous, foreign-inspired projects can lead to viable rural development. Hunter in his assessment of CADU has argued, however, that the program's policy was not to develop local institutions but to bypass them altogether. In his view the program's paternalism towards in- digenous organizations has been its major shortcoming. This difference in perception is another example which points to the need for specification of more assessable criteria by which to judge institutional achievements, since they are so crucial for the realization of rural development objectives. 18. In 1965 VSC was dissolved and replaced by the Village Settlement Division (VSD) of the Ministry of Lands, Settlements and Water Development. 138 DESIGN OF RURAL DEVELOPMENT Observers of KTDA in Kenya attribute KTDA's administrative effec- tiveness and productive efficiency to similar strict discipline and little polit- ical interference in management. By the 1971/72 period KTDA had reached 66,500 smallholders. Since KTDA caters to small farmers-their average tea plot is about 0.4 hectares- it would appear to have been highly effective - in realizing the objective of broad participation in growth. However, the very discipline imposed on farmers-for example KTDA's emphasis on not providing credit for tea seedlings (so as to include only the motivated - farmers and to foster mobilization of local savings)-seems to have ex- cluded some of the poorest farmers from growing tea. Often they have neither the credit nor the financial ability to divert the best part of their small plot to tea cultivation or to bear the gestation lag in returns from tea. When the resource constraints are different between farmers, obviously there are some conflicts in growth and participation, particularly when the trained manpower available for devising alternative strategies is limited. An effective solution to these conflicts requires monitoring performance of programs from the point of view of participation and developing appropri- ate strategies over time so as to broaden the distribution of benefits. Instead of following such a systematic, sequential approach to participation, the tendency of an indigenous administration may often be either to neglect participation altogether, as has been the case in Ethiopia until recently, or to adopt a strategy participatory in style rather than in substance. The decline in tobacco production in Tabora, Tanzania, subsequent to IDA financing in 1970 illustrates the latter case. The decline was prompted by the com- mencement of a policy to move smallholder tobacco farmers into ujimaa villages and to collectivize tobacco cultivation. The government faced con- siderable resistance to villagization and collectivization from tobacco farm- ers, adversely affecting tobacco production as well as the distribution of benefits. As a result, the administrators had begun to show flexibility in the extent of collectivization and the rate at which it will be introduced.19 Direction and Coordination in the Transfer of Functions to the Regular Administration Transfer of the functions of the autonomous project authorities to the regular administration raises questions, not only of the possible conflicts between sociopolitical and economic objectives, but also about the logistics of the effective transfer of functions to ensure a high level of administrative performance. These questions are of particular importance because, unlike - project authorities administered by expatriates which turn towards the center of power in the capital city, it is the improvement of the lower levels of government administration and other institutions that is critical for the - long-run performance of the programs. To the extent that project authorities undertake functions which the ex- - isting administration is ill-suited or ill-prepared to carry out, transfer may involve at least a temporary, adverse effect on the performance of the proj- ect. If adequate attention is not given to bridging the gap between the proj- ect authority and the existing governmental structure at an early stage, the project's performance may be significantly affected in the long run. 19. See Chapter IX of this study. ADMINISTRATION: AUTONOMOUS PROJECTS 139 Integration involves administrative choices and improvements at several levels. First, there are the considerations related to overall direction and coordination, an important feature of the project authority. Second, there are questions related to the transfer of individual components carried out by the project. Some components-such as roads, soil conservation, boreholes, community development, health clinics, housing, and train- ing- involve transfer of responsibility to the regional or local government administration. Others (as, for instance, input and output marketing as well as credit distribution) involve transfer of responsibility of a commercial nature. Third, there are questions related to local participation and integra- tion of the local committee structure created by the project with the existing local government adrninistration. LLDP in Malawi is at such a transitional stage where the gaps between its accomplishments and the existing rural administrative capability in the region are substantial. In Lilongwe difficult problems of overall direction and coordination arise because the administrative units of the regular gov- ernment structure-namely, regional and district administrations and agri- cultural divisions-do not correspond to the units and sectors of the proj- ect.20 Though this has perhaps given the project a measure of freedom from entrenched bureaucratic groups, it has also raised such elementary ques- tions as (a) which levels of administrative units in the Ministry of Agricul- ture should be strengthened to perform which specific functions of policy and field administration coordination, and (b) the more difficult questions of how to implement these changes in administrative