DISCUSSION PAPER Report No.: ARU 67 Farm/Non-Farm Linkages in Rural Sub-Saharan Africa: Empirical Evidence and Policy Implications by Steve Haggblade Peter Hazell James Brown Research Unit Agriculture and Rural Development Department Operational Policy Staff World Bank May 1987 The views presented here are those of the author(s), and they should not be interpreted as reflecting those of the World Bank. The authors are staff members and consultant of the World Bank. However, the World Bank does not accept responsibility for the views expressed herein which are those of the authors and should not be attributed to the World Bank or to its affiliated organizations. The findings, interpreta- tions, and conclusions are the results of research supported in part by the Bank; they do not necessarily represent official policy of the Bank. The designations employed and the presentation of material in this document are solely for the convenience of the reade: and do not imply the expression of any opinion whatsoever on the part of the World Bank or its affiliates concerning the legal status of any country, territory, area or of its authorities, or concerning the delimitation of its boundaries, or national affiliation. FARM/NON-FARM LINKAGES IN RURAL SUB-SAHARAN AFRICA: EMPIRICAL EVIDENCE AND POLICY IMPLICATIONS Steve Haggblade Peter Hazell James Brown Agriculture and Rural Development Department May 20, 1987 11 TABLE OF CONTENTS Page I. Introduction.................... . . . . 1 II. Profile of the Rural Non-Farm Sector....... . . . . . . . . . 4 A. Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . 4 B. Weight of Non-Agricultural Activities Africa's Rural Economy 6 1. Employment . . . . . . . . . . . . . . . . . . . . . . . . . 6 2. Income . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 3. Income distribution. . . . . . . . . . . . . . . . . . . . .12 C. Characteristics of Rural Non-Farm Enterprises (RNFE). . . . . .21 1. Size distribution. . . . . . . . . . . . . . . . . . . . . .21 2. Capital requirements . . . . . . . . . . . . . . . . . . . .26 3. Seasonality. . . . . . . . . . . . . . . . . . . . . . . . .28 4. Composition. . . . . . . . . . . . . . . . . . . . . . . . .31 5. Gender.. . . . . . . . . . . . . . . . . . . . . . . . . . .39 III. Structural Transformation of the Rural Non-Farm Economy. . . . . .41 A. An Empirical Profile of the Structural Transformation . . . . .41 1. Megatrends. . ............ . . . . . . . . ..41 2. State of the transformation........... . . . . ..50 3. Changes in rural nonfarm activity..............55 a. by size of locality . . . . . ..............55 b. by distance from urban cl.nters..............59 c. regional differences. . ....................61 d. changes over time . . . ..................65 e. conclusions............... .. . . . . . .6.1 4. Correlates of growth in the rural nonfarm economy. .... .71 B. Magnitude of the Farm-Nonfarm Linkages. . . . . . . ...... 3 1. Capital flows..83 a. investment of agricultural surplus in nonfarm ......83 activities b. investment of nonfarm surplus in agriculture. ......84 2. Labor flows. .........................................88 3. Backward linkages from agriculture to rural input......89 suppliers a. current backward linkages . . . . . . . . . . ...89 b. the future . . . . . . . . . . . . . . . . . . . .. . . 92 c. idiosyncratic design adaptation ..............93 4. Forward linkages from agriculture to processors. ..94 and distributors 5. Consumption linkages . . . . . . . . . . . . . . .97 6. A caution....................... . . . . 3110 111 TABLE OF CONTENTS (CONT.) Page C. Probable Magnitude of the Agriculture-to-RNFE Multipliers 113 1. Direct estimation................ . . . . . 113 2. Modeling agriculItural multipliers........ . . . . . 115 a. the model ...... 115 b. size of the multiplier... . . . . 118 c. estates versus smallholders and the size of the . 121 multiplier d. relative importance of consumption and....... . . 123 inter-industry linkages in the multiplier 3. Macroeconomic consequences and simulations...... . . . 125 IV. Constraints and Inducements to Growth of Rural.... . . . . . 127 Non-Farm Enterprises A. Demand. . . . . . . . . . . . . . . . . . . . . . . . . . . . 127 B. Supply Constraints. . . . . . . . . . . . . . . . . . . . . . 130 C. Existing Policies and Programs . . . . . . . . . . . . . . . 132 1. Policies . . . . . . . . . . . . . . . . . . . . . . ... . 133 2. Programs . . . . . . . . . . . . . . . . . . . . . . . . . 135 a. program profiles. . . . . . . . . . . . . . . . . . . . 135 b. project evaluations . . . . . . . . . . . . . . . . . . 142 V. Policy and Research Implications ... ..... 150 APPENDIX......................... . . . . . . 161 BIBLIOGRAPHY........... . . . . . .... . ..... . . . . 169 '4 iv LIST OF TABLES 1. Primary Employment in African Rural Nonfarm Enterprises .. .. ...7 2. Secondary Employment in African Rural Nonfarm Enterprises.. .. ..9 3. Share of Nonfarm Earnings in Total Rural Income .. .. .......10 4. Rural Income Distribution, Kano State Northern Nigeria .. ... ..13 5. Rural Income Distribution, Lesotho .. ... .......... ..14 6. The Effect of Women's Nonfarm Earnings on Hypothesized .. ... ..15 Rural Income Distribution, Zaria Rlegion Northern Nigeria 7. Rural Income Distribution, Botswana. .. ... ... ...... ..16 8. Rural Income Distribution, Sierra Leone. .. .. ..........17 9. Rural Income Distribution,, Kenya. .. ....... ........18 10. Frequency of Nonfarm Activity by Human Resource Level, Zambia. ..19 11. Distribution of Employment. in African Manufacturing, by Firm Size. 22 12. Size Distribution of African Rural Nonfarm Enterprises .. ... ..24 13. Composition of Workforce Employed in African RUral........25 Nonfarm Enterprises 14. Capital Investment in African Rural Nonfarm Enterprises. .. .....27 15. Composition of Capital Stock in Sierra Leonean Small Industry. .-29 16. Magnitude and Composition of African Rural Nonfarm Employment .. 32 17. Ac tivity Breakdown of Rural Manufacturing Employment. .. .....37 18. Female Participation in Rural Nonfarm Enterprises.........40 19. Degree of Association Beteen Rural Nonfarm Employment .. .. ...44 and Agriculture in Rural Kenya 20. Density and Composition of 'African Rural Nonfarm Employment, . 46 by Size of Locality 21. Composition or' African Rural Nonfarm Employment,.. ........47 by Size of Locality 22. Density of Rural Manufacturing and Service Employment,......48 by Size of Locality, Zambia 23. Density of Rural Manufacturing and'Service Employment,......49 by Size of Locality, Sierra Leor.-, 24. Spatial Distribution of Consumer Purchases in Rural Kenya......58 25. Spatial Distribution of Rural Nonfarm Employmen,.........60 Bouake Region Ivory Coast 26. Regional Distribution of Rural Nonfarm Employment, Kenya .. ....64 27. Trends in the Structure of Nonfarm Employment .. .. ...... ..67 Sierra Leone 1974-80 28. Trends in the Structure of Rural Nonfarm Employment,.......69 Rwanda 1978-83- 29. Sources of Start-up Capital in African Rural Nonfarm Enterprises .85 30. Sources of Working Capital in African Rural Nonfarm Enterprises. 86 31. Inventory, of Small Farmer Equipment in Sub-Saharan Africa. .. ....91 32. Probable Relative Magnitude of Backward and Forward Agricultural .95 Linkages in Rural Africa 33. Rural Household Expenditure Behavior in Selected Countries . . . .100 V LIST OF TABLES (CONT.) Page 34. Marginal Budget Shares by Per Capita Expenditure Decile .. .. ...103 in Muda, 1972 35. Marginal Budget Shares by Per Capita Expenditure Decile .. .. ...104 in Gusau, 1976 36. Marginal Propensity to Consume by Income Class .. ..... ...108 in Rural Sierra Leone 37. Farm Household Gross Cap.ital Formation, Ondo State Nigeria . . . .112 38. Estimated Agricultural Growth Multipliers . . . . .. .. .. .. .120 Given Varying Parameter Estimates 39. Excess Capacity in African Nonfarm Enterprises .. ..... ...129 40. Principal Business Problems as Assessed .. .. ........ ..131 by Rural Nonfarm Entrepreneurs 41. The Differential. Impact of Policy-Induced Factor Price. .. ....136 Distortions on Large and Small Nonagricultural Enterprises 42. Summary of Rural Nonfarm Program Interventions .. ..... ...140 in Sub-Saharan'Africa 43. Criteria for Comparing Alternative Sorghum Beer. .. ... .....149 Production/Distribution Channels in Botswana APPENDIX TABLES PaQe A.1 RNFE Correlations, Basic Data .. .... .............162 A.2 Regional Variation in Farm and Nonfarm Income.. .. ... ....163 Togo arid Sierra Leone A-3 Review of Rural African 'Consumption Studies .. ..... .....164 A.4 Estimated Range of Agriculture's Rural Value Added Share .. .. ..165 in Gross Output A.5 Estimated Range of Nonfarm Rural Value Added Share. .. ... ....7 in Gross Output A.6 Estimates of Multiplier Parameters Based on Regional .. .. ....168 Input-Output Tables for Two Regions of Ivory Coast 34. Vi LIST OF FIGURES Figure 1. Seasonal Patterns of Labor Use in Farm and Nonfarm Activities. ............ . . . . . . . . . 30 Figure 2. Evolution of Nonfarm Enterprises in the Rural Economy. . . . 42 Figure 3. interactions between Locality Size and Location in the . . . 63 Composition of RNFE, Northern Nigeria Figure 4. Interactions between Locality Size and Location in the . . . 64 Composition of RNFE, Ivory Coast Figure 5. Correlation between RNFE and Rural Population Density, . . . 78 Figure 6. Correlation between RNFE and GDP Per Capita..... . . . . 79 Figure 7. Correlation between RNFE and Agricultural Income Per Capita. 80 Figure 8. Cross Regional Corrlation between Rural Nonfarm and. . . . . 82 Agricultural Income, Togo Figure 9. Cross Regional Correlation between Rural Aonfarm and . . . . 83 Agricultural Income, Sierra Leone Figure 10. Marginal Budget Shares of Locally Produced Goods and. . .108 Services, by Per Capita Income Expenditure Decile, Sierra Leone Figure 11. Schematic of Botswana's Sorghum Beer...... . . . . .. 149 Production/Distribution System LIST OF BOXES Box 1. Efficiency Gains from Specialization in Rural Sudan . . . . . . 55 Box 2. Gari Processing in Ghana: The Role of Rural Towns and Infrastructure in the Development of Rural Nonfarm Enterprises. 74 Box 3. Botswana's Sorghum Beer Industry: Subsector Analysis. . . . . .147 in Action 1 I. INTRODUCTION1 Based largely on evidence from Asia, Johnston, Kilby, and Mellor advocate a rurally oriented development strategy driven by growth in small farmer agriculture (Johnston and Kilby, 1975; Mellor, 1976; Mellor *and Johnston, 1984). They argue that a small farmer strategy will generate rapid, equitable and geographically dispersed growth because of substantial, labor intensive linkages with the rural nonfarm economy. Using examples from India, Pakistan, and Taiwan, Johnston and Kilby (1975) highlight the potential importance of production linkages. In addition to farmer demand for fertilizer and construction inputs, they emphasizp the importance, for the rural economy, of other backward linkages from small farm agriculture to local blacksmiths and equipment suppli2rs. We would like to thank the many individuals who assisted us at various stages in producing this report. Ram Agarwala, Dennis Anderson, Hans Binswanger, Eric Crawford, Bill Jaeger, Peter Kilby, Carl Liedholm, Peter Matlon, John Page and Thelma Triche all assisted in identifying promising veins of both data and relevant analysis. We likewise benefitted early on from discussions with Stanley Please, Ram Agarwala and the entire convened staff of the Special Office for Africa (SOA). Because the data demands of this study were considerable, we are particularly grateful to Yacob Fisseha, Peter Kilby, Carl Liedholm and Chris Windheuser, all of whom made special efforts to assist us in procuring elusive data. In helping sift and shape a large body of information, Nandinee Kutty and Kwesi Amissah provided invaluable research assistance. Carl Liedholm and Vijay Vyas both pored over our ponderous initial draft, offering incisive commentary and helping us to clarify our presentation and thinking in countless ways. To all we issue our thanks as well as absolution from any implication in remaining errors of fact or interpretation. We alone shoulder responsibility for whdt remains. 2 But farmers also purchase consumer goods. Mellor (1976) and Mellor & Lele (1973), in particular, have called attention to the potential power of agricultural consumption linkages. Drawing on Indian data, they conclude that middle-sized peasant farmers - to a much greater extent than their large scale and urban counterparts - spend incremental income on labor- intensive rurally produced goods, thereby generating important second-round demand multipliers. Hirschman's (1959) early indictment of agriculture as a low-linkage, underpowered engine of growth erred, according to Mellor (1976), because it ignored these important agricultural -consumption linkages. Several empirical studies have documented the power of farm-nonfarm linkages in Asia. Based on data from India, Rangarajan (1982), Mellor and Lele (1973), and Mellor and Johnston (1984) estimate economy wide agriculture-to-nonfarm income multipliers to be in the range of about .7. And Bell, Hazell and Slade (1982), examining exclusively rural growth linkages in the Muda River Region of Malaysia, compute an agriculturally induced rural nonfarm income multiplier of .8. Employment multipliers are also substantial. Evidence from the Philippines (Gibb, 1974) and India (Krishna, 1976; Mellor and Mudahar, 1974) places agriculture-to-nonfarm employment elasticities between 1 and 1.3. Although decomposition of the growth multipliers is rare, Bell, Hazell, and Slade (1985), in their Muda study, do provide a breakdown, estimating the consumer demand generated by agricultural growth to be double the production linkages. They provide, thus, the-major empirical testament to the relative power of consumption linkages. 3 As the world turns its attention to Sub-Saharan Africa, government leaders and donors alike view small farmer agriculture as the necessary centerpiece of development attention. Equity, nutrition and poverty considerations argue persuasively for such a focus. In addition, many would like to believe that a small farmer strategy will generate maximum growth rates, Asia-style, through linkage multipliers with the rural economy. But no one has yet marshalled the empirical evidence necessary for assessing the power of agricultural growth linkages in Africa. It may now be possible to do so given the accelerating accumulation, over the past 20 years, of a body of detailed survey work examining the structure of Africa's rural nonfarm economy (Kilby, 1962, Liedholm, 1973; Steel, 1977; Chuta and Liedholm, 1979; Page, 1979; Anderson and Leiserson, 1980; Page and Steel, 1984; Liedholm and Kilby, 1986; Liedholm and Mead, 1987). This paper aims to examine that evidence in order to assess the nature and magnitude of farm-noifarm links in rural Sub-Saharan Africa. First, the paper will systematically review empirical evidence on the structure of the African rural nonfarm economy. It will then examine changes over time, as well as differences across countries, in an effort to assess quantitatively the magnitude of the agricultural growth multipliers in Africa. The paper concludes with a brief review of policies and programs that will be necessary in achieving maximum farm-nonfarm growth linkages. 