responsibility. Tradi- tionally, the region and the district have been the administrative units coor- dinating a number of technical and social service departments. Therefore, the headquarters of the district administration are also usually located in the same town as the headquarters of the district councils. The district, therefore, has several advantages in coordinating interdepartmental func- tions. In contrast, divisions, which are larger administrative units than the districts, have traditionally been the center of agricultural activities. Since agriculture constitutes the backbone of LLDP, divisions are the more ap- propriate coordinating point to reinforce staffing for improved planning and implementation of the agricultural activities. However, coordination of the project activities involves the resolution of policies as well as the admin- istration of field staff. Under the present administrative setup, the policy formulation could best be done at the regional level. In transferring individual components, many other problems arise. For example, coordination between the Ministries of Agriculture and Public Works is necessary during road construction. The maintenance of roads, on the other hand, has to be transferred to the appropriate body (i.e., the upkeep of major thoroughfares must be made the responsibility of the Min- istry of Public Works, while district and local roads should be transferred to the district and village councils). These transfers require strengthening the appropriate units with manpower and financial resources so that they may perform their functions satisfactorily. Health clinics, of course, involve coordination between district and village councils, the health ministry, and- if they involve any self-help- the Ministry of Community Develop- 20. In LLDP the unit is the basic administrative unit and is smaller than the agricultural divisions. It is viewed as a potential growth center consisting of forty to fifty villages. 140 DESIGN OF RURAL DEVELOPMENT ment. The same applies to women's extension which involves health, nutri- tion, child care, and so forth. Which of these various ministries or depart- ments should be assigned to pay staff salaries, supervise field administra- tion, or perform the central coordinating function are less important issues which need to be resolved. The transfer of commercial services is equally important. In Malawi the Agricultural Development and Marketing Corporation is responsible for in- put delivery and output marketing. However, as discussed in Chapter VI, the performance of ADMARC seems to have been less than satisfactory. The extent to which ADMARC will take over the major responsibility of ensuring the continued expansion of LLDP's delivery systems remains to be seen. Administration of credit after LLDP phases out is even more difficult. This is not only for the various reasons discussed earlier in Chapter V but also because, at present, there is no logical administrative organization that can take over the credit function. A more diversified and low-cost approach to provision of inputs and credit (i.e., an approach which develops local commercial institutions and involves group sales, commercial credit, pri- vate trading channels, and personal savings) still remains largely unex- plored. Organization of medium-term credit for purchase of capital goods will also need attention as investment demand increases. The role of the various committees created under the project authority and of their relationship with the more traditional district councils raises a different set of questions. The village-level committees are meant to be par- ticipatory bodies. Their purpose is to provide a forum for communicating new ideas, for identifying the needs and aspirations of the rural people, and for fostering self-help efforts. But, as pointed out in Chapter V, these com- mittees have at times been used for recovery of loans which promotes dis- sension between local groups. Guy Hunter, in his assessment of LLDP's ad- ministrative structure, therefore recommended that as far as possible local committees should not be used to serve objectives other than those for which they were formulated.2' The status of the higher level consultative committees vis-a-vis the district councils in planning, implementation, and maintenance of activities is also unclear at this stage. This is partly because LLDP's administrative units do not correspond with the regular local gov- ernment institutions. For example, there is no governmental equivalent to the project's unit centers which would constitute the growth nuclei. However, even more important, the role of the local leadership and its rela- tion to the political party or other authority that may be assigned to the various project and non-project committees for mobilization and allocation of resources remains to be specified. Concluding Statement The problems faced by parallel administrative structures are many and are admittedly complex. They include the hiring and training of national staff; establishing rapport with the indigenous institutions to improve effec- tiveness of administration; and, finally, the task of integration with the 21. Based on discussions with Hunter. ADMINISTRATION: AUTONOMOUS PROJECTS 141 regular governmental administration without jeopardizing the performance of the project. As has been pointed out in this chapter, without the resolu- tion of many of these problems, the gap between the initial promise of the integrated programs and their long-term performance can be substantial, despite the considerable initial capital and trained manpower input that they involve. Realizing the full potential of donor-aided programs must mean not only short-term production increases but also the development of effective rural administration and the concurrent training of indigenous manpower. To this enrd, far greater national involvement in planning and implementation