4 II. PROFILE OF THE RURAL NONFARM SECTOR A. Definitions Before embarking on a review of rural nonfarm activities, it is helpful to specify what we mean by both "rural" and "nonfarm". Nonfarm activities are easier to pinpoint. They are taken to include all economic activity apart from crop and livestock production. Thus nonfarm activities encompass services, construction, mining, commerce, and manufacturing. They include agroindustries which transport, process, package Qr distribute agricultural commodities. Some studies make a different distinction - between "on-farm" and "off- farm" activities. Analysis of this type includes in off-farm work all economic endeavor a person engages in outside the confines of his or her own fields. Thus "off-farm" activities include agricultural wage labor on other peoples' farms, wage labor in manufacturing, commerce or services, as well as any owner operated nonagricultural activities. This review is primarily concerned with the conceptual distinction between agricultural and nonagricultural activities. Consequently, where possible in using these studies, we separate out the nonfarm, that is the nonagricultural, from the agricultural component of off-farm work. Conceptions and definitions of rural vary dramatically, as Anderson and Leiserson (1980), in particular, have emphasized. Certainly the transition from rural to urban traverses a continuum of settlement patterns, population and functional densities, making any attempt to partition the universe necessarily arbitrary. Although the United Nations has defined as 5 "rural" any settlement under 20,000 in population (United Nations, 1969), few African countries have adopted that standard. Many include a mix of locality size and economic characteristics in their official definitions leading to a range of locality size cutoffs ranging from as small as 150 person settlements in Zimbabwe's 1969 census definition to implicit cutoffs as high as about 10,000 person agglomerations in Mauritania and Benin. As the notes to Table 16 indicate, official African government definitions of rural normally include localities up to about 3,000 to 5,000 in population. In practical terms, the definition is important, because the density of nonfarm activities changes dramatically with size of locality. So arbitrary differences in definitions of rural can lead to measurement of apparently widely different nonfarm employment densities, even in identical economies. Conceptually, we prefer to think of rural as depending on both the function and size of a locality. We will consider as rural any locality that exists primarily to service an agricultural hinterland. Urban areas, in contrast, operate economies that are driven by manufacturing, government or some other motor independent of agriculture. Given this view, rural areas can include rural towns of substantial size, perhaps as large as 50,000. Admittedly this definition is difficult to operationalize across a wide number of countries, as it requires intimate knowledge of the regional economies of each. And in most cases, we are prisoners of the rural-urban delineations made by those who have collected the economic data of interest in this review. In order to deal with this problem and also to emphasize the spatial features that are so important to,the development of the rural 6 nonfarm economy, whenever possible we will provide analyses which highlight differences in nonfarm activities across locality size and region. B. Weight of Non-Agricultural Activities in Africa's Rural Economy 1. Employment Respecting individual country definitions of "rural", Table 1 indicates that rural nonfarm enterprises provide the primary source of employment for between 3 and 73% of the labor force in rural Sub-Saharan Africa. Although the range is wide, the nonfarm shares of.rural employment typically fall in the 10-20% range. As a preliminary indicator, these percentages appear lower than the 20-30% figures reported elsewhere for Asia (Chuta and Liedholm, 1979). But as many observers have noted, figures from both continents should be viewed as minimum estimates, because they frequently exclude rural towns (Anderson and Leiserson, 1980; Chuta and Liedholm, 1979). The wide range in African rural nonfarm employment shares arises not only from differing definitions of rural or from different rural economic structure, but also because of different perceptions of female workforce participation. Thus Western Nigeria and Benin's shares of nonfarm activity in rural employment attain 73% and 41%, because survey designers in those countries have classified 75% to 85% of active women as working primarily outside of agriculture, for the most part, as we will see later, in commerce. Chad, a Muslim society, represents the opposite extreme. Women appear to participate in little but agriculture. Given the large women's nonfarm earnings measured by female researchers in Muslim Northern Nigeria 7 TABLE 1 PRIMARY EMPLOYMENT IN AFRICAN RURAL NONFARM ENTERPRISES (RNFE) Percent of Rurala Labor Force Employed in RNFE Countries Total Male Female Benin (1961) 41 10 77 Cameroon (1976) 10 ---- Chad (1964) 3 4 .1 Ghana (1960) 27 13 89 Ivory Coast, Bouake Region (1963) 10 -- -- Kenya (1970) 28 -- Malawi (1977) 11 17 4 Mali (1976) 8 -- -- Mauritania (1977) 21 -- -- Mozambique (1963) 9 19 1 Nigeria, 3-District W. State (1966) 73 19 87 Rwanda (1978) 5 -- -- Senegal (1970-71) 18 -- -- Sierra Leone (1974, males 1976) 14 19 -- Tanzania (1978) 6 10 2 Togo (1970) 26 -- -- Uganda, 4 Villages (1967) 20 -- -- Zimbabwe (1982) 20 -- -- aSee Table 16 for details of individual country definitions of rural. SOURCE: Benin: Republique du Dahomey (1964). Cameroon: Republique Unie du Cameroun (19xx). Chad: Republique du Tchad (1966). Ghana: Ghana (1964). Ivory Coast: Republique de C6te d'Ivoire (1965a). Kenya: I.L.O. (1972), cited in Chuta and Liedholm (1979). Malawi: Malawi Government (1980). Mali: Republique du Mali (1980). Mauritania: Republique de Mauritanie (1979). Mozambique: Republica Popular de Mozambique (1983). Nigeria: Mueller and Zevering (1970). Rwanda: Republique Rwandaise (1978). Senegal: Republique du Senegal (1973). Sierra Leone: Male 1976 from Byerlee, et al. (1977); 1974 Aggregate from Thomas (1983). Tanzania: United Republic of Tanzania (1982). Togo: Republique Togolaise (1974). Uganda: Brandt, et al. (1972). Zimbabwe: Zimbabwe (1985). 8 (Simmons, 1976a), it is. possible that the very low Chad figure stems not only from different economic participation rates but also, in part, from reporting to enumerators socially accepted perceptions of the appropriate role of women. Thus the role of women matters a great deal in assessing the magnitude of rural nonfarm employment. As Table 2 shows, seasonal and part-time nonfarm activities employ a further 4 to 20% of th-e rural male workers and as much as 30% of the total rural labor force. Looking exclusively at agricultural households, farm management surveys in numerous African countries have found 135% to 65% of farmers with secondary employment in nonfarm enterprises. In a similar vein, labor time allocation studies of farm households have found 15% to 40% of total family working hours devoted to income generating nonfarm. activities (Cleave, 1974; Anthony et.al., 1979; Norman, 1972; Hill, 1977; Luning, 1967; Okafor, 1983). 2. Income Income measures the importance of rural nonfarm activities in a way that takes account of both seasonal and part-time nature of much of the activity. As Table 3 shows, between 25% and 30% of African rural income typically comes from nonfarm sources. And because nonfarm activities are monetized to a much larger extent than agricultural production, nonagricultural earnings constitute an even larger share of cash income. Panel B of Table 3 indicates that nonfarm income regularly accounts for 30 to 50% of rural cash incomes. 9 TABLE 2 SECONDARY EMPLOYMENT IN AFRICAN RURAL NONFARM ENTERPRISES (NFE) Percent of Rural Population with Secondary Employment in NFE Countries Total Males Females Benin (1961) 7 -- -- Chad (1964) -- 4 -- Western Nigeria, Three States (1966) -- 20 -- Sierra Leone (1976) -- 11 -- Tanzania (1980) 17 -- -- Zambia, 2 Regions 30 -- -- -- = not available SOURCE: Benin: Republique du Dahomey (1964). Chad: Republique du Tchad (1966). W. Nigeria: Mueller and Zevering (1970). Sierra Leone: Byerlee, et al. (1977). Tanzania: Collier et.al. (186). Zambia: Hedlund and Landahl (1983). 10 TABLE 3 SHARE OF NONFARM EARNINGS IN TOTAL RURAL INCOME IN SUBSAHARAN AFRICA (percent) Agricultural Nonfarm Country Income Income Total A. Total Rural Income Botswana (1974-75) 56-76 24-44 100 Ghana, Akwapim (1965-66) 58-72 14-28 100 Ghana, Eastern Region (1967-68) 58-82 18-42 100 Ghana, Heads of Household in E. Region 25-50 50-75 100 Ghana, Five Villages 70-84 16-30 100 Ivory Coast, South East Region (1963) 71 29 100 Kenya (1969) 69 31 100 Kenya (1976) 72 28 100 Lesotho (1977-78) 23 77a 100 Northern Nigeria, Kano Region 72 28 100 Northern Nigeria, Zaria Region 78 22 100 Sierra Leone (1974) 64 36 100 Sierra Leone (1975-76) 81 19 100 Tanzania (1969) 79 21 100 Tanzania (1975) 77 23 100 B. Rural Cash Income Ghana, Akwapim (1965-66) 55-7& 22-45 100 Ghana, Cocoa Farmers (1968-70) 77 23 100 Ghana, Eastern Region (1967-68) 45-68 32-55 100 Ghana, Heads of Household in E. Region 17-45 55-83 100 Ivory Coast, Bouake Region (1970) 58-76 24-42 100 Liberia, Western Region (1973) 69 31 100 Tanzania (1969) 64 36 100 Tanzania (1975) 57 43 100 Zambia, Luapuala Province (1977) 79 21 100 Zambia, Three Provinces (1982) . 76 24 100 aRemittances account for 2/3 of rural income. Domestic nonfarm income is 12 percent of total or 34 percent of domestic rural value added. 11 Table 3 (CONT.) SOURCES: Botswana: Chernichovsky et al. (1985) p. 51. Ghana, Akwapim: Dutta-Roy & Mabey, Household Budget Survey in Ghana, cited in ILO (1982b) p. 51. Ghana, Cocoa Farmers: Buxton, Cash Income of Cocoa Farmers in Ghana, cited in ILO (1982b) p. 57. Ghana, E. Region: Dutta-Roy, Eastern Region Household Budget Survey in Ghana, cited in ILO (1982b) p. 53. Ghana, Household heads: Ibid. p. 54. Ghana, Five Villages: Boateng and Wukunya, Weija Dam Project, cited in ILO (1985d) p. 36. Ivory Coast: Republique de C8te d'Ivoire (1965a). Kenya:- ILO (1982c). Lesotho: National Accounts, cited in ILO (1982d) p. 16. Liberia: Ministry of Planning and Economic Affairs (Liberia), cited in ILO (1982) p. 58. Nigeria, Kano Region: Matlon (1977); Zaria Region: Norman, Pryor and Gibbs (1979). Sierra Leone: Chuta and Liedholm.(1979). Sierra Leone: ILO (1981c) p. 278. Tanzania: ILO (1982a). Uganda: Massell and Parnes (1969) p. 315. Zambia, Luapaula: Maimbo (Zambia), cited in ILO (1982q). Zambia, Three Provinces: Due and Mindenda (1985). 12 3. Income Distribution In Asia, distributional considerations appear to magnify the importance of rural nonfarm income. Kilby and Liedholm (1986), in a recent review, show that in Korea, Taiwan, and Thailand, nonfarm income accounts for a larger share of total income for landless households than it does for large landholders. And for a more limited group of countries, the authors find the same J-curve across income, as opposed to land holding, distributions. The J-shaped distributions lead to equity embellished arguments for promoting rural nonfarm activities. The available Africa data, displayed in Tables 4 through 10, do not show such a clear pattern of equity enhancement from rural nonfarm earnings. They show, in the cases of Northern Nigeria and Lesotho, clearly an opposite trend. Higher income groups derive a greater share of their earnings from nonfarm sources than do the poor. Less detailed evidence from rural Tanzania (Collier, Radwain and Wangwe, 1986) and from farm families in Zambia (Ghai and Radwan, 1984), Uganda (ILO, 1985g) and Zimbabwe (Weinrich, 1970) suggest similarly inequitable trends. Support for the equity enhancing view of rural nonfarm income comes from rural Botswana and Zaria Region, Northern Nigeria. Displayed in Tables 6 and 7, these data indicate that poor households depend more heavily on nunfarm incomes than do the wealthy, a tendancy also identified among farm households in Gambia (ILO, 1985d). Simmons' synthetic data from Zaria Region, displayed in Table 6, offer tantalizing insights into the distributional importance of women's nonfarm earnings. These data compare income distribution profiles before and after 13 TABLE 4 RURAL INCOME DISTRIBUTION, KANO STATE, NORTHERN NIGERIA (THREE VILLAGES) (in percentages) Income Deciles Economic Activit f 1 2 3-4 5-6 7-8 9 10 Agriculture Own Farm 80 75 77 77 75 60 63 Wages 8 4 4 5 1 4 10 Total Agriculture 88 79 81 82 74 64 64 Nonagriculture Nonagricultural Income 12 21 19 18 26 36 36 Grand Total 100 100 100 100 100 100 100 .SOURCE: Malton, Income Distribution among Farmers in N. Nigeria, p. 44. 14 TABLE 5 LESOTHO, RURAL INCOME DISTRIBUTION, THABA TSEKA-PHUTHIATSANA PROJECT, 1975/76 (in percentages) Income Categories Economic Activity 0 - 200 200 - 600 600 - 1,000 1,000+ Agriculture Crops 39 16 3 4 Livestock 30 21 6 12 Total Agriculture 69 37 9 16 Nonagricultural Remittances 8 53 81 71 Nonfarm 23 10 9 13 Total Nonagricultural 31 63 90 84 Grand Total 100 100 100 100 SOURCE: Van Der Wiel, cited in ILO (19821) p. 46. 15 TABLE 6 THE EFFECT OF WOMEN'S NONFARM EARNINGS ON HYPOTHESIZEDa RURAL INCOME DISTRIBUTION, ZARIA REGION, NORTHERN NIGERIA, 1970/71 (percent) Dan Mahawayi Village Hanwa Village Low Middle High Low Middle High Income Income Income Income Income Income A. Income, Excluding Women's Off-farm Earnings Farm Income 74 83 76 89 84 68 Off-farm Income Male adults 26 27 23 11 16 32 Total Without 100 100 100 100 100 100 Estimated Women's Earnings B. Income, Including Estimated Women's Off-farm Earnings Farm Income 50 66 64 42 49 56 Off-farm Income Male adults 17 19 20 5 9 26 Female adults 33 16 16 52 42 19 Total off-farm 50 34 36 57 51 44 Total "Household" 100 100 100 100 100 100 Income aOriginal household survey data did not include female off-farm earnings because of difficulties in obtaining such data. Women's nonfarm earnings were estimated in a later separate survey effort and the two data sets consolidated to produce hypothesized total "household" incomes. SOURCE: Norman, Simmons and Hays (1982). 16 TABLE 7 BOTSWANA RURAL INCOME DISTRIBUTION, 1974/75 (percent of rural income) Income Categories 0.5 - 10 15 - 50 60 - 95 97 - 99.7 Economic Activity Percent Percent Percent Percent Agriculture Crops 6 9 4 0 Livestock 5 7 32 64 Gathering & Hunting 18 8 2 2 Total Agriculture 77 ~7" 7u 716 Nonagricultural Manufacturing 5 8 1 0 Services 3 1 1 1 Trade 0 2 2 22 H. Benefits 10 7 3 2 Transfers 21 14 4 0 Property Income 0 0 10 0 Additional Income 17 10 7 6 Taxes 0 -2 -2 -7 Total Nonagricultural 55 40 26 24 Not Classified Wage Employment 15 36 36 9 Grand Total 100 100 100 100 SOURCE: Botswana (1976). "The Rural Income Distribution Survey in Botswana 1974-75," Central Statistics Office. 17 TABLE 8 SIERRA LEONE RURAL INCOME DISTRIBUTION, 1975 (percent of rural income) Income Terciles Low Middle High Agricultural Income Annual Crops 58 63 61 Tree Crops 22 18 19 Livestock .4 .2 .1 Total Agriculture 80 81 80 Nonagricultural Income Small Scale Industry 5 4 7 Hunting, gathering, 2 8 9 fishing Tradinga 2 1 1 Total Nonagriculture 7TT 7 Not Classified Hired Labor 11 6 3 Grand Total 100 100 100 aSurvey excludes households whose primary source of income is trading. SOURCE: Matlon et.al. (1979). TABLE 9 KENYA, RURAL INCOME DISTRIBUTION, 1974-75 (in percentages) Income (in shillings) 8,000 and Income Sources 0 - 9,000 1,000 - 1,999 2,000 - 2,999 3,000 - 3,999 4,000 - 5,999 6,000 - 7,999 Over Farm Operating Surplus 23 44 53 56 61 61 64 Nonfarm Operating Surplus 16 11 10 12 12 6 11 Remittances from Relatives 29 18 16 11 9 6 3 Other Gifts 3 3 2 1 3 2 1 Total nonfarm 48 32 28 24 24 14 15 Regular Erployment 8 19 5 13 9 19 19 Casual Employnent 21 15 14 6 6 6 2 Total unclassified 29 25 19 19 15 25 Total 100 100 100 100 100 100 100 SOURCE: Kenya, Statistical Abstract (1976) p. 139. 19 TABLE 10 FREQUENCY OF NONFARM ACTIVITY BY HUMAN RESOURCE LEVEL, ZAMBIA, 1976: PERCENT OF HOUSEHOLDS EARNING SOME INCOME FROM EACH ACTIVITY Households Groups, Classified by Resource Level 1 2 3 4 Slightly Above Above Activity Disadvantaged Subsistence Average Average All Brewing 33 37 32 17 33 Fishing 25 21 7 5 19 Charcoal Burning 4 5 1 -- 4 Woodcarving 5 12 5 3 7 Brickmaking 4 12 16 8 9 Trading 5 5 7 17 6 Store Owner -- -- 3 14 2 Wage Employment 8 6 24 17 10 Total 84 98 95 81 90 SOURCE: Marter and Honeybone (1976), p.43. 20 estimated corrections for missing female nonfarm earnings," originally omitted because of the difficulty of obtaining responses to questions about women's earnings in that Muslim region. The resulting income profiles suggest that women's nonfarm earnings are not only substantial but also most important among low income households. It is the women's earnings that turn the nonfarm contributions to household income larger in low than in high income families. In addition to highlighting the distributional importance of women's nonfarm earnings in Northern Nigeria, this simulated finding casts doubt on the direction of the Kano Region rural nonfarm income profiles in Table 4, since those include only male nonfa.rm earnings (Matlon, 1977). The Kenya and Sierra Leone data, presented in Tables 8 and 9, do not permit a definitive judgement about the income distribution effects of rural nonfarm income, primarily because they lump together all "wage income," combining both agricultural and non-agricultural wage earnings. But they, along with the Zambia data in Table 10 and the Botswana evidence in Table 7, do highlight important income distribution differences both among sources of nonfarm income and across regions. In the Southern African countries of Zambia and Botswana, income from trading and commerce accrues largely to wealthy rural households. That finding is consistent with popular images of wealthy general store owners in those countries. The apparently opposite result in Sierra Leone stems from the omission of predominantly trading households from the rural sample selection in that study (Matlon et.al., 1979). 21 Manufactur'ing income, in both rural Botswana and rural Zambia, does work in favor of income equality, providing a larger share of income for poor than rich. In both countries, female dominated food processing - about 80% of it sorghum and maize beer brewing - accounts for the bulk of this manufacturing income (Botswana, 1976; Milimo and Fisseha, 1986). But in Sierra Leone, where indigenous beer is not brewed, manufacturing income accrues primarily to the wealthy. Once again, female dominated activities swing the distributional balance. While blanket equity arguments supporting promotion of rural nonfarm activities cannot be imported from Asia to Africa, some activities - frequently female dominated food preparation - do appear to enhance income equality. And even if the impact on relative inequality is uncertain or negative, in terms of absolute poverty alleviation nonfarm earnings cannot but help improve the economic status of the poor. C. Characteristics of Rural Non-Farm Enterprises (RNFE) 1. Size distribution The overwhelming majority of rural nonfarm enterprises are extremely small. In fact, studies of manufacturing - the most closely studied segment of the nonfarm economy - uniformly find the great majority of total rural and urban employment in firms employing less than 10 workers. Table 11 indicates that firms with less than 10 workers employ between 60 and 95% of the entire manufacturing labor force in the half dozen countries for which evidence is available. 22 TABLE 11 DISTRIBUTION OF EMPLOYMENT IN AFRICAN .MANUFACTURING, BY FIRM SIZE (percent) Firm Size (number of workers) Large Scale Small Scale 50 or More 10-49 -Below 10 Country and Date Engaged Engaged Engaged a Ghana (1970) 15 1 84a Kenya (1969) 41 _10 49 Nigeria (1972) 15 26 59a Sierra Leone (1974) 5 5 90 Tanzania (1967) 37 7 56a Zambia-1985 16 1 83 aComputed as a residual, the difference between employment recorded in labor force or population surveys and establishment surveys. SOURCE: Chuta and Liedholm (1979); Page (1979); Milimo and Fisseha (1985). 