of rural development programs is necessary than was possible in the autonomous projects reviewed under ARDS. However, if participation of nationals is to be meaningful in the long run and if pro- grams are to benefit from knowledge of local needs and local constraints, it is also essential that rural people and local-level officials, who possess such knowledge, play a major role in planning and implementing rural develop- ment programs. The improvement of administrative capacity and technical competence at the regional and lower levels is necessary to ensure that sociopolitical considerations do not become the exclusive criteria for the allocation of development resources. In the following chapter the alterna- tive of self-generated, internal improvement of existing administrations is discussed. CHAPTER IX Forms of Rural Development Administration 2: Nationally Planned Programs Objectives of national administrations tend to be broader than those of project authorities. In addition to increasing production and incomes, na- tional administrations may use rural development programs as a vehicle to broaden political participation, reinforce political patronage, or realize ideological objectives. These multiple and, at times, conflicting objectives have often rendered overall rural development strategy inconsistent and sometimes even self-defeating. Government planners, therefore, have the difficult job of reconciling the- needs for growth as well as for broadening participation and of finding a coherent rural development strategy which is politically feasible and economically sound and can be implemented admin- istratively. Such planning is by no means easy to realize. Priorities have to be set with regard to an appropriate production mix (i.e., between cropping com- binations, livestock activities, and small industrial enterprises) and with regard to an appropriate activity mix (i.e., a balance between productive and social services). The production mix must take into account the technical constraints outlined in Chapters 11 and III which vary widely from area to area. The activity mix must also be highly area specific and must depend upon the existing availability of productive facilities and upon the need and support for social services. Yet while the sheer complexity and diversity of local socioeconomic systems necessitate individually tailored development programs, this same complexity and diversity require considerable coor- dination between the many departments and agencies usually responsible for rural development. National planning, thus, has to reconcile two ap- parently conflicting but basically interdependent goals: (a) the need for de- centralization to take into account local potential and constraints and to channel the knowledge and energies of millions of rural people into the de- velopment process; and (b) the need for central control to foster national integration, regional specialization, and outside stimulus. In the past, national administrations have rarely been able to strike a proper balance between these twin goals. Centralized control of the rural ADMINISTRATION: NATIONALLY PLANNED PROGRAMS 143 development planning process has tended to predominate, often leading to what Chambers has called "planning without implementation."' The ten- dency has been to draw up broad sectoral plans that are little more than shopping lists and may bear little relation to local resource endowments. Because of this planning vacuum, there has frequently been "implementa- tion without planning,"2 i.e., self-help groups or activist administrators channeling resources into local programs that are at times in conflict with national objectives. What explains such uncoordinated planning? To some extent the centrist tendency has been necessitated by the shortage of administrative capacity in many African countries. The limited trained manpower has led to a con- centration of expertise at the center, inevitably weakening the capacity at the local level. Central control of the administration has often also been a result of the narrow political base of many of the governments and the cor- responding lack of national identity among large segments of the popula- tion. Frequently, these tendencies towards centralized administration have, of course, been reinforced by inadequate appreciation of the need for area- based planning. This chapter analyzes the performance of the planning and implementa- tion of the Special Rural Development Program (SRDP) in Kenya and of the ujdimaa movement in Tanzania. It discusses the efforts being made in these two countries to improve planning and implementing capacity, the nature of the problems encountered in increased national participation in planning and implementation, and the possible ways in which some of these problems could be dealt with in future rural development programs. The purpose of the discussion is to raise three questions: 1. To what extent have the existing administrative structures been able to carry out the development task effectively? 2. To the extent that there are deficiencies, how can these administrative structures be made more innovative and effective in initiating and ad- ministering the growing number of rural development programs? 3. How can rural development programs help improve administrative capacity? The Special Rural Development Program in Kenya Until recently, the administration of rural development programs in Kenya has been highly centralized. This has resulted in several problems in the design and implementation of various specific projects. First, the hierarchical structure of the various ministries involved in rural develop- -ment has resulted in most decisions having been made in Nairobi and 1. Robert Chambers, "Plianning for Rural Areas in East Africa: Experience and Prescrip- tions," in Rural Administration in Kenya, ed. David K. Leonard (Nairobi: East African Literature Bureau, 1973), pp. 15-18. 2. Ibid., pp. 18-19. For an interesting example of implementation without planning, see the discussion of the Kisii cattle dip program in Frank W. Holmquist, "Implementing Rural De- velopment Projects," in Developmel7t Adnministration: The Kenyaln Experienice, eds. Goran Hyden, Robert Jackson, and John Okumu (Nairobi: Oxford University Press, 1970). pp. 201-32. 144 DESIGN OF RURAL DEVELOPMENT passed down to the field staff. As a result, much of the rural development planning has been based on inadequate knowledge of local constraints. Sec- ond, different programs have been the responsibility of different agencies. Given the generally inadequate interministerial coordination, many ser- vices have been duplicated while others have been completely omit- ted.3 Third, while there has been general agreement that the government should pursue the course of African socialism, i.e., the "participation of all people in the task of nation building as well as in the enjoyment of the fruits of progress,"4 the specific means for achieving these goals have often been poorly articulated. Because no simple, clear-cut solutions existed to the problems outlined above, a conference was held in Kericho in 1966 on the interrelated issues of education, employment, and rural development. As an outgrowth of the proposals made at the much-debated Kericho conference, the government initiated the Special Rural Development Program to increase inter- ministerial cooperation and to bring about more realistic divisional plan- ning. As originally conceived in 1967, SRDP was to involve a number of comprehensive pilot programs which would pull together all aspects of rural development within a coordinated strategy. These programs were to be specific, concrete, and based upon detailed baseline surveys of the divi- sions involved. The decision-making locus was to be shifted away from the center. The special nature of SRDP lay in its capacity for testing the effect on rural development of various additional efforts over and above extant programs and policies.5 Since all of SRDP's programs were also to be under- taken with a view to their replicability elsewhere in Kenya, they were to be handled through the existing government machinery and with only minimal additional resources. While donors would be invited to participate, it was intended that they would form a consortium from which the program would allocate funds for the various divisional projects. As SRDP developed, it departed from its initial objectives. However, its accomplishments and shortcomings provide useful insights into the process of program planning and implementation and into the role of communica- tion and coordination (a) between various ministries, (b) between the civil servants and researchers involved in planning, (c) between the central and the local-level bureaucracy, (d) between the bureaucracy and the rural peo- ple, and (e) between national administrators and donors. 3. Hyden notes the proliferation of specialized civil service ministries in Kenya (there were twenty in 1969) and the consequent problem of interministerial coordination. The prolifera- tion of ministries is partly a result of a need to patronize large numbers of supporters in a polit-- ically heterogeneous society. See Goran Hyden, "Basic Civil Service Characteristics," in Devel- opment Administration: The Kenyan Experience, eds. Goran Hyden, Robert Jackson, and John Okumu (Nairobi: Oxford University Press, 1970), pp. 3-25. 4. Government of Kenya, African Socialism and Its Application to Planning in Kenya, Ses-- sional Paper no. 10 (Nairobi: Government Printers Office, 1965). 5. According to a statement approved by the National Rural Development Committee on January 23, 1970, the purpose of SRDP was "(a) to increase rural income and employment op- portunities and (b) to establish procedures and techniques for acceleration and self-generating rural development programs which can be repeated in other similar areas and, in particular, to improve the developmental capacity of Kenyan government officials in the field." It is this lat- ter purpose which has made SRDP unique. As the IDS study of SRDP concludes, it is "a truly remarkable and innovative program because it seeks to build the functions of Research and Development into the government machinery for rural development." Institute for Develop- ment Studies, An Overall Evaluation of the Special Rural Development Programme, 1972, Occa- sional Paper no. 8 (Nairobi: University of Nairobi, 1973), p. xii. ADMINISTRATION: NATIONALLY PLANNED PROGRAMS 145 Administrative Structure of SRDP Overall responsibility for the program was held by the Ministry of Fi- nance and Economic Planning (MFEP). In order to coordinate the program at the top, the National Rural Development Committee (NRDC) was orga- nized in 1969. This comcmittee was made up of the secretaries of MFEP, the Ministry of Agriculture, the Treasury, and the other ministries involved in rural development. Interdepartmental cooperation was to be facilitated by the establishment of a system of linkmen in the various ministries. When problems of coordination between ministries arose, the appropriate link- man-usually a senior officer in his ministry-was to be contacted and a decision worked out. T'he coordination of activities within the specific proj- ect area was the responsibility of the area coordinator, who was also an ad- ministrative officer in the provincial administration. In addition, the area coordinator was to serve as a communication link between the divisional, district, and provincial administrations. In order to provide input from local people, the District Development Committees (DDC) and the Provincial Development Committees (PDC) were brought into the decision-making process. NRDC laid down very specific procedures for the preparation of SRDP programs. First, a provincial planning officer (PPO) was to conduct a survey of the proposed