23 And rural firms are so small many analysts refer to them as "micro- enterprises." As Table 12 shows, average employment in rural nonfarm enterprises regularly averages less than two, the owner plus one additional worker. In countries such as Burkina Faso, Ghana, Nigeria, Sierra Leone and Zambia, one-person enterprises constitute about half of all nonfarm enterprises; and well over 95% employ five people or less. Although differences are not large, the limited available evidence - from Ghana, Sierra Leone, and Zambia - does suggest that manufacturing enterprises tend to be slightly larger than other nonfarm activities. Trading establishment follow, with service establishments being the smallest (Thomi and Yankton,1985; Liedholm and Chuta, 1976; Milimo and Fisseha, 1986). Large scale firms establishing operations in rural areas are frequently limited to major trading establishments - either large distributors or wholesalers of major agricultural commodities - and large resource-based manufauturers. Large rural manufacturers generally confine their operations to weight losing production processes - such as lumbering, sugar production and oil extraction - or the preliminary processing of perishable commodities such as tea, cocoa and cqffee. Studies in Ghana (Andrae, 1981), Kenya (Freeman and Norcliffe, 1985), and Zaire (Congo, 1968) cle*arly point to the concentration of large-scale rural manufacturing in lumbering, mining, and food processing. In West and Central Africaj rural nonfarm establishments employ primarily family laborers and apprentices. As Table 13 indicates, hired workers from outside the family comprise less than 20% of nonfarm 24 TABLE 12 SIZE DISTRIBUTION OF AFRICAN RURAL NONFARM ENTERPRISES (percent of firms) Firm Size, by Employment Average Employment a Per Enterprise 1 2-5 6+ Burkina Faso, Eastern Province (1978) 1.8 56 Ghana, Rural Towns (1984) n.a. 46 37 17 Kenya, Central Province (1983) 1.8 -- -- -- Kenya', Meru District (1978)b 2.8 36 43 10 Kenya, b- Bangowa District (1978) 3.0 22 69 9 Sierra Leone (1974)c 1.6 42 53 5 Zambia, Six Ruralc Provinces (1984) 1.6 68 30 1 alncludes owner operator. bInformal sector only, but including establishments such as sawmills employing 20 people on average. CIncludes only manufacturing and repair enterprises employing less than 50 workers. SOURCES: Burkina: Wilcock and Chuta (1982). Ghana: Yankson (1985). Kenya, C6ntral Province: Freeman and Norcliffe (1985). Kenya, Meru and Bangowa Districts: Williams and McClintock (1981). Sierra Leone: Liedholm and Chuta (1976). Zambia: Milimo and Fisseha (1985). 25 TABLE 13 COMPOSITION OF WOR(FORCE EMPLOYED IN AFRICAN RURAL NONFARM ENTERPRISES (percent of rural nonfarm employment) Owner Plus Family Hired Coops and Workers Apprentices Workers Others Total BurKina Faso, Eastern Region (1980) 95 3 2 0 100 Kenya, Meru District (1978) 30 3 63 0 100 Kenya, Bangowa District (1978) 46 3 53 0 100 Malawi (1977) 34 0 66 0 100 Mozambique (1980) 20 -- 75 5 100 Nigeria, Western Region (1970) 39 50 11 0 100 Mid-West Region (1971) 30 63 7 0 100 Rwanda (1978) 29 10 69 0 100 Tanzania (1978) 11 0 84 0 100 Zambia (1985) 92 5 3 0 100 SOURCE: Burkina: Wilcock and Chuta (1981). Kenya: Williams and McClintock (1981). Malawi: Republique du Mali (1980). Mozambique: Republica da Mozambique (1983). Nigeria: Aluko (1972) and (1973). Rwanda: Republique Rwandaise (1978). Tanzania: Tanzania (1982). Zambia: Milimo and Fisseha (1986). 26 employment on average in these regions. But in East and Southern Africa, with their settler economies and a long history of wage labor on plantations and,in mines, the percentage of nonfarm wage labor appears much higher, in the region of 60% of rural nonfarm employment. Unfortunately, the Malawi, Mozambique and Tanzania censuses did not specifically include apprentices as a job category in their classification scheme, so their hired labor category may include apprentices as well. Consequently, it is not possible to determine whether their categorization reflects a real absence of apprenticeship or simply a lapse in survey design. Either census definitions do not provide sufficient job category discrimination in Southern Africa or history has left a legacy of differential employment structures across the continent. 2. Capital Requirements Not only are rural nonfarm enterprises small in terms of employment per enterprise, they require modest amounts of capital as well. As Table 14 shows, avc,age capital per enterprise commonly ranges between $500 and $4,000 in African rural nonfarm activities. But requirements vary considerably among activities. High capital activities such as grain milling, sawmills, wholesaling, gasoline stations, photography and pharmacies demand on the order of $1,000 to $10,000 per enterprise; while pottery, leatherwork, weaving, small scale trading, personal services and repairs require in the range of $50 to $600. Because capital includes not only the cost of equipment but also buildings, and working capital, investment costs depend heavily on the nature of the business premises and whether they are owned or rented. 27 TABLE 14 CAPITAL INVESTMENT IN AFRICAN RURAL NONFARM ENTERPRISES (1980 $U.S. per enterprise) Burkina,a Kenya Rural Botswana Eastern Province Central Province Sierra Leone Manufacturing Tailoring $ 566 $ 191 $ 1,528b $ 571 Beer Brewing 110 -- 9,785 -- Milling 2,460 3,378 7,376 -- Carpentry 126 429 5,878 676 Baking 7,500 88 -- 1,527 Blacksmith -- 69 -- 909 Metal Working -- 204 1,489 -- Commerce Gas Stations -- 5,000 6,132 -- Wholesaling -- -- 13,689 -- Pharmacies -- 1,514 -- Bars -- 584 4,246c General Retailers -- 521 4,212 Services Restaurants -- 49 -- Photography -- 983 -- -- Barbers -- 210 515 -- Bicycle Repair -- -- 1,552 -- aStart-up capital. All others capital at time of interview. bAverage employment is eight per enterprise as opposed to one to two in Botswana. Probably includes building. cIncluded restaurants as well. SOURCE: Botswana: Haggblade (1982). Burkina: Wilcock and Chuta (1981). Kenya: Freeman and Norcliffe (1985). Sierra Leone: Liedholm and Chuta (1976). 28 Buildings, for example, account for about 80% of the investment costs in Botswana and Sierra Leonean bakeries. For this reason, and because it is more comfortable in hot climates, many rural entrepreneurs prefer to work outdoors, under trees or simple lean-tos. Displaying data for a range of nonfarm activities in Sierra Leone, Table 15 decomposes total rural enterprise investment costs into their three constituent elements. In addition to giving a feel for the relative magnitudes of each component of investment costs, the table illustrates the tendancy of businesses in larger towns to operate from more sophisticated premises than their counterparts in small rural localities. 3. Seasonalit Rural nonfarm enterprises experience marked seasonal fluctuations in activity. As Figure 1 shows, levels of rural nonfarm activity generally run countercyclically to the agricultural calendar. Nonfarm enterprises typically achieve peak activity in the dry season immediately after harvest when both labor inputs and cash demand for nonfarm products are most freely available in the rural economy. While Figure 1 depicts nonfarm labor profiles in Nigeria, similar seasonal profiles have been observed in Burkina (Delgado, 1979), Ivory Coast (Cote d'Ivoire, 1965a and 1967a), Ghana (Yankton, 1985), Gambia (ILO, 1985c), Botswana (Chernicovsky et.al. 1985), Sierra Leone (Liedholm and Chuta, 1976) and in other parts of Nigeria (Mueller and Zvering, 1969). Despite the marked seasonality, numerous time allocation studies have found that nonfarm troughs rarely decend to zero. Even during the peak' agricultural season, nonfarm activities can occupy as much as three to four hours a day (Eicher and 29 TABLE 15 COMPOSITION OF CAPITAL STOCK IN SIERRA LEONEAN SMALL INDUSTRY, 1974/75 (Leones) Locality Size Less Industrial Components of Than 2,000- 20,000- Over Category Capital 2,000 20,000 100,000 100,000 Tailors Building 57 102 288 911 Equipment 128 165 244 469 Working capital 0 16 36 26 Total capital input 185 283 568 1,406 Carpentry Building 57 488 292 Equipment 118 .261 91 Working capital 0 100 .221 Total capital input 175 849 604 Blacksmiths Building 57 351 1,785 1,171 Equipment 123 426 816 1,277 Working capital 0 62 6 108 Total capital input 180 839 2,607 2,556 Baking Building --- 409 Equipment 72 --- Working capital --- 75 Total capital input 556 --- SOURCE: Liedholm and Chuta (1976). 30 FIGURE 1 SEASONAL PATTERNS OF LABOR USE IN FARM AND NONFARM ACTIVITIES, DAN MAHAWAYI, NORTHERN NIGERIA 1966-67 Percentage of total annual man hours in farm and nonfarm work 9 8 Percentage of annual hours spent in farm work 7 6 5 4, \Percentage of annual - 1..,, /hours spent in 1 nonfarm work 3 - 2 1- Apr May Jun Jul Aug Sept Oct Nov Dec Jan Feb Mar Apr Rainy > - Dry - -- Planting --------- Harvesting ) +- Weeding -- Source: Norman (1973) as rendered by World Bank (1978a). 31 Baker5 1982; Cleave, 1974; Norman, 1974; Matlon et.al. 1982). While in the aggregate nonfarm activity fills in the troughs in the agricultural calendar, a disaggregated view of the nonfarm economy reveals several exceptions to that general tendancy. Evidence from Sierra Leone, for example, indicates that rural blacksmithing activity reaches its peak during the height of the agricultural season, as farmers require new tools and repair services for farm equipment. Bakqng also reaches its peak at this time, probably because its closest substitute, rice, is least available prior to harvest (Liedholm and Chuta, 1976). 4. Composition Aggregate employment data for rural Africa remain in short supply, because most studies have focused on limited portions of the rural economy. Many have concentrated exclusively on farm households; while others, interested in the nonfarm economy, commonly enumerate manufacturing and repair services, rarely including other services or commercial establishments. Nonetheless, some available data do allow a comprehensive view of the rural economy. Coming mainly from population censuses, they are presented in Table 16. Amid wide variation, these data indicate that commercial establishments typically predominate, employing 30 to 40% of the rural nonfarm workforce. Services and manufacturing comprise about 25% each, with construction and mining accounting for the remainder. Because manufacturing has frequently been singled out for special investigation, it is possible to provide a more detailed decomposition of that one quarter of the rural nonfarm economy. The breakdowns in Table 17 TABLE 16 MAGNITUDE AND COMPOSITION OF AFRICAN RURALa NONFARM EMPLOYMENTb A. Enployment Percentages 1 2 2a 2b 2c 2d 2e 2f 3 AGRIC. NON-AGRI Minin Mfr. Constr. Comerce* Service Other TOTAL Countries ( % of total emp.) (percentage of rural non-agricultural errployrrent) AFRICA Benin, 1961 58.6 41.4 0.0 17.2 1.2 78.2 3.3 0.0 100.0 Cameroon, 1976 89.6 10.4 0.2 23.8 8.6 16.1 31.4 19.9 100.0 Chad, 1964 96.9 3.1 0.0 12.7 8.1 20.8 '58.4 0.0 100.0 Ghana, 1960 73.2 26.8 5.7 28.8 9.1 44.4 12.0 0.0 100.0 Ivory Coast, Bouake Region, 1963 90.0 10.0 0.0 34.1 0.0 50.7 15.2 0.0 100.0 Malawi, 1977 89.0 11.0 0.8 24.9 14.9 23.5 19.4 16.4 100.0 Mali, 1976 91.7 8.3 0.8 39.8 0.8 9.0 14.9 34.7 100.0 Mauritania, 1977 79.2 20.8 1.9 17.3 5.8 34.1 40.9 0.0 100.0 Mozanbique, 1980 91.0 9.0 0.0 50.9 5.6 17.6 25.9 0.0 100.0 Rwanda, 1978 95.1 4.9 9.5 23.4 13.2 11.0 40.2 2.7 100.0 Senegal, 1970/71 82.3 17.7 2.4 34.3 4.3 38.3 20.5 0.2 100.0 Sierra Leone, 1974 86.2 13.8 5.4 20.3 7.9 44.9 21.4 0.0 100.0 Tanzania, 1978 93.9 6.1 0.0 19.6 0.0 9.6 23.1 47.7 100.0 Togo, 1970 74.5 25.5 0.1 18.6 4.2 26.7 7.9 42.4 100.0 Zimbabwe, 1982 80.0 20.0 0.0 46.6 0.0 8.1 27.4 17.9 100.0 ASIA India, N. Arcot Region, 1981 75.0 25.0 100.0 Philippines, Neuva Ecija Prov., 1971 52.0 48.0 100.0 Thailand, 1975/76 83.0 17.0 100.0 Table 16 (CONT.), Page 2 B. Employmfent Densities (enployment per 1000 population) 1 2 2a 2b 2c 2d 2e 2f 3 Countries AGRIC. NON-AGRI MiningMfr. Constr. Coamrce* Service Other Grand Total AFRICA Benin, 1961 265.6 188.0 0.0 32.4 2.2 147.1 6.3 0.0 453.7 Cameroon, 1976 359.3 41.7 0.1 9.9 3.6 6.7 13.1 8.3 401.1 Chad. 1964 302.1 9.5 0.0 1.2 0.8 2.0 5.6 0.0 311.6 Ghana, 1960 277.8 101.8 5.8 29.4 9.3 45.2 12.2 0.0 379.7 Ivory Coast, Bouake Region, 1963 424.4 47.0 0.0 16.0 0.0 23.8 7.1 0.0 471.4 Malawi, 1977 374.0 46.3 0.4 11.5 6.9 10.9 9.0 7.6 420.3 Mali, 1976 333.9 30.3 0.3 12.1 0.2 2.7 4.5 10.5 364.1 Mauritania, 1977 249.6 65.6 1.3 11.3 3.8 22.4 26.8 0.0 315.2 Mozambique, 1980 450.6 44.5 0.0 22.6 2.5 7.8 11.5 0.0 495.1 Rwanda, 1978 544.0 27.8 2.6 6.5 3.7 3.1 11.2 0.8 571.8 Senegal, 1970/71 121.3 26.1. 0.6 9.0 1.1 10.0 5.4 0.1 147.4 Sierra Leone, 1974 530.4 85.0 4.6 17.3 6.7 38.2 18.2 0.0 615.4 Tanzania, 1978 425.6 27.6 0.0 5.4 0.0 2.7 6.4 13.2 453.2 Togo, 1970 268.5 91.9 0.1 17.1 3.9 24.5 7.3 38.9 360.3 Ziffbabwe, 1982 233.0 58.2 0.0 27.1 0.0 4,7 15.9 10.4 291.1 ASIA India, N. Arcot Region, 1981 253.0 83.0 336.0 Philippines, Nueva Ecija Prov., 1971 132.0 120.0 750.0 Thailand, 1975/76 625.0 125.0 252.0 Table 16 (CONT.), Page 3 aRural as defined in each population census. Normally localities with less than about 5000 persons. See attached notes for specific country definitions. bEmployrent is defined as economically active population primarily engaged in each activity. Seasonal employment. See attached notes for labor force definitions. *Includes Transport. SOURCES: Benin: Republique du Dahomey (1964). Caneroon: Republique du Cameroun (1980). Chad: Republique du Tchad (1966). Ghana: Ghana (1964). Ivory Coast: Republique de Cote d'Ivoire (1964). Malawi: Malawi Government (1980). Mali: Republique du Mali (1980). Mauritania: Republique Islamique de Mauritanie (1979). Mozambique: Republica Popular de Mocambique (1983). Rwanda: Republique Rwandaise (1978). Senegal: Republique du Senegal (1973). Sierra Leone: Thomas (1983). Tanzania: United Republic of Tanzania (1982). Togo: Republique Togolaise (1974). Zinbabwe: Zimbabwe (1985). 35 TABLE 16 NOTES AND REFERENCES SOURCES: Benin: Republique du Dahomey (1964). Cameroon: Republique du Cameroun (1980). Chad: Republique du Tchad (1966). Ghana: Ghana (1964). Ivory Coast: Republique de Cote d'Ivoire (1964). Malawi: Malawi Government (1980). Mali: Republique du Mali (1980). Mauritania: Republique Islamique de Mauritanie (1979). Mozambique: Republica Popular de Mocambique (1983). Rwanda: Republique Rwandaise (1978). Senegal: Republique du Senegal (1973). Sierra Leone: Thomas (1983). Tanzania: United Republic of Tanzania (1982). Togo: Republique Togolaise (1974). Zimbabwe: Zimbabwe (1985). NOTES: Benin: Rural = not explicitly ciefined. Implicitly includes population in localities of less than 9500 persons. Economically active all people over 15 years of age. Women's participation rate = 73% vs. 95% for men. Cameroon: Rural = all areas outside of provincial, departmental, arrondissement and district headquarteres and outside selected settlements over 5,000 with highschools, hospitals or train stations. Economically active = population six and over. Women's participation = 38% vs. 62% for men. Chad: Rural = all population living outside of prefecture headquarters or in localities of less than 5000 in the South and less than 3000 in the North. Economically active = all people over 15 years of age. Women's participation rate = 28% vs. 94% for men. Ghana: Rural = localities under 5,000 according to official definition. But economic activity breakdown only available for towns under 10,000. Data in table refer to towns under 10,000 as rural. Economically active = all people over 15 years of age. Women's participation rate = not given explicitly but is evidently 33% lower than men's. Ivory Coast, Bouake Region: Rural = entire Bouake region including rural towns. Economically active = population aged 15-59. Women's participation rate = not specified. Malawi: Rural = undefined. Rural towns taken as all urban areas outside of Lilongwe, Blantyre and Zomba. Economically active = all persons over 10 years of age. Women's participation rate = 44% vs. 52% for men. 36 Table 16 (CONT.) Mali: Rural = areas outside of legally designated "communes". Implicitly settlements over about 3,000. Economically active = not defined. Women's participation rate = not given. Mauritania: Rural = not defined. Taken as all regions except 3 with over 85% population designated as urban. Essentially all localities under 10,000. Economically active = all persons over 12 years of age. Women's participation rate = 25% vs. 82% for men. Mozambique: Rural = all areas outside 12 major cities. Twelve city sizes not cited. Economically active = all persons over 12 years of age. Women's participation rate = 81% vs. 78% for men. Rwanda: Rural = areas outside prefecture headquarters and two other localities designated as urban. Economically active = all persons over 7 years of age. Women's participation rate = not given. Senegal: Rural = localities less than 1000. Economically active = all persons over 6 years of age. Women's participation rate = 23% vs. 60% for men. Sierra Leone: Rural = census definition, all towns under 5000. Since employment not broken down by town, rural defined here as all districts with at least 75% of their population in localities of under 2000. Economically active = all persons over 12 years of age. Women's participation rate = equal to that of men. Each accounts for 50% of the labor force. Tanzania: Rural = regional and district headquarters plus towns over 5000. Economically active = all persons over 5 years of age. Women's participation rate = 55% vs. 54% for men. Togo: Rural = all areas outside of Lome and 6 legally designated "communes". Economically active = population over 12 years of age. Women's participation rate = not given. Zimbabwe: Rural = population living in towns under 2,500. During 1969 census urban = any locality over 150 if majority of males employed in non-agricultural occupations. Economically active population over 15. Women's participation rate = 48% vs. 80% for men. 37 TABLE 17 ACTIVITY BREAKDOWN OF RURAL MANUFACTURING EMPLOYMENT (in percents) Sierra Burkinaa Zambiab Kenyac Nigeriad Leonee Item 1980 1986 1977 1972/73 1976 Food/Agri cultural f Processing 55 55 22 2 5 Clothing Products 25 6 22 56 53 Wood Products 1 32 43 11 19 Metal Products 8 4 6 4 19 Other 21 3 7 27 4 bEastern ORD, 1980. Six Rural Provinces, 1986. dCentral Province Village Centers, 1977. Four States, 1972/73. eAll Rural, 1976. Includes only bakeries. SOURCE: Burkina Faso: Chuta and Wilcox (1982); firms below 50 persons. Kenya: Norcliffe, Freeman, and Miles (1984); firms below 50 persons. Nigeria: Aluko et al. (1972 and 1973); firms below 50 persons; enterprise, not employment percentages. Sierra Leone: Liedholm and Chuta (1976); firms below 50 persons. Zambia: Milimo and Fisseha (1985). 38 indicete that, within manufacturing, food preparation, tailoring, carpentry and metal working activities predominate. The importance of food preparation - particularly beer brewing, small restaurants and prepared foods - varies enormously across studies, in part due to differences in activity coverage and failure of some studies to interview within households. Differences in survey coverage complicate cross country comparisons not only in manufacturing but even more so for the remainder of the rural nonfarm economy. They make it difficult to meaningfully compare employment percentages across surveys, because activity shares are computed as percentages of different bases depending on the comprehensiveness of the survey coverage. One solution to this problem is to compute employment densities per 1,000 members of population. Because it measures all surveyed activities against a common yardstick, this procedure allows comparison of relative activity intensities for all activities enumerated, even if a survey's activity coverage is only partial. This approach is taken throughout much of this paper, especially in Section III.A. where we explore features of structural 'transformation using cross section data. Using this approach, the nonfarm employment densities in Table 16 provide tantalizing initial insights into differences in the Asian and African rural nonfarm sectors. The data, in Panel B, suggest that Asian rural nonfarm employment densities lie in the range of 80 to 120 per 1,000 population, while the Africa densities only in West Africa reach this level. Elsewhere they are much lower, more frequently on the order of 40 to 50 per 1,000. Thus, initially, it appears that the Asian rural economy 39 supports about double the nonfarm activity found in East, Central and Southern Africa. 5. Gender Across much of Africa, women participate actively RNFE. Yet social and religious norms vary considerably from country to country and in some cases tightly shape the economic options available to women. This is clearly the case in Muslim countries, as the Chad data in Table 1 indicate. And even in non-Muslim countries, convention frequently dictates that women participate in activities that can be done in the home, with low capital requirements,. and with skills they already have. This leads to their predominance in food preparation, restaurants, basket making, and in some regions, ceramics, weaving and tailoring. Of course across West Africa, and in many other regions as well, women also play a major role in retailing and trade. As Table 18 indicates, women account for a substantial proportion of both employment and management in rural enterprises, between 50 and 60% in Zambia and Ghana. Because of differential access to education and because the obligations of child rearing combine with social expectations, this is far higher than female participation rates in formal sector employment. For Ghana and Kenya, the two countries in Table 18 for which it is possible to directly compare, women's share of formal sector employment reaches only 10% and 14% respectively compared to 54% and 25% shares in informal, small enterprises. (Steel, 1977; ILO, 1972). So opportunities in small private businesses afford African women a major economic outlet. An important question to be examined below is how projected changes in the African rural economy will affect these predominantly female nonfarm occupations. 40 TABLE 18 FEMALE PARTICIPATION IN RURAL NONFARM ENTERPRISES IN SUBSAHARAN AFRICA: FEMALE PARTICIPATION AS PERCENT OF TOTAL a bc Zambia Kenya Ghanac Owners Employed Employed Employed Manufacturing Traditional beer 99 91 -- -- Ceramics 92 85 -- -- Garments 75 73 26 65 d Foods 69 52 15 86 Carpentry 12e 12e 4 21 Metals 6. 4 4 1 Commerce Vending/Retailing 59 60 38 -- Trading 31 24 -- Services Restaurants 37 47 18 -- Bars 12 17' -- -- Total 60 54 25 54 aRural, Six Provinces, 1985. bCentral Province Market Centers, 1977. dTotal National Manufacturing, 1970. eIncludes primary extraction. fAll forest based production, charcoal production plus carpentry. Includes metals and carpentry together. SOURCE: Zambia: Milimo and Fisseha (1985) (Includes only enterprises employing 50 or less.); Kenya: Norcliffe and Freeman (1979); Ghana: Steel (1977). 41 III. STRUCTURAL TRANSFORMATION OF THE RURAL NONFARM ECONOMY A. Empirical Profile of the Structural Transformation 1. Megatrends In evaluating the present structure of Africa's rural nonfarm enterprises as well as the potential for farm-nonfarm growth linkages, it is useful to first step back and review the grand lines of the economic transformation that has been taking place there over long stretches of time. Figure 2 describes that general evolution over a time horizon of many centuries. It begins long ago, with a countryside populated by self- sufficient, primarily agricultural households which provide for all their own needs. There are no market exchanges, but within each household family members engage in both farm and nonfarm activities. By fiat, the head of household allocates labor and capital among activities. Specialization, a central feature of economic growth, emerges very early on in response to technical change, opportunities for trade with outside regions, introduction of new crops, population growth and a host of other factors. Diversification out of agriculture is a key feature of this specialization. In the absence of secure importable food supplies, specialization requires, as a precondition, increased agricultural productivity per worker in order to avoid community starvation. As population grows and opportunities arise, households begin to concentrate on some activities, beyond their own consumption needs, based on locational advantages, historical access to key inputs or technology, or the proclivity and talents of individual members. Households begin commodity Figure.-2 42 Evolution of Rural Nonfarm Enterprises (RNFE) in the Rural Economy cops RNFE Time = 1 Production for sale Livesto - Production for own consumption E Time = 2 Liveg, k RNFE Crop Time =3 LivestockUb' Urba-, RNFE Enter. RNFE Crops E prise, Time R4 Rural Areas Rural Villages Rural Towns Urban Areas 43 and resource exchanges. They trade goods and services with other family groups, and their share of consumption from own production declines. In parts of Africa, social controls develop through which certain ethnic groups or castes specialize in specific nonfarm activities such as metal work, tailoring, weaving, or other nonfarm activities (See, for example, Hill, 1967). Initially, local exchanges among households are regulated by social protocols. Families exchange goods and services through socially monitored reciprocal giving. They transfer capital through rotating cr,edit societies regulated by social norms and pressure (Bauman, 1977; DeLancey 1978a; Miracle, Miracle and Cohen, 1980). They exchanges labor through reciprocal cooperative work parties (Erasmus, 1956, Moore, 1975). Ultimately markets develop - labor markets, rural financial markets, and commodity markets. They operate alongside social allocation mechanisms and ultimately supplant them. But the time frame is very long. Even today in parts of rural Africa, families exchange as much labor through reciprocal work parties as they do through the hiring of nonfamily wage labor (Weinrich, 1975). And capital flows remain predominantly intra- family (See Tables 29 and 30). In virtually all of Africa, capital exchanges across families are channeled through rotating credit societies and other socially regulated institutions far more frequently than through formal financial markets (Miracle, Miracle and Cohen, 1980). As specialization continues, some nonfarm enterprises split off from the farm household, the economic analog of mitosis. Evidence from Kenya, displayed in Table 19, suggests that transport, fin.ancial services and 44 TABLE 19 DEGREE OF ASSOCIATION BETWEEN RURAL NONFARM EMPLOYMENT AND AGRICULTURE IN RURAL KENYA Percent of Rural Nonfarm Entrepreneurs Working At Least One Day Per Week Activity In Agriculture Transport & Storage 14.80 Services: Financial 20.00 Metal Products 33.30 Services: Community 38.50 Wood Products 45.00 Repairing 50.00 Food, Beverage and Tobacco Products 53.20 Wholesale/Retail 60.60 Resource Extraction 66.70 Construction and Building Materials 66.70 Other Manufacturing 72.70 Fiber Production and Clothing 73.90 SOURCE: Kenya: IRS Nonfarm Module 2, 1977-78 (unpublished), cited in Freeman and Norcliffe (1985) p. 71. 45 metal working are among the first activities to split off from the farm household, while weaving and tailoring, resource extraction and constru.ction remain integrated longer. The growing proportion of nonfarm economic activity and the emergence of independent nonfarm enterprise go hand in hand with spatial concentration of population in village centers and rural towns. The concentration of people and activities in village towns stimulates the emergence of primarily nonfarm households by generating a volume of demand sufficient to support specialized non-agricultural production units. Certainly all available evidence indicates dramatically higher proportions of nonfarm activity in rural towns than in dispersed rural settlements (See Matlon 1977, p.30; Liedholm and Chuta, 1976 p.42; Tables 20-23). Moreover, the emergence of primarily nQn-farm households accelerates the move to a market economy. Given the difficulties of rural communication and transport, social norms and household heads become physically unable to exert control over input allocation and production decisions across distances. Hence the physical separation of farm and nonfarm activities accelerates the move to market exchanges of both inputs and outputs. Because they are specialized, nonfarm households are inclined to sell their wares'for cash in order that they may purchase food and other necessities. Even in areas where farm production remains almost entirely consumed within the producing household, nonfarm goods are traded for cash (See Ancey, 1974 for evidence from Ivory Coast.) Sutter (1977) has suggested that the rise of nonfarm employm6nt lies at the heart of the breakdown of the extended family system of economic resource management in 46 TABLE 20 DENSITY AND COMPOSITION OF AFRICAN RURAL NONFARM EMPLOYMENT, BY SIZE OF LOCALITY (employnent per 1,000 population) 1. 2. 2.a. 2.b. 2.c. 2.d. 2.e. 2.f. AGRIC. NON-AG. Mining Mftr. Constr. Connerce* Serv. Non.Oesig Benin, 1961 Rural 266 188 0 32 2 147 6 0 Urban 37 313 0 59 13 189 53 0 Cameroon, 1976 Rural 359 42 0 10 4 7 13 8 Rural towns 142 152 0 27 11 41 60 12 Urban 8 270 1 52 24 81 95 18 Chad, 1964 Rural 302 10 0 1 1 2 6 0 Rural towns 132 131 0 20 15 34 62 0 Ghana, 1960 Rural 278 102 6 29 9 45 12 0 Rural towns 55 239 14 45 25 112 43 0 Urban 38 473 13 77 65 207 112 0 Malawi, 1977 Rural 374 46 0 12 7 11 9 8 Rural towns 152 177 0 19 27 56 60 14 Urban 42 287 0 62 38 69 98 19 Mali, 1976 Rural 334 30 0 12 0 3 5 10 Rural towns 112 128 4 25 5 32 53 9 Urban 29 - 207 9 23 11 53 101 11 Mauritania, 1977 Rural 154 40 1 7 2 14 17 0 Urban 14 371 43 35 48 95 149 0 Mozambique, 1980 Rural 451 44 0 23 2 8 12 0 Urban 122 241 0 76 11 71 82 0 Nigeria, 3 Western States 1966 Rural 22/ 355 0 188 0 147 9 10 Rural towns 124 349 0 119 0 177 40 12 Rwanda, 1978 Rural 544 28 3 7 4 3 11 1 Rural towns 324 185 0 20 14 37 110 3 Urban 149 364 3 39 70 68 174 9 Senegal, 1970/71 Rural 121 26 1 9 1 10 5 0 R).ral towns 111 69 0 15 3 27 24 0 Urban 16 170 1 28 10 55 76 0 Sierra Leone, 1974 Rural 530 85 5 17 7 38 18 0 Rural towns 691 309 54 70 19 115 51 0 Urban 123 877 4 90 70 450 263 0 Togo, 1970 Rural 312 107 0 20 5 28 8 45 Rural towns 71 268 1 51 17 94 42 63 Urban 10 384 0 64 23 118 93 86 *Includes transport plus commerce. SOURCE: As in Table 16.. 47 TABLE 21 COMPOSITION OF AFRICAN RURAL NONFARM EMPLOYMENT, BY SIZE OF LOCALITY (percent) 1. 2. 2.a. 2.b. 2.c. 2.d. 2.e, 2.f. 2.g. AGRIC. NON-AG. Mining Mftr. Constr. Camerce* Serv. Non.Desig. Total Benin, 1961 Rural 59 41 0 17 1 78 3 0 100 Urban 11 89 0 19 4 60 17 0 100 Cameroon, 1976 -ara - 90 10 0 24 9 16 31 20 100 Rural towns 48 52 0 18 7 27 40 8 100 Urban 3 97 0 19 9 30 35 7 100 Chad, 1964 Rural 97 3 0 13 8 21 58 0 100 Rural towns 50 50 0 15 12 26 47 0 100 Ghana, 1960 Rurat 73 27 6 29 9 44 12 0 100 Rural towns 19 81 6 19 11 47 18 0 100 Urban 7 93 3 16 14 44 24 0 100 Ivory Coast, Bouake Region 1963 Rural 96 4 Rural towns 26 74 Urban 11 89 Malawi, 1977 Rural 89 11 1 25 15 23 19 16 100 Rural towns 46 54 0 11 15 32 34 8 100 Urban 13 87 0 22 13 24 34 7 100 Mali, 1976 Rural 92 8 1 40 1 9 15 35 100 Rural towns 47 53 3 20 4 .25 41 7 100 Urban 12 88 4 11 5 25 49 5 100 Mauritania, 1977 Rural 79 21 2 17 6 34 41 0 100 Urban 4 96 12 10 13 26 40 0 100 Mozambique, 1980 Rural 91 9 0 51 6 18 26 0 100 Urban 34 66 0 32 5 30 34 0 100 Nigeria, 3 Western States 1966 HuraN 39 61 0 53 0 41 3 3 100 Rural towns 26 74 0 34 0 51 12 4 100 Rwanda, 1978 RurTl 95 5 9 23 13 11 40 3 100 Rural towns 64 36 0 11 8 20 60 2 100 Urban 29 71 1 11 19 19 48 3 100 Senegal , 1970/71 Rural 82 18 2 34 4 38 21 0 100 Rural towns 62 38 0 21 4 40 34 0 100 Urban 8 92 0 17 6 32 44 0 100 Sierra Leone, 1974 Kura! 86 14 5 20 8 45 21 0 100 Rural towns 69 31 17 23 6 37 17 0 100 Urban 12 88 0 10 8 51 30 0 100 fogo, 1970 Trfal- 75 25 0 19 4 27 8 42 100 Rural towns 21 79 0 19 6 35 16 24 100 Urban 2 98 0 17 6 31 24 22 100 *Includes transport plus comerce. SOURCE: As in Table 16. 48 TABLE 29 DENSITY OF RURAL NONFARM EMPLOYMENT BY SIZE OF LOCALITY, RURAL ZAMBIA 1985a (employment per 1000 population) Standard Enumerated Rural Towns Small Towns Areas 0-12,000 12-50,000 Total Rural, (dispersed) Inhabitants Inhabitants Six Provinces Manufacturing 1. Traditional beers 38.57 19.96 9.50 32.06 2. Sweet beer 2.34 3.83 1.98 2.20 3. Foods- 3.11 11.46 9.38 3.71 4. Woodworking/ 28.54 7.64 5.65 23.18 Charcoal 5. Garments 2.52 8.99 7.62 3.04 6. Metalworking 3.21 1.22 1.39 2.68 7. Ceramics 2.06 0.36 0.87 1.69 8. Leatherworking 0.37 0.77 0.63 0.38 9. Other 0.09 0.31' 0.22 0.11 Total 80.80 54.54 36.22 69.03 Commerce 1. Vending 7.10 42.39 43.37 11.07 2. Trading 2.29 14.72 13.96 3.63 Total 9.39 TTTT TT4 T.70 Services 1. Bars/taverns 1.10 1.52 1.93 1.08 2. Restaurants 0.00 1.79 1.30 0.20 3. Repairs 0.64 2.42 2.60 0.83 4. Other 3.44 1.26 1.37 2.86 Total 95.37 118.65 100.76 88.69 Grand Total 186 230 194 172 aThis study covered the rural areas of six of Zambia's eight provinces. Excluded were the Copperbelt and Lusaka provinces. The study area accounted for 60 percent of Zambia's population. SOURCE: Milimo and Fisseha (1986), pp. 34, 47; and 1980 Population census data. 49 TABLE 23 DENSITY OF RURAL MANUFACTURING AND REPAIR EMPLOYMENT BY SIZE OF LOCALITY, SIERRA LEONE 1974 (employment per 1,000 population) Size of Locality Under 5,000- Over 2,000 20,000 100,000 1. Tailoring 8.92 10.19 8.62 2. Carpentry 5.31 4.09 1.25 3. Blacksmithing 5.52 0.66 0.13 4. Matmaking 4.25 0.00 0.00 5. Spinning & weaving 2.55 0.27 0.00 6. Baking 1.49 1.25 0.59 7. Goldsmithing 1.06 0.45 0.20 8. Gara Dyeing 0.00 1.19 0.58 9. Wood carving 0.64 0.15 0.01 10. Vehicle repair 0.00 1.07 2.09 11. Shoemaking/repair 0.42 0.36 0.47 12. Welding and fitting 0.00 0.07 0.31 13. Watch repair 0.00 0.23 0.28 14. Radio repair 0.00 0.15 0.22 15. Other 0.85 0.76 3.49 Total 31.00 20.89 18.24 SOURCE: Liedholm and Chuta (1976), p. 15 and Thomas (1979). 50 Nigeria. Although causality is difficult to prove, and alternative explanations have been advanced (Norman, Simmons and Hays, 1982), there is little doubt that increasing nonfarm activity, a declining proportion of intra-household in total resource allocation decisions, arid increased rural monetization go hand in hand. In some cases, rural nonfarm activities do not develop organically from the specialization of polyvalent rural households but instead are implanted by outsiders. In fact, outsiders - Asians in'-East Africa, Lebanese and Greeks in West Africa, and white settlers in East and Southern Africa - have instigated much of the growth in African rural nonfarm activities. The source of growth in rural nonfarm activity - organic or implanted from outside - has strong implications for both the size structure and technology used in rural nonfarm enterprises. Ultimately, rural nonfarm enterprises must face competition from the outside, either from enterprises in growing nearby rural towns or from urban or imported goods. Manufacturers, especially those producing easily transportable items, face the stiffest competition. Yet rural services remain insulated, because by their nature they are difficult to move across space. As economies mature, the magnitude and composition of rural nonfarm enterprises becomes dependent on the competitiveness of local suppliers. That competitiveness, in turn, depends on a host of factors that will be explored shortly. 