project area to discover local constraints and potentials. Based on this data, the PPO, in consultation with the district heads of tech- nical departments, was to draw up a general project strategy. These plans were then to be forwarded to NRDC, where they would be evaluated with assistance from the concerned ministries. Approved plans would be returned to the PPO for clearance with DDC and PDC. After clearance had been received, detailed plans were to be drawn up and forwarded to NRDC. The committee in turn would secure the necessary approval from the departmental headquarters. Only when this process was successfully com- pleted would implementation begin- the sequencing and phasing of which were largely the responsiblity of the area coordinator. Coordination Between Ministries The objective in establishing NRDC was largely to bring sufficient prestige and attention to SRDP so as to facilitate the interministerial cooperation vital to the program's success. Such cooperation was not easy to achieve. While MFEP was generally supportive of SRDP, both the Treas- ury and the Ministry of Agriculture were ambivalent towards the program. NRDC lacked adequate staff and authority to resolve these tensions quickly. Thus, despite substantial prodding, NRDC found it difficult to convince the Treasury either of the novelty of SRDP's content or of the validity of its purpose.6 Consequently, the Treasury blocked the release of funds for the program in June 1970, despite the fact that the program had been discussed for three years at the national level and had been announced publicly. It became readily apparent that further delays in implementation might jeopardize the entire effort. The urge to press on led NRDC to aban- don its earlier positionl of seeking external assistance towards a central con- sortium in favor of allowing donors to fund specific geographic units. The 6. See J. R. Nellis, "The Administration of Rural Development in Kenya," in East Africa Journal, vol. 9, no. 3 (March 1972), pp. 10-17. 146 DESIGN OF RURAL DEVELOPMENT Norwegian Agency for International Development (NORAD) had been in- terested in backing SRDP in the Mbere district and received the green light from NRDC. FAO, SIDA, USAID, and other agencies were permitted to fund SRDP in other districts. While it enabled the program to get under way, this form of funding com- promised SRDP's original goals of experimentation and replicability. Some of the donors demanded replanning of the individual projects as a precondi- tion of their support.7 In addition, the donors offered technical as well as fi- nancial assistance, which raised more questions about the programs' replicability. It was only in January 1971, after the Treasury had been amalgamated with the Ministry of Economic Planning, that the government finally re- leased funds for SRDP. By this time, however, the external agencies were heavily involved in specific projects. Other problems have occurred with the linkman system in the various ministries, largely because ministry staff has tended to be skeptical of SRDP. Many of the linkmen have had only nominal functions and have often lacked experience and the authority within their respective ministries necessary to acquire resource commitments to the program. Nevertheless, progress has been made in smoothing interministerial coordination. A per- sonalized network has developed among the various linkmen, NRDC, and the area coordinators. This has helped circumvent the bureaucratic hierarchy and overcome delays. In addition, SRDP has developed a report- ing system to close the gap between project planning and implementation. Coordination Between Civil Servants and Researchers In order to prepare SRDP plans, extensive baseline surveys were carried out in the target districts by researchers of IDS in Nairobi.8 SRDP surveys were useful for understanding the characteristics of the regions studied and for pointing out that major economic breakthroughs of a technological miracle variety could not be expected and that the problems in rural devel- opment could be solved only by slow, steady, incremental change on a broad front.9 The planning, however, suffered from considerable delays. While mem- bers of IDS were insistent about the minimum data needs for planning, ad- ministrators in Nairobi became impatient that nothing useful was resulting from the surveys. The division of responsibility between the civil servants and researchers in the formation of strategy, as distinct from the conduct of baseline surveys, remained unclearly specified. As a result, the bureaucrats 7. For example, subsequent to its involvement, USAID replanned the program in the Vihiga district. According to IDS's evaluation, under American prodding the program was focused on the promotion of a single product- hybrid maize. Though an original objective of the program had been to involve all classes of farmers, the American-designed maize credit package was made available only to those farmers cultivating between 0.81 and 1.62 hectares of maize. Moreover, this particular crop is unlikely to be suitable for poorer and less skilled farmers. The IDS evaluation of the Vihiga program concluded that it had "lost most of its ex- perimental content..." and had become a straightforward development project. See An Overall Evaluation of the Special Rural Development Programme, 1972, Section F, p. 16. 8. J. Heyer, D. Ireri, and J. Moris, Rural Development in Kenya (Nairobi: East African Publishing House, 1971). Although the surveys were conducted in fourteen districts which were initially selected, only six districts had programs under way in 1971. 9. Ibid. ADMINISTRATION: NATIONALLY PLANNED PROGRAMS 147 felt that there was no emphasis on setting priorities, while the researchers, being cognizant of variability, emphasized principles of planning rather than specifying the content of individual plans. SRDP's planning experience, nevertheless, emphasizes