2. State of the transformation The evolution of Africa's rural economy has proceeded at differing rates across the continent. Even so, some general orders of magnitude can be attached to each of the key features of-the structural transformation. 51 First, monetization of the rural economy measures a key feature of the transformation, the degree to which market sale of goods has supplanted intra-household exchange. Available evidence shows a surprisingly consistent 60% cash share in total rural transactions. Countries as diverse as Ivory Coast, Zambia, Uganda, Kenya, Tanzania and Burkina all fall within 3 percentage points of 60%; and only .Sierra Leone has 'a monetized share of rural economic exchanges as low as 52%"(ILO 1982i; ILO 1983d; Collier, Radwan and Wangwe, 1986; Bessel and Roberts, 1971; Massel and Parnes, 1969; Byerlee and King, 1977). Disaggregating reveals substantial differences between rates of monetization of agricultural and nonagricultural commodities. Nonfarm entrbpreneurs exchange the overwhelming. majority of their goods and services for cash, while agricultural commodities are monetized to a smaller extent. Farmers commonly market 70% to 90% of their cash crops, although continent wide they sell only about 30% of all foodcrop production (FAO, 1986; Cleave, 1974). Foodgrain marketing ranges from about 15% of total production in the Sahel to 25% in humid Central .and West Africa and as high as 50% in Zimbabwe (FAO, 1986; Cleave, 1974; Srivastava and Livingstone, 1983; Norman, Pryor and Gibbs, 1979; Delgado, 1979; Hedlund and Landahl, 1983; Wilcock and Chuta, 1982; Anthony et.al.,1979; Leurquin, 1960). A second feature of structural transformation, diversification of the rural economy out of agriculture and into nonfarm activities, has proceeded at a variable rate across the continent as the data in Table 3 attest. 52 From those data, it appears that nonfarm income commonly accounts for about 30% of total earnings in rural Africa. Economic mitosis, the splitting off of nonfarm enterprises from integrated agricultural households, constitutes a third feature of the transformation of Africa's rural economy. It measures the proportion of nonfarm activity that takes place in specialized nonfarm,households, indicating roughly the magnitude of capital and labor flows among activities that must be regulated by markets or communal social mechanisms rather than by direct decision of an integrated household head. The rate of household specialization can be roughly estimated from Tables 1 and 2. Since 80 to 90% of Africa's rural labor force works in agriculture and since 30 to 50% works full or part time in nonfarm activity, 30 to 60% of the population must work in both. So as a rough general notion, we can suggest that about 45% of African rural nonfarm activity takes place within integrated farm-nonfarm households, while about 55% is run from specialized nonfarm firms. These magnitudes square well with specific figures available for rural Kenya (Mbithi and Chege, 1973). Labor is a key factor of production in rural Africa. In fact, most observers identify peak season labor bottlenecks as the principal constraint to increased farm production (Eicher and Baker, 1982; Anthony et.al., 1979; Mellor, Delgado and Blackie, 1987). Hence, the effective functioning of labor markets is central to efficiency and growth. The current state of rural labor markets can be summarized in five propositions. First, the hiring of wage labor is more common in nonfarm than in agricultural enterprises. Nonfarm entrepreneurs commonly hire (or 53 apprentice) 50% to 70% of their workers from outside the family, while the figure in agriculture averages closer to 15-20% (Table 13; Eicher and Baker, 1982; Byerlee, 1980; Spencer and Byerlee, 1976; Norman, 1972)). Second, the use of wage labor in agriculture is far more prevalent in cash cropping regions than elsewhere. Cash crop farmers commonly hire on the order of 25% of total labor input (Cleave, 1974), while growers of food crops engage closer to 10% from outside the family (Cleave, 1974; Norman, Pryor and Gibbs, 1979; Matlon, et.al. 1979; Oates, 1984; Robertson and Hughes, 1978; Weinrich, 1974; Anthony et.al., 1979; Collier and Lal, 1986; Collier, Radwan and Wangwe, 1986). Third, from the foregoing we estimate that about 20% of Africa's rural labor force is channeled through labor markets (60% of nonfarm workers x a 15% employment share plus 15% of agricultural labor x an 85% rural employment share = 22% of rural employment), while household heads allocate the remaining roughly 80% directly within the family. Fourth, the prevalence of communal labor exchange has diminished precipitously in recent years (Moore, 1975; Erasmus, 1956; Cleave, 1974), today normally accounting for only about 5% of all farm labor input (Cleave, 1974; Norman, Pryor and Gibbs, 1979; Collier et.al. 1986; Weinrich, 1974) and probably less in RNFE where the practice appears limited to female labor exchange in rotating beer parties in Southern Africa (Curtis, 1973). Finally, there is some debate about the efficiency of Africa's rural labor markets (Berry and Sabot, 1978; Collier, Radwan and Wangwe, 1986). While efficiency would undoubtedly improve with access to better rural communication and transport systems, it is evident that todays labor 54 markets do direct substantial flows of workers seasonally, regionally and internationally in a process that improves aggregate welfare. Box 1 gives an example of how efficiency gains arise from the specialization and labor market development that accompany the rural transformation. Box 1. Efficiency Gains from Specialization in Rural Sudan -A recent study from Sudan (ILO, 1985f) highlights the benefits arising from economic specialization and the accompanying development of rural labor markets. It compares a matched set of Sudanese farm households, alike in all respects except that the second group has access to significantly greater nonfarm employment due, they speculate, to special skills and training not possessed by other households. Because of the nonfarm earnings, these households earn greater total income than their otherwise identical compatriots. The households with nonfarm earnings hire wage labor to replace family members absent from the farm, thus increasing their overall earnings by substituting low cost wage labor for family members who can 'earn far more off the farm. The hired farm laborers gain as well, presumably agreeing to work on others' farms only when wages exceed what they could earn elsewhere. Income gains to both groups not only raise rural income directly, they also generate additional rural incomes through their expenditures and the consequent second round consumption multipliers. Thus specialization motivates labor market development, labor markets facilitate specialization and the cumulative effect of this transformation is to directly increase rural incomes. A review of rural capital flows completes our snapshot of the structure of Africa's rural economy. Available evidence indicates that well over 90% of all capital is ch.anneled outside the formal banking system, either within families or across households through informal and traditional credit systems. The share of remittance income accruing to rural households - 11% of total rural income in Kenya, 13% in Botswana, 55 and 71% in Lesotho - attests to the importance of intra-family flows and the close links between urban and rural economies (Tables 5,7,9; Kenya, 1978; Colclough and Fallon, 1983; ILO 1982L). These magnitudes suggest that intra-family flows predominate in the rural economy. While data on sources of agricultural investment funds are scarce, analagous information from nonfarm enterprises indicates that family funds account for about 90% of investment capital (Table 29). Informal credit markets and socially regulated rotating credit systems do provide investment funds for equipment (Miracle, Miracle and Cohen, 1980); but they appear most common as sources of working capital and consumption. Studies in Nigeria have found rotating credit societies furnishing up to a third of rural tr.aders' working capital needs (Table 30). 3. Changes in Rural Nonfarm Activity a. Differences by size of locality Consistent with this view of the structural transformation taking place in rural Africa, the empirical evidence in Tables 20 and 21 document the important.role played by emerging rural centers. Because many census definitions of rural exclude towns as small as 3,000 and 5,000, these two tables maintain an urban category for countries whose data do not permit a three-tiered breakdown. In these cases, even "urban" areas include a large proportion of settlements that can be classified as rural towns. The data paint a picture of dramatic increase in the importance of nonfarm activities in rural towns and urban centers. The only exception is Western Nigeria where unusual settlement patterns arise because of Yoruba cocoa farmers' preferences for town living (Berry, 1975). Their 56 urban domicile dilutes nonfarm employment densities in Nigeria's rural towns. Comparing among nonfarm activities, Table 20 shows that the frequency of commercial and service employment increases by far most rapidly with size of locality. Moving from dispersed rural settlements to rural towns, commerc-ial and service employment densities frequently grow by multiples of 5 and 10. Although manufacturing densities also rise, they more commonly grow by a factor of 2 or 3. Because commercial and service densities increase so much faster than manufacturing, the nonfarm employment percentages displayed in Table 21 report a declining proportion of manufacturing in total nonfarm employment as locality size increases. But it is important to disaggregate, because within sectors - especially within manufacturing - some activities fare better than others. As Tables 22 and 23 show, some nonfarm activities fade rapidly as urbanization proceeds and local producers increasingly face competition from larger scale outside suppliers. Spinning, weaving, pottery, woodcarving and mat making decline dramatically in importance with locality size, undoubtedly the victims of competition from cheap manufactured substitute goods. At the other extreme, activities that flourish as town size increases include vending, trading, restaurants, food preparation, and repair work of all sorts. Only one food processing activity runs counter to the trend of generally increasing employment density in larger rural localities. Zambian home beer brewing declines markedly with locality size. This results from government regulations severely limiting home brewing in and - 57 from increased competition from factory-brewed indigenous beers, both of which pose a greater threat to home brewers in larger towns than in smaller settlements (Jules-Rosette, 1982). Thus policy, as well as the economics of production and distribution, can substantially influence the profile of rural nonfarm activity, in this case constraining by far the largest nonfarm employer in rurnl Zambia. Middle performers in rural towns include tailoring, metal working and carpentry. Tailoring employment increases steadily with town size in rural Zambia. In Sierra Leone, it is necessary to add that along with essentially constant employment density across localities, tailors operating in larger settlements work longer hours and earn higher returns (Liedholm and Chuta 197G, pp.16,42,43). Blacksmithing appears most prevalent in small localities, probably because of proximity to farmer repair needs; while metal working activities requiring more substantial equipment investment and throughiut to remain solvent are concentrated more heavily in the rural and urban towns. Carpentry can only be judged from the Sierra Leone data in Table 23, because the Zambian data unfortunately group carpenters together with charcoal manufacture. The Sierre Leone data do indicate a modest increase in carpentry activity in rural towns as compared with dispersed rural settlements. Looking at the rural nonfarm economy-from the consumer's viewpoint, rather than the producer's, Table 24 reinforces several notions from the business school marketing literacure. Displaying data from Kenya, the table indicates thac consumers are far more likely to procure inexpensive and frequently purchased items close to home. Hence they purchase 58 TABLE 24 SPATIAL DISTRIDUTION OF CONSUMER PURCHASES IN RURAL KENYA, 1977 (percent households primarily purchasing from each source) Source of Supply Food Clothing Shoes Pots/Pans Beds/Tables 1. Own holding 47 0 2 1 3 2. Other households 5 0 0 1 6 3. Rural shop (duka) 17 22 21 14 15 4. Market center 28 54 42 65 32 5. Town 2 24 35 19 45 Total 100 100 100 100 100 SOURCE: Integrated Rural Survey Nonfarm module, cited in Freeman and Norcliffe (1985), pp. 43-50. - 59 virtually all food locally. On the other hand, they buy consumer durables, especially expensive and infrequently purchased items such as furniture, in rural towns where volumes are sufficient to sustain specialized carpentry shops. Furniture manufacturers, because their product is bulky (space-gaining) and difficult to transport gain considerable competitive advantage by *producing close to the final consumer. Together, Tables 23 and 24 suggest that carpenters fare particularly well in rural towns where they can service a sufficiently large catchment area to sustain regular operation while transport barriers protect them from urban competitors. b. Distance from urban centers Evidence from Ivory Coast reinforces previously developed notions about which rural nonfarm activities are most likely to be bulldozed by coropet-tion from urban and imported manufacturers. Corroborating the locality size evidence in Tables 22 and 23, Table 25 indicates clearly that basket making and weaving diminish precipitously in importance with proximity to major urban centers. Pottery making also declines, although to a lesser extent. Because these activities constitute the three largest rural industries in Bouake Region, it is not surprising that manufacturing in the aggregate may fare less well than commerce and services as rural structural transformation proceeds. Certainly it is necessary to distinguish between declining industries - such as pottery, weaving, basket making, mat making, and wood carving - and ascendant manufacturing activIties such as food processing, and to a lesser extent tailoring, carpentry and metal work. 60 TABLE 25 SPATIAL DISTRIBUTION OF RURAL MANUFACTURING BOUAKE REGION, IVORY COAST, 1970 (employment per 1,000 population) Kilometers from Bouake (population 110,000) 0-10 10-15 15-20 20-25 25+ Region 1. Basketmaker 6.24 8.75 16.42 19.46 40.74 11.71 2. Weavers 9.73 11.70 13.76 15.72 17.75 10.14 3. Potters 2.65 4.62 3.83 3.38 4.22 2.97 4. Builders (cement) 2.75 2.86 1.89 1.77 1.90 1.79 5. Builders (banco) 1.68 1.76 2.21 1.25 1.06 1.33 6. Dyeing 0.00 0.53 2.30 6.79 1.90 1.58 7. Tailors 1.56 1.19 2.30 1.53 1.48 1.24 8. Wood carving 0.59 1.35 0.86 0.64 1.06 0.74 9. Carpentry 0.59 0.33 0.54 0.88 0.84 0.43 10. Shoemakers 0.00 1.47 0.09 0.00 0.84 0.43 11. Mechanics 0.25 0.20 0.90 0.76 0.42 0.40 12. Blacksmiths 0.25 0.25 0.63 0.64 0.23 0.31 13. Jewelers 0.00 0.08 0.00 0.12 2.74 0.25 14. Hairdresser 0.00 0.00 0.14 0.00 0.00 0.03 Total 26.21 35.17 46.00 53.06 75.24 33.37 SOURCE: Ancey (1974), p.116. 61 Figures 3 and 4 illustrates how locality size and distance from urban centers interact to affect the density of nonfarm activities. Because the spatial and size distribution of localities do not coincide, the two interact in influencing the pattern of nonfarm activity. in any particular settlement. c. Regional differences Within countries, the resource base, rainfall, soil types, population density, and incomes vary dramatically among regions. So too does the composition and prevalence of rural nonfarm activity. Resource based activities, such as sawmilling and mining, and weight losing manufacturing processes such oil extraction and sugar refining are frequently tied to the region in which the raw material is located. Andrae (1981) and Freeman and Norcliffe (1985) document the locational tie between resource availability and sawmilling in Ghana anu Kenya. And regional data from Sierra Leone show clearly that the importance of fishing income, a resource based rural nonfarm activity, varies dramatically among r2gions depending on resource availability (Matlon et.al., 1979). Displaying data from Kenya, Table 26 documents the substantial regional variation in both the level and composition of rural nonfarm income. Freeman and Norcliffe (1985) suggest that high nonfarm earnings in Central Province mirror the greater level of agricultural prosperity in that region. With the aid of Panel 3 in Table 26, it is possible to explore their hypothesis further. While the very high level of wood manufacture in Central Province is clearly related to timber resources 느 FIGURE 4 IWTERACTIONS BETWEEN LOCALITY SIZE AND LOCATION IN THE COMPOSITION OF RURAL NONFARM ACTIVITY KANO STATE, NORTHERN NIGERIA ý1 ýJ N F R M l N C"t-,A E FA R Kil -V ll L-, l Cýi'lili'll F.ýliRM I t---] -1 E LU Lu Lij lý. ri r-J AOO POPULATION 65 3,740 21,960 KM FROM MAJOR 0 13 10 64 TABLE 26 REGIONAL DISTRIBUTION OF RJRAL NONFARM ACTIVITY, KENYA 1977/78 Rift Total Central Coast Eastern Valley Nyanza Western Rural 1. Average Nonfarm Incofe/ 461 425 329 405 437 394 403 Household (Kenya shillings) 2. Source of income per family holding (percent) a. nonfarm surplus 7 18 14 9 6 5 9 b. farm operating surplus 52 27 56 74 70 51 59 c. employrent (ag + nonag) 29 27 20 13 19 27 22 d. remittances and gifts 12 27 10 5 5 17 10 e. Total 100 100 100 100 100 100 100 3. Percent of Population Involved in Each Nonfarm Activity a. resource extraction 8.2 6.6 13.5 24.9 5.1 16.0 13.0 b. food processing 11.7 17.6 .13.5 17.3 20.4 14.9 16.4 (brewing) 4.5 0.4 5.8 13.3 13.1 9.1 7.7 c. textiles and weaving 9.8 13.9 15.1 8.2 10.1 5.6 10.9 d. wood product manufacture 19.0 7.4 13.3 8.5 14.3 8.2 11.5 e. metal manufacture 1.5 0.5 0.8 0.9 0.7 0.2 1.0 (blacksmith) 0.7 0.4 0.2 0.4 0.4 0.1 0.3 (farm implements) 0.0 0.1 0.2 0.3 0.4 0.2 f. trading 13.6 3.8 9.6 12.2 16.1 11.9 10.9 g. repairing 8.0 6.7 6.0 7.8 9.0 10.5 7.8 (farm equipment) 0.1 0.0 0.3 0.1 0.5 0.2 0.2 h. other services 23.9 38.9 20.1 12.4 21.0 23.2 22.8 S"JRCE: Government of Kenya (1977); Freefmn and Norcliffe (1985). 65 rather than agriculture, the metal manufacture and repair data provide some tantalizing glimmers of evidence on backward linkages from agriculture. They suggest that implement manufacture, not large anywhere in rural Kenya, is virtually nonexistent in Central Province, the most densely populated region and that with the highest level of agricultural activity. But blacksmithing and farm equipment repair are twice as common there, albeit at low absolute levels. This suggests that agricultural prosp!rity may be more likely to bring with it repair linkages than implement -manufacture. Of course the proximity of Central Province to Nairobi and the consequently easy penetration of urban manufacturers may prevent growth of rural manufacturing that would otherwise grow up in a more remote region. Given that Freeman and Norcliffe's data list percentages of enterprises, without any size weighting - by employment or value added - these suggestions can only be regarded as very tentative hypotheses. Ongoing work by Kilby (1987) may shed further light on regional differences in the informal sector segment of Kenya's rural nonfarm employment. d. Changes over time Time series data are limited and fragile. Nonetheless, those that do exist can be used to further explore the nature of Africa's rural structural transformation. Using aggregate labor force data for major subregions of the world, both Anderson and Leiserson (1980) and ILO (1979c) make cautious judgements about long-term trends in rural nonfarm employment. They conclude that over the past 20 years rural nonfarm employment has been 66 growing more rapidly than agricultural employment in all regions of the world, including Africa. Therefore, they reason, the share of nonfarm activities in total rural employment has been growing as well. Although both sets of authors acknowledge the frailty of the underlying data, their conclusions are comfortingly consistent with the view of a rural structural transformation involving increasing specialization and diversification out of agriculture. For individual countries, time series data on trends in rural nonfarm employment are limited to Sierra Leone, Rwanda and Kenya. All have their limitations but point in generally consistent directions. The Sierra Leone data are based on rigorous survey data of identical areas and firms, first in 1974 and subsequently in 1980 (Chuta and Liedholm, 1982). Unfortunately, sample sizes in the 1980 follow-up survey were not sufficiently large to permit inferences on trends in the most dispersed rural settlements, those below 2,000 in population. Nor do they include data for commercial, non-repair service firms, or for any enterprise employing over 50 workers. In practice the firm size cutoff should not pose a problem of incomplete coverage except in Freetown, the size category of cities over 250,000. The results from Sierra Leone, displayed in Table 27, lead to several important conclusions. First, repair services and food processing, have grown most rapidly, both overall and in small and medium sized rural towns. Second, while manufacturing in general appears to have declined in the smallest - the 2,000 to L0,000 person - localities, tailoring and welding have held constant; and they, along with carpentry, have grown 67 TABLE 27 TRENDS IN THE STRUCTURE OF NONFARM EMPLOYMENT, SIERRA LEONE, 1974-1980 Locality Size a a a 2,000-20,000 10,000-25,000 250,000+ 1. Repairs 15.0 5.2 15.0 a. Radio 22.0 19.0 b b. Motor Vehicle 17.0 2.4 20.0 c. Watch b b 13.0 2. Food Processing 14.0 33.0 21.0 a. Bakery 3.4 14.0 32.0 b. Other 39.0 c b 3. Woodwork 0.5 8.8 7.6 a. Carving 0.0 b 24.0 b. Carpentry -0.6 8.6 3.6 c. Other c b 0.0 4. Clothing -0.7 1.8 5.0 a. Tailoring 0.0. 3.8 4.9 b. Tie Dyeing -3.0 -8.5 b c. Shoemaking -4.3 16.0 6.2 d. Other b b 0.0 5. Metalwork -5.8 9.4 10.0 a. Welding 0.0 23.0 6.6 b. Blacksmithing -5.5- 1.8 22.0 c. Goldsmithing b b b d. Other 0.0 0.0 0.0 6. Other Manufacturing b 7.2 b Total Small Manufacturing and repair -2.4 6.0 5.7 aCompound annual rate of employment growth. bGreater than zero but less than one percent. cValue for 1974 is zero. SOURCE: Chuta and Liedholm (1982), pp. 104-105. 68 very rapidly in the medium sized rural towns, those between 20,000 and 250,000. These observations are consistent with the spatial evidence suggesting that repair services and food processing grow most rapidly with the development of rural towns. Tailoring, carpentry and metal working activity also increases, but more rapidly in middle-sized rather than very small rural settlements. The Rwanda time series data, tracing changes in rural nonfarm employment between 1978 and 1983, are more conjectural than those from Sierra Leone, since they splice together two very different data sets, the 1978 population census and the results of a more recent agricultural census. The timing as well as availability of information on part-time and seasonal labor differed among the two studies. And because occupational classifications also varied, the author (Vanvalli, 1985) had to make some judgements about how to aggregate and compare the two sets. Perhaps because of its part-time nature, home beer brewing was excluded from the manufacturing employment classification. The exclusion is unfortunate since beer brewing is by far the largest generator of rural nonfarm income, accounting for about 6.5% of total national GDP (Rwanda, 1982). Possibly the exclusion of brewing in both the beginning and ending years will leave the results of the remaining sectors valid. The great advantage of the Rwanda data is that they provide a comprahensive view of the rural economy, including commerce, services and manufacturing. With these caveats, the Rwanda data generate several interesting conclusions. First, they once again point to service, commerce and restaurants as the fastest growing nonfarm sectors. Construction has also 69 TABLE 28 TRENDS IN THE STRUCTURE OF RURAL NONFARM EMPLOYMENT, RWANDA, 1978-1983 Employment, Full-time Equivalents Compound Annual T&7 _ 1983 Growth Rate 1. Personal Services 20,164 49,061 19.5 a. Servants -- 31,246 b. Night Guards -- 17,815 2. Commerce, Restaurants and Hotels 14,957 31,339 15.9 a. Import-Export 142 -- b. Wholesaling 158 -- c. Retailing 10,761 15,954 d. Hotels, Bars, Restaurants 3,896 15,385 3. Construction 16,219 29,108 12.4 a. Builders 14,241 18,923 b. Plumbing and Water 247 3,000 c. Bridges and Roads 1,575 7,185 d. Other 156 -- 4. Manufacturing 29,144 48,647 10.8 a. Woodworking 6,439 17,492 --Carpentry -- 16,338 --Lumbering -- 1,154 b. Brick and Tile Manufacturing 14,014 12,215 c. Clothing 4,073 10,539 d. Metalwork 1,079 7,069 --Mechanics -- 3,062 --Blacksmiths -- 1,831 --Ferrailleur -- 1,185 --Bicycle Repair -- 985 e. Food Processing 3,019 1,338 --Butchers -- 1,338 f. Other 520 -- 5. Mining 11,389 11,000 0.0 6. Public Services 30,322 23,891 - 4.7 7. Transport, Storage 3,139 1,154 -18.0 Total Rural Nonfarm Employment 126,169 203,785 +10.1 SOURCE: Vanvalli (1985). 70 grown rapidly, perhaps because of major highway construction during 1983. Fourth fastest, but still apparently growing rapidly is manufacturing. Here, too, woodworking, metal work and textiles seem to have fared the best. The apparent decline in food processing, however, must be discounted because of the exclusion of home beer brewing. Time series data from Kenya are limited to th'e "informal sector" enterprise censuses undertaken regularly since 1978 and the Census of Manufacturers which collects data every five years on firms employing 10 or more workers. Both surveys are run by the Central Bureau of Statistics (CBS), but by different divisions within CBS. While the raw data from both sets can, in principle, be broken down by province, locality, and size of locality, knowledgeable observers indicate that there is substantial overlap in their coverages. Hence it is not possible to marry the two sets together without double counting firms at the small end of the formal spectrum. This difficulty, common in many countries, Drecludes generation of a reliable recent time series - or even overall profile - for either rural, urban or aggregate.nonfarm employment in Kenya. Restricting his investigations to the informal sector, Kilby (1987) has begun exploring trends in Kenyan rural and urban nonfarm employment. His initial findings suggest a .98 correlation between informal sector employment (both urban and rural) and wage employment. From this, he tentatively concludes that activity in small scale manufacturing, service and commercial establishments is driven by demand generated by low income wage laborers. 71 In a further interesting observation, Kilby calls attention to a spurt in informal sector employment in 1984/85. He suggests it may be due to the imposition in 1984 of a 17% sales tox which large firms find more difficult to evade than do small enterprises. Once again, the power of policy interventions becoales measurable. e. Conclusions Although some of the data are fragile, both the time series and cross section evideice point to several key conclusiona, First, services, commerce, and food )rocessing activities are the largest, the most rapidly growing, and the most ascendant nonfarm activities in emerging rural towns. Within manufacturing, basket making, weaving and spinning, pottery and wood carving consistently show signs of decline, both spatially and over time. But manufacturing activities other than food processing and preparation - especially carpentry, tailorinj and modorn metal working - appear to grow as well, by filtering up to regional towns and out of dispersed rural settlements. 4. Correlates of Growth in the Rural Nonfarm conomy Several features of the rural transformation clearly go hand in hand with growth in the economy's rural nonfarm component. First, the development of rural towns and the rise of nonfarm enterprises appear intimately intertwined. In part this may be due to the benefits of size - economies of scale and scope - as well as the prospects for centralized, cost effective provision of key produc. ive and in:titutional infrastructure. 72 The level of infrastructure - roads, telephones and 'telegraph, electricity, water, banking and transport facilities - emerges as a second likely stimulant to the growth of rural nonfarm enterprises. With the development of a cash economy, information, people, and funds must be transmitted across space in response to differential needs and productivities. Decreases in the cost of information and transport flows can only improve the efficiency with which rural labor and financial markets channel inputs into activities yielding the highest returns. Moreover, decreased transport costs open up rural resources and markets to viable exploitation. This can only facilitate movement to a more specialized, productive rural economy. Third, and also related, we expect that population density is. also positively correlated with the level of development of the rural nonfarm economy. Population density makes possible more rapid attainment of minimum efficient scales for full-time specialization in given activities. And emergence of a service sector depends on close physical proximity between purveyors and clients. Fourth we speculate that per capita income will be positively associated the level of rural nonfarm activity. Engel's Law mandate' this tandem movement for the overall economy, and if definitions of rural remain liberal enough to encompass the nonfarm growth in rural towns, then increases in income will be reflected in an increasing share of nonfarm in total rural income. Finally, and central for this review, is the relationship between agricultural production and nonfarm activity. For two reasons, we expect 73 growth in agricultural income per capita to be positively associated with increasing rural nonfarm activity. First, increased farm productivity is normally a precondition for rural specialization if labor is to be released to nonagricultural pursuits without lowering food production and intake in the rural community. And second, if rural growth multipliers exis.t, both the production and consumption linkages entrained by agricultural growth will stimulate the rural production of nonfarm goods and services in response. Consequently we hypothesize that, as in Asia, rising levels of rural nonfarm income will be in part caused by, and therefore correlated with, increases in farm income. Empirical evidence presented in the preceding sections has provided strong support for Hypothesis 1 concerning the importance of rural towns in the development of rural nonfarm enterprises. Tables 20 and 21 furnish particularly persuasive documentation of the strong relationship between rural towns and rural nonfarm enterprises. While data that would permit testing of Hypothesis 2 are not readily available, the example in Box 2 describes the key role rural towns and their infrastructure have played in the development of commercial gari processing in Ghana. The discussion in Box 2 draws heavily on a recent study by Kreamer (1986). Box 2. Gari Processing in Ghana: The Importance of Rural Towns in the Development of Rural Nonfarm Enterprise In less than ten years, Ghana's gari processing industry has surged to prominence in areas outside its traditional regional and ethnic strongholds. This rapid flowering - a response to successive years of drought - affords an unusual opportunity to view the stages and dimensions of the rural industrialization process that would 74 normally require review of a much longer time horizon. Initially, growers themselves turned to processing their cassava to preserve its value. This nonfarm household activity was incorporated into the combined farm and nonfarm labor demands of the family; it was seen initially as an activity ancillary to agriculture. Production was seasonal, and countercyclical to farm labor demands. The evolution to increasing degrees of specialization and stability of production is characterized by developments in .three basic areas: markets, organization and technology. The development of each of these requires the presence of critical volumes of users, suppliers and support systems, all of which which require separating the enterprise from the farm family and locating in areas of concentrated activity, namely rural towns. Access to key infrastructure - transport and piped water - as well as- organizational advantages of concentration proved particularly important to the emergence of gari processing in rural towns. Marketing and Organization In the case of gari, the process of specialization began with the move from grower-processors to independent processors. It was at this stage that raw material supply first appeared as a constraint, and the processors had to forego time in processing to travel about looking for supplies of cassava. Intermediaries emerged in the form of contractors whose function it was to assemble cassava in rural areas and bring it to processors. However, the volume of cassava required to support this function is much greater than the requirements of an individual processor. In the rural town, processors grouped themselves together into associations and appointed a representative, the secretary, who negotiates with contractors and oversees allocation of the cassava among members. Only with the concentration of processors in the rural town could the efficiency of raw material marketing be improved in this manner. Similarly, being located in a trade and transport center, processors have access to information concerning other markets and developments which affect their enterprise. Moreover, they gain access to key productive infrastructure, particularly piped water. Some processors, or grovos of processors througn contractors, become assemblers of larger quantities of gari for transport and sale in more distant markets. In the first instance, the benefit of this initiative accrues to the processors or contractors who initiated the entry into distant markets, but geographic integration of the production, pr.ocessing and distribution network in this manner progressively stabilizes supplies and prices. 75 Tech no-1ogy In the traditional methoQ ;f making gari, the only significant equipment is a rasp, against wihich the cassava is grated by hand, and a roasting pan in which the fermented and pressed cassava is cooked. Grating of the roots is very time consuming 'and tiring work, and this proved to be the first bottleneck to processors who had adequate supplies of fresh roots. The same associations which were formed to improve the raw material supply of individual processors also serve an important role in the adoption of technology. The power grater has capacity far in excess of the needs of any one processor, and costs more than most could afford; but 20 to 25 processors grouped together can pay a small fee each to an entrepreneur to gain access to a power grater for the twenty minutes or so that will save four hours of manual labor. It is also common for this miller/ entrepreneur to provide the custom services of a corn mill which is used to break up the lumps of intermediate product (Agbeliku and Gariku) which are otherwise normally thrown away. There is another important role of the rural town in the technical transformation described above: it provides access to and interaction with the small scale manufacturers and metalworkers who can convert ideas and models from other location into the capital equipment required for technical innovation. Even where size and cost do no dictate any form of association among processors, their proximity to entrepreneurs in the capital equipment field will have a direct bearing on innovation. For example, a simple hand screw press has gained widespread popularity among gari makers. It is a simple adaptation from another processing activity, but it reduces the ardor and time required to extract the fluid during fermentation of the cassava. This interaction between manufacturers and the user of the equipment is a process of adaptation and innovation which, left to itself, results in the designs and capacity that best reflect their combined constraints and opportunities. Thus the rapid recent expansion of Ghanaian gari processing took place in rural towns. The convergence there of infrastructure, technology and sufficient population densities enabled rapid response to new opportunities for rural nonfarm growth. ************************************************************************ A sufficient number of cross section observatibns have been assembled to permit at least a crude initial attempt at exploring the third, fourth and fifth hypothesized correlates of rural nonfarm growth - populution 76 density, per capita income and agricultural income per person. 7he plots in Figures 5, 6 and 7 display the relationships for the 15 countries for which data are available. The raw data appear in Table 16 and Appendix Table A.1. As would be expected given the differences in labor force definitions, and especially in the treatment of female employment, the data house considerable noise. And the narrow range of income and population densities lead to considerable bunching of the data, complicating inferences at higher income levels and population densities. Nonetheless, there does appear to be a positive relationship between rural nonfarm employment density and all three hypothesized correlates. The population density plots in Figure 5 show a clear upward trend with the exception of Rwanda, the outlier in the lower right corner. Its extremely high population density, unusual topography and dispersed settlement pattern make Rwanda's rural economy atypical among African countries. Rural population is dispersed evenly across the country's one thousand hill tops and the few settlements of any size - even rural towns of only a few thousand in population - have been officially classified as urban by Rwandan census officials. Consequently, Rwanda's RNFE densities appear artificially low at the same time population density, extremely high by African standards, and a marked lack of rural settlements make it very much different than other countries. Apart from Rwanda, we see a positive correlation between rural nonfarm activity and population density. Similarly, Figures 6 and 7 suggest a positive relationship between both per capita GDP a(A agricultural income on the one hand and RNFE FIGURE 5 RELATIONSHIP BETWEEN RNFE AND RURAL POPULATION DENSITY 190 - -~~-- -- 180 170 160 150 z 140 o 1 J F3O -D 120 CL o 110 S100 90 80- 70- 60 z 50 40 30 ¤ 20 10-¤ 0 20 40 60 80 100 120 140 160 180 RURAL POPULATION DENSITY (POP/KM SQ) FIGURE 6 RELATIONSHIP BETWEEN RNFE AND GDP PER CAPITA 190 -- 180- 170 160- 150 z 140-4 0 130 120 0 110 100D 0 9 0 0 90- 80 70 60 - 50 40 30 20 10 0 .10.3 0.5 0.7 0.9 1.1 1.3 1.5 (Thousands) GDP PER CAPIrTA ($1 980) FIGURE 7 REL ATION\1SHIP BETfWEEN PNFE- AND AGRICULTURAL GDP PER CAPITA ($1980) 190 - 180 170- 160- 150 140- 0 130 120 o 110 100 0 90 80 704 60 zi- 50-1 40-i 30 -LP 201 10-D 0 200 400 600 AGRICULTURAL INCOME PER CAPITA ($1980) 80 density on the other. The first provides corroboration of structural transformation and the effect of Engel's law, while the second supports belief in rural agricultural growth linkages. Because these cross-country comparisons are obscured by differences in definitions of rural, differences in measured female participation and different policy environments, a clearer way to isolate posited relationships would be to compare data among rural regions within a single country. We are fortunate to have the necessary regional breakdowns for rural Togo and rural Sierra Leone (ILO, 1982e; Matlon et.al., 1979). Displayed in Figures 8 and 9, they allow us to directly assess the relationship between farm and rural nonfarm income. The clear upward trend in Togo suggests that higher levels of nonfarm income are found in more prosperous agricultural regions. And the Sierra Leone data show a similar positive relationship between farm and nonfarm income with the exception of one outlier, a coastal region in which nonfarm fishing income accounts for over a third of total rural income. That resource based outlier apart, both countries show a strong correlation between per capita farm and nonfarm incomes across regions. Because nonfarm activities may be sustained by forces outside of agriculture - government or natural resources, for example - these correlations do not prove a causal link between growth in agriculture and growth in RNFE, but they are at least consistent with the belief in strong farm-nonfarm growth linkages. FIGURE 8 REGIONAL CORRELATIONS BETWEEN RURAL NON-FARM AND AGRICULTURE INCOME PER CAPITA TOGO 1980 R N 25- F 244 23 I 22- N 21- C 20 091 m ++ M181 E 17- co P16- E 15 R 14 + 13 C 12- A 1 T 8 Å 6- 0 5 0 4 0 3 2- C F 0 0 10 20 30 40 RURAL FARM INCOME PER CAPITA (000 CFA) FIGURE 9 REGIONAL CORRELATIONS BETWEEN RURAL NON-FARM AND AGRICULTURE INCOME PER CAPITA SIERRA LEONE 1975 R 301 + N F I N V C 0 M E 20 + P E + + R c+ A P I T L 0 N E S -- _ _ _ _ _ _ T_+ + 0 10 20 30 40 50 60 70 80 90 ±00 RURAL FARM INCOME PER CAPITA (LEONES) J j + 83 B. Magnitude of the Farm-Nonfarm Linkages As a prerequisite for estimating the magnitude of agricultural growth multipliers, as well as an aid in gaining a fuller understanding of the nature of farm-nonfarm linkages, it is useful to examine in some detail available evidence on the strength of intersectoral linkages in rural areas. Five different linkages are important, two in factor markets and three in product markets. The factor market linkages involve capital and labor flows between agriculture and nonfarm enterprises. Product markets inclu.de backward production linkages from agriculture to nonfarm equipment suppliers, forward production linkages from agriculture to processors and distributors, and consumer demand linkages generated as a result of increasing farm incomes. 1. Capital flows a. Investment of agricultural surplus in nonfarm activitiess Although capital flows clearly move in both directions - from farm-to- nonfarm and from nonfarm-to-farm enterprises - most observers believe the outflow from agriculture to be the larger. Certainly at an aggregate level, a large accumulation of evidence suggests that surpluses have been consistently transferred out of agriculture through fiscal, crop pricing and trade policies (World Bank, 1981). Marketing boards and export levies typically tax away 30 to 50% farmer cash crop prices (World Bank, 1981). Because estimating the magnitude of total effective agricultural taxation remains a complex undertaking, few such calculations have been undertaken anywhere. While Lee (1971) and Ueno (1985) have computed the rates of resource extraction from agriculture in Taiwan and Japan, Sharpley (1981) 84 provides the only African evidence. Using data from 1969 to 1975, she finds steadily increasing government resource extraction from agriculture, resulting in 1975 in agriculture supplying 76% of the country's gross capital formation. Given the structure of government expenditures, it is very likely that most African governments have transferred these agricultural surpluses primarily to urban areas and into nonagricultural activities. And of course, many observers remain concerned about the productivity the investments made with those transfers (World Bank, 1981, 1983; Nellis, 1987). Private investors also channel investment funds from agriculture to nonfarm activities. As shown in Table 29. evidence from Kenya an,d Sierra Leone suggests that agricultural surpluses account for between 15 and 40% of nonfarm investment funds. But since most survey teams do not explicitly ask rural entrepreneurs the sectoral origin of their investment funds, that figure may well be higher. Anecdotal evidence certainly suggests a frequent flow of funds from agriculture to nonfarm activities (Freeman and Norcliffe, 1985; Williams and McClintock, 1981). b. Investment of nonfarm surplus in agriculture Surpluses generated in nonfarm activities likewise furnish funds for productivity enhancing investments agriculture. Kitching (1977), in a review of sixteen farm management surveys from East Africa, finds repeated evidence of nonfarm earnings playing a crucial role in farmer acquisition of productive agricultural assets, especially land. The weight of the evidence, he believes, supports his hypothesis that previous nonfarm earnings is the single most important determinant of farm income. Collier 85 TABLE 29 SOURCES OF START-UP CAPITAL IN AFRICAN RURAL NONFARM ENTERPRISES (percent of enterprises obtaining capital fra each source) Own Savings Friends & Money Country Total of which from Agric. Relatives Lenders Banks Other Total Ghana, Rural Toins (1984) 0 -- 87 3 7 3 100 Kenya, RIDC Clients (1972) 100 (16) 0 0 0 0 100 Nigeria (a) Kwara State (1970) 96 - a 0 2 2 100 (b) Mid-West State (1970) 88 -- 10 1 a 1 100 (c) Western State (1970) 98 -- 2 a a 0 100 Sierra Leone (1974) (a) Rural Towns 61 (14) 21 0 0 18 101 (b) Rural Areas 65 (38) 9 3 0 24 101 Tanzania, 20 Toins .(1966/67) 78 -- 15 0 1 7 101 aLess than one percent. SOURCE: Ghana: Yankton (1985). Kenya: Child (1977). Nigeria: Aluko et al. (1972) and (1973). Sierra Leone: Liedholm & Chuta (1976). Tanzania: Schadler (1968). 86 TABLE 30 SOURCES OF WRKING CAPITAL IN AFRICAN RURAL NONFARM ENTERPRISES (percent of enterprises obtaining capital from each source) Reinvested Earnings Rotating Friends Personal Money Credit and Country Savings Lenders Banks Societies Relatives Other Total Kenya, RIDC Clients (1972)a 86b 0 0 0 0 14 100 Nigeria, Food Marketeers (a) Anaffbra State (1981/82) 12 11 3 12 21 410 100 (b) South Western (1981) 39 2 d 36 15 8 100 Sierra Leone (1974) (a) Rural Towns 73 0 0 0 0 27 100 (b) Rural Areas 77 12 0 0 0 11 100 aCapital for expansion. b24 percent of which from reinvested earnings plus savings from agriculture. cIncludes loans from crops and trade associations. dLess than one percent. SOURCE: Kenya: Child (1977). Nigeria, Ananera: Adeyemo (1935). Nigeria, Southwest: Olupokawbi (1981). Sierra Leone: Liedholm and Chuta (1976). and Lal (1986) reach a similar conclusion. They infer from Kenyan data that off-farm earnings, primarily urban wages, are the major correlate of agricultural innovation among farm families, presumably because nonfarm earnings supply funds for productive investment in agriculture. Berry (1975) reports a similar flow of nonfarm earnings playing an important role in the expansion of Nigerian cocoa farming in the 1930's and 40's. Likewise, Haugerud (1984) finds that nonfarm earnings are a key determinant of investment in Kenyan agriculture. And Massel and Parnes (1969), in explaining the extraordinarily high nonfarm earnings they found in Ugandan farm families, nonfarm earnings on the order of 65% of total household income, report similar conclusions. They indicate that a number of local businessmen had recently expanded from nonfarm business into agricultural enterprises, financing the agricultural expansion with nonfarm earnings. Because the farmers remained active in both sectors, nonfarm business accounted for a preponderant share of household income. The bulk of this admittedly fragmentary evidence comes from East Africa and appears frequently tied to investment in cash tree crops. But it does show that nonfarm to farm investment flows do occur w.ith some regularity. Ultimately, the fungibility of funds makes the question of source of investment monies irrelevant unless it can be argued that timing of nonfarm revenue availability makes them uniquely suited for investment transfers to agriculture. This may be the case for urban wage earnings, but since peak farm and rural nonfarm earnings coincide during the post harvest period, it is difficult to ascribe temporal liquidity advantages to rurai nonfarm surpluses. 88 Because funds are fungible, probably the best way to measure the relative investment contributions from farm and nonfarm sources is to measure the contribution of each to rural cash incomes. Since the income shares from Table 3 indicate roughly a 30-70 split, it is probable that nonfarm earnings generate about 30% of rural investments, while agriculture shoulders the remainder of the load. Thus, while th.e nonfarm- to-farm flows may be very important in certain instances, the farm-to- nonfarm flows likely predominate. 2. Labor Flows Labor in rural Africa regularly flows back and forth between farm and nonfarm activities. As Figure 1 indicates, and a host of other studies corroborate, nonfarm labor inputs move contracyclically to demands of the agricultural calendar. Competition for labor and the post harvest peak in nonagricultural demand combine to enforce this juxtaposition of peaks and troughs. Tables 1 and 2 indicate that agriculture employs between 80% and 90% Africa's rural labor force. Nonfarm activity provides primary employment for the remaining 10% to 20% of rural laborers, as well as secondary employment for a further 20% to 30%. Combining these data with estimates of secondary employment in agriculture, we have estimated earlier (Section III.A.2) that 30 to 60% of the rural labor force works in both farm and nonfarm activities. This range represents the sizeable magnitude of the labor flows moving back and forth between the rural farm and no.nfarm sectors. 89 Our earlier estimates suggest that about 20% of all labor flows pass through labor markets, including no more than 5% through communal labor exchanges. The remaining 80% of all rural labor is allocated within families by household heads. 3. Backward Linkages from Agriculture to Rural Input Suppliers a. Current backward linkages Agriculture requires production 'inputs, some of which can be supplied by rural enterprises. The Sierra Leone seasonal data clearly indicate that blacksmithing activity is at its peak during the height of the agricultural season because of the demand for repair services (Liedholm and Chuta, 1976). And the locality size data in Table 23 also indicate a predominance of blacksmithing in the communities below 2,000 in population, presumably because of the necessity of being near the farmers who require the repair services. The type and magnitude of backward linkages varies depending on agricultural technology, size of holding, type of crop, and whether production is irrigated or rainfed. Appendix Table A.4 gives a feel for the range of differences among hand hoe agriculture and animal traction, small and large holder farms, tree crops and annuals, irrigated and rainfed cultivation. In general the backward linkages appear to be weak, in all likelihood far weaker than those measured in Asia. To see why, recall that Johnston and Kilby (1975), in evaluating the magnitude of backward linkages in Asia, identify fertilizer, followed by equipment and finally cement and building materials as the three key production inputs to 90 agriculture. Fertilizer, by far the largest of these inputs in Asia, requires petroleum, huge amounts of capital and a large market to service. In Africa, probably only Nigeria can aspire to fertilizer production, while in the much larger Asian countries it may more frequently be possible. Of course, even where they exist, fertilizer production linkages are far more likely to be urban than rural. Moreover, the second and third production linkages, cement plus building materials, and farm implements, are also likely to be less powerful in Africa than in Asia. Topography and hydrology severely limit irrigation potential in Africa. Shallow river basins coupled with low population density limit the potential for cost effective irrigation in most of Africa (Delgado, 1986). Hence the canal building and maintenance demanded in Asia is not required in Africa. The famous rural pump, engine and tubewell industries which blossomed in the Punjab are not likely to be induced in Africa where that equipment is simply not needed (Child and Kaneda, 1975). Furthermore, animal traction and other farm implements are less prevalent in Africa than in Asia. Hand hoe agriculture prevails in the humid subtropical regions where tse tse flies prevent animal traction. Even in arid West Africa, where disease does not threaten livestock, only about 10% of all households practice animal traction (Mellor, Delgado and Blackie, 1987; Ohm, Nagy and Sawadogo, 1985). Of course temperate climate and history lead to far higher rates of animal traction - and even mechanization - in East and Southern Africa. The equipment inventories displayed in Table 31 indicate that the prevalence of animal traction 91 TABLE 31 INVENTORY OF SMALL FARMER EQUIPMENT IN SUB-SAHARAN AFRICA: PERCENT OF FARM HOUSEHOLUS OWNING EACH ITEM Botswana Gambia Tanzania Zimbabwe 1975-76 1981 1981 1982 Small hoe -- -- 97 -- Large hoe -- 35 -- Axe -- -- 87 -- Machete 83 -- Plow 69 44 -- 83 Planter 6 73 -- 8 Cultivator 4 40 -- 54 Wagon or cart 11 24 -- 43 Tractor -- -- 5 aThree Rice Growing Regions. bEight Villages. cTwo Regions. SOURCE: Botswana: Chernicovsky, Meuller and Lucas (1985); Gambia: ILO (1985c); Tanzania: Oates (1984; and Zimbabwe: ILO (1985g). -92 rises to between 50 and 90% in those regions. Debates continue over the profitability of both mechanization and animal traction (Sargeant et.al. 1979; Jaeger, 1986; Johnston, 1981; Binswanger, 1982 and 1985). b. The future Of course, the potential for production linkages lies not necessarily in todays-technology but in what will be required to generate the first round of agriculturally propelled growth tomorrow. Much like the two shoe salesmen visiting a country of barefoots, two very different conclusions can be drawn from the current low rate of product utilization. The optimistic salesman concludes that the market potential is huge, while the other dismisses prospects as dismal citing the absence of a proven clientele. Johnston (1981) stands most visible among optimists. Evaluating the prospects for agricultural equipment manufacture in Kenya he states, "The magnitude of the potential market is suggested by the fact than an estimated 84% of Kenya's arable area is still cultivated with hand tools compared to 12 percent cultivated with oxen and 3.5% by tractors." But a close reading of even his review of prospects for equipment manufacture in East Africa reveals that his only examples come from China and Taiwan. The question of what tomorrow's technology will look like is difficult to answer definitively. It is increasingly clear that peak season labor bottlenecks constrain output in most areas, so some form biological or mechanical solution will have to address that problem area (Eicher and Baker, 1982; Mellor, Delgado and Blackie, 1987). If the most economic solutions turn out to be mechanical, backward nonfarm linkage 93 multipliers may grow. The multipliers computed below use a range of estimates of equipment inputs to generate a sense of the magnitude of the growth linkages that can be reasonably expected from this quarter. c. Idiosyncratic design adaptation Over and above the employment and income multipliers, Kilby and Liedholm (1986) propose one further virtue of rural nonfarm equipment suppliers. Using examples from India, Taiwan and Thailand, they point to prospects for "idiosyncratic design adaptation", the tailoring of equipment to local soil and topography. They describe how interaction between farmers and rural blacksmiths led to development of Thailand's Prapradaeng power tiller and a wide array of highly specialized agricultural implements in Taiwan. While government sponsored implement research is underway in numerous African countries (Eicher and Baker, 1982; Johnston, 1981; World Bank, 1986), only in Ghana do we have tentative evidence of private sector design adaptation for agriculture. A rotary punch planter for maize was apparently successfully adapted from a prototype obtained from the International Institute for Tropical Agriculture in Nigeria (Quansah, 1985). Even here, a technical institute in Kumasi appeared to play a key role. Much better documented are examples of successful innovation on behalf of rural nonfarm enterprises. Kilby (1962) describes how experimentation around 1920 by a Nigerian baker and metal worker led to domestic production of a dough brake at 20% of the cost of the imported article, launching development of the entire Nigerian bread industry. Kreamer (1986) describes how similar interaction between gari processors -94 and metal workers in Ghana led to development of low cost presses as well as simple mechanical graters, both of which were instrumental in expansion of the industry (See Box 2). Analagous efforts have led to development of inexpensive strainers used by home beer brewers in Botswana (Haggblade, 1984). 4. Forward Linkages from Agriculture to Processors and Distributors The forward linkages from agriculture to local processors and distributors appear to be far larger than those on the input side. Table 32 attempts to break out current rural nonfarm activity into three categories - those activities unrelated to agriculture, those supplying inputs and those processing outputs. Because many firms service a multiplicity of clients, this is not possible to do with great precision. Metal workers, for example, fix bicycles, automobile parts, and manufacture gates, door hinges and bars for windows. The problem is most severe with. commerce. Much rural commerce clearly involves distribution of domestically produced agricultural goods. But a large share also distributes goods imported into the rural regions, both consumer goods as well as equipment. Three conclusions emerge from Table 32. First, input supply is far smaller than the forward processing linkages from agriculture. Even if all metal working, blacksmithing and metal repair were related to agriculture, the forward processing and distribution links would be at least 10 times as great. Food processing is clearly important. Although bread baking is based on imported wheat flour, virtually all other processing activities involve transformation of local agricultural 95 TABLE 32 PROBABLE RELATIVE MAGNITUDE OF BACKWARD AND FORWARD AGRICULTURAL LINKAGES IN RURAL AFRICA 1. Backward Linkages from Agriculture Distribution Implement I pl ement Imported 1. Z'abia (1985)a Manufacture Repair Inputs Total a. Dispersed Settlements 3.21 -- 3.21 b. Rural Towns - 12,000 1.22 1.22 c. Rural Towns - 12,000-50,000 1.39 -- 1.39 d. Total Rural 2.68 -- 2.68 Sierra Leone (1974)a a. Settlements under 2,000 5.52 -- 5.52 b. Rural Towns, 5,000-20,000 .73 -- .73 Kenya (1977)b Central Province .23% 5.52% 2.11% 7.86% 2. Forward Linkages from Agriculture Distribution Food Leather- and 2. Zairbia (1985)a Processing Restaurants Butchers work Storage Total 2/1 a. Dispersed Settlements 44.02 0.00 -- .37 4.23 46.82 15 b. Rural Towns - 12,000 35.25 1.79 -- .77 25.70 63.51 52 c. Rural Towns - 12,000-50,000 19.86 ).30 -- .63 25.80 47.59 34 d. Total Rural 37.97 0.20 -- .38 6.60 45.15 17 Sierra Leone (1974)a a. Settlerents under 2,000 0.00 b. Rural Towns, 5,000-20,000 0.00d Kenya (1977) Central Province 5.86% ---4.34 3.75e 5.05 19.00 2.4 96 TABLE 32 (CONT.) Six Provinces: Employment per 1,000 population. b Percent of rural nonfarm enterprises. cDistribution and trade estimated equal to locally produced foods share of total rural cash expenditure. Taking 30 percent as a conmn share of cash in total food expenditure, and a total (cash plus subsistence) average budget shares of .70 for locally produced food and .05 for imported food, yields a 45 percent share of locally produced foods in total rural cash expenditure. d0nly baking enufferated among food-processing activities. Baking is based on imported wheat flour. SOURCE: Zambia: Milio and Fisseha (1985); Sierra Leone: Liedholm and Chuta (1976); Kenya: Freeman and Norcliffe (1985). 97 production. Brewing is the largest nonfarm employers in many of the countries for which we have evidence, for example Zambia, Burkina, Botswana and Rwanda. Milling is also consistently important. Oil extraction, sugar production, tea drying and packaging, coffee and cocoa dehulling and drying are often performed in rural areas by large scale enterprises. And while we know these large scale activities take place predominantly in rural areas, the small scale orientation of much of the past research causes those activities to elude the statistical net. Hence even substantial figures in Table 32 understate the magnitude of rural agroprocessing. Distribution of agricultural products generates undoubtedly the largest of the nonfarm production links to agriculture. .Retailing is a labor intensive activity which accounts for a majority of female employment in much of West Africa (see sources to Tables 1 and 2). If retailing of local agricultural produce is approximately proportional to the share of marketed production in total rural consumer cash expenditure, then about 45% of rural retailing can be assumed to be a forward distribution link from agriculture. That would put the forward linkages over 10 times the backward links imply that roughly two-thirds of the forward links come from distribution, the remainder from processing. 5. Consumption Links In a dynamic situation where farm incomes are increasing, the consumer demand linkages emanating from these increases have the potential to be an important stimulant to the growth of the rural nonfarm economy. As per capita incomes rise, the demand for local services (especially transport, personal services, home improvement and religious functions) 98 and higher quality foods (especially livestock and horticultural products) typically increases more rapidly than the demand for foodgrains. The Asian experience suggests that the production of these commodities and services is labor intensive, hence rural employment in the non-foodgrain sector increases quite rapidly with per capita farm incomes (see for example, Mellor 1976; Gibb 1974; Bell, Hazell and Slade 1982). These patterns of growth may depend on a combination of high population density and adequate transport facilities to make rural services accessible to the villages, and so may be less applicable to large parts of Sub-Saharan Africa. One might hypothesize that in the absence of relatively cheap, labor-intensive goods and services in rural areas, then as per capita incomes increase, demand will shift less rapidly-from foods to services in Sub-Saharan Africa than in Asia, with consequent weaker demand linkages to the rural nonfarm economy. Although there have been numerous household expenditure surveys in Africa (See Table A.3), these are generally difficult to compare because of differences in the way income or expenditure is measured, differences in household definitions and the time period covered, differences in commodity groupings, and differences in functional forms used to estimate marginal propensities and income elasticities. In addition, few studies compute any measures other than average budget shares. Of the available studies, the most comparable were undertaken for farm households in Gusau, Northern Nigeria (Hazell and Roell 1983), rural households in Sierra Leone (King and Byerlee 1977), and households in three villages in Zaria Province, Northern Nigeria (Simmons, 1976). These are summarized in Table 99 33 together with comparable results -or farm households in the Muda region of Malaysia (Hazell and Roell 1983) and North Arcot district, South India (Hazell and Ramasamy, 1986). Expenditures patterns in Gusau, Northern Nigeria and rural Sierra Leone are very similar. The three villages Simmons studied in Zaria Province were deliberately chosen because of their close location to the urban center of Zaria, and as a consequence they have easier access to, off-farm employment and to a wider range of consumer goods and services than are available in most rural villages in Africa. Not surprisingly, the expenditure patterns in these villages are quite different from those in Gusau and Sierra Leone, with greater diversification of the budge into nonfoods, especially services. Unfortunately, data on consumption of manufactured goods are not reported. While there is a basic similarity in tne breakdown between food and nonfood aggregates in Gusau, rural Sierra Leone and North Arcot, the most significant difference is that the South Indian households allocate substantially larger shares of incremental income to services and social and religious obligations, largely at the expense of food. This difference provides support for the hypothesis that services are more cheaply available in Asian countries, though a more rigorous test would also require allowing for possible differences in per capita income levels. The Muda region is visibly richer than Gusau, rural Sierra Leone or North Arcot. 'This is reflected in the more inelastic demand for food and the greater diversification of the average and incremental budget into TABLE 33 RURAL HOUSEHOLD EXPENDI111RE BEHAVIOR IN SELECTED COUNTRIES Average Budget Share Marginal Budget Share Expenditure Elasticities Rural Zaira North Rural Zaira Nurth Rural Zaira North Gusau Sierra Province Mudiz Arcot Gusau Sierra Province Muda Arcot Gusau Sierra Province Muda Arcot N. Nigeria Leone N. Nigeria Malaysia S. India N. Nigeria Leone N. Nigeria Malaysia S. India N. Nigeria Leone N. Nigeria alaysia S. India Cawxodity Group Food Alcohol and Tobacco 80.7 73.7 56.5 66.7 78.2 76.1 67.9 37.7 63.0 0.94 0.92 0.57 0.81 Clothing and Footwear 7.2 7.0 11.4 5.8 4.2 8.9 7.4 8.1 7.7 1.24 1.06 1.39 1.85 Consumner Expendables 4.3 -- -- 3.4 3.1 4.4 3.7 2.4 1.02 -- 1.09 0.77 Housing 0.3 -- 2.6 4.1 n.a. 0.4 12.4 n.a. 1.40 -- 1.02 n.a. Transport 1.9 2.2 1.3 1.8 2.8 2.7 3.0 3.1 3.4 -- 1.41 1.36 1.67 1.22 Durables 1.1 -- 2.1 0.6 1.4 7.1 1.9 1.25 3.35 3.43 -- Education and Health 1.1 1.4 -- 2.9 1.9 1.6 0.8 5.2 2.4 1.42 0.57 1.79 1.26 D Services and Social and Religious Oblications 3.3 4.3 13.5 13.1 9.1 4.4 8.1 22.7 19.3 1.33 1.88 1.73 2.12 Locational Group Locally Produced Foods 75.3 69.0 -- 46.4 63.0 70.3 66.1 24.6 48.5 0.93 0.96 0.53 0.77 Nonfoods 8.4 8.8 -- 18.1 17.4 11.3 12.3 36.9 30.8 1.34 1.40 2.05 1.77 Regional,Imports Foods 5.4 -- -- 20.3 12.3 5.8 -- 13.1 12.0 1.07 -- 0.65 0.98 22.2 21.6 0.97 Nonfoods 10.9 -- -- 15.3 7.4 12.6 -- 25.4 8.6 1.16 -- 1.66 1.17 Nontradables 24.7 -- -- 23.5 17.6 32.0 -- 40.6 24.1 1.30 0.87-1.62 1.73 1.37 SOURCE: Gusau, Northern Nigeria, 1976-77 (Hazell and Roell 1983). Sierra Leone, Rural (King and Byerlee 1977). Zaria Province, Northern Nigeria, 1970-71 (Sianmons 1976). Muda, Malaysia, 1972-73 (Hazell and Roell 1983). North Arcot, South India, 1982-83 (Hazell and Rainssany 1986). 101 nonfoods. Note also the similarity of the importance of services in the budget to North Arcot and Simmon's three villages near Zaria. Again, the ready availability of labor-intensive services may be an important determinant of revealed expenditure behavior. In the lower part of Table 33, commodities have been grouped by their location of origin. Tradables and nontradables are defined with respect to the regions studied; they are not necessarily tradables or nontradables in the international markets. To a regional economy, local household expenditures on imported goods (from outside the region) represent a direct leakage which reduces the size of the local income and employment multipliers. But as Siamwalla (1982) has argued, if incremental income is spent on locally produced goods that could be exported from the region at a constant price, that expenditure repres.ents a loss in export proceed and is as much a leakage as if the money were expended upon imported goods. Thus the stimulative effect of increased household expenditure on the local eccnomy depends crucially on the expenditure share allocated to locally produced goods and services that are also nontradables. The average budget share for regional nontradables is about the same in Gusau as in Muda, and a little higher than in North Arcot. At first blush, this would suggest that the stimulatory effect of the consumption linkages on the local nonfarm economy should be about the same in the three regions. However, local nonfoods are about twice as important in the composition in the two Asian cases. The difference is even more exaggerated in the marginal budget shares; the share of incremental income allocated to local nonfoods is about three times larger in Muda and,North 102 Arcot than in Gusau. The results for rural Sierra Leone also seem to confirm that the demand for local nonfoods is greater in Asia than in Africa. Consequently, whether the demand linkages for local nontradables are as important in Africa as in Asia for the generation of indirect rural employment seems to depend critically on the labor intensity of the local nontradable foods demanded. In Asia, these products (mostly livestocks and horticultural crops) tend to be labor-intensive and remunerative to the poor. In Gusau and Sierra Leone there is little direct evidence, but certainly livestock are kept largely under extensive, if not nomadic, grazing systems which are not noticeably labor-intensive. However, these are important issues which need to be examined with more detailed studies than can be mustered here. The importance of the avai-lability of nonfood goods and services as a determinant of the household expenditure patterns becomes even clearer when the expenditure behavior across income deciles within countries is compared. Tables 34 and 35 summarize the relevant findings reported by Hazell and Roell for Gusau and Muda. In Muda, the marginal budget shares for all food groups decline sharply between the bottom and top per capita expenditure deciles. The marginal budget share for total food, alcohol, and tobacco declines from 67 percent for the bottom decile to 14 percent for the top decile. The decline is particularly sharp for cereals and cereal products, and hence, for home-produced foods. In contrast, the marginal budget share for total food, alcohol, and tobacco does not change significantly between expenditure deciles in TABLE 34 MARGINAL BUDGET SHARES BY PER CAPITA EXPENDITURE DECILE IN MUDA, 1972/73 (in percents) Per Capita Expenditure Decile Group 1st 2nd 3rd 4th 5th 6th 7th 8th 9th 10th Commodity group Food, alcohol, and tobacco 67.39 57.94 51.98 49.61 45.57 41.88 37.63 35.27 27.71 13.89 Cereals and cereal products 21.88 18.42 15.64 15.34 13.41 12.27 10.39 9.43 6.22 10.53 Fruits, vegetables, and nuts 9.85 8.75 8.30 8.02 7.60 6.98 6.79 6.49 5.94 4.42 Meat and fish 12.19 10.54 9.52 9.11 8.52 8.17 7.24 7.16 5.73 3.46 Eggs and dairy products 2.61 2.38 2.14 2.01 1.85 1.76 1.57 1.52 1.42 0.96 Clothing and footwear 7.92 8.20 8.33 8.38 8.14 8.07 7.94 7.70 7.87 7.44 Consumer expendables 4.58 4.25 4.10 4.13 3.95 3.79 3.67 3.59 3.52 3.10 Housing 2.51 5.50 7.88 7.64 10.06 10.55 12.35 13.76 15.47 20.29 Transport 2.33 2.53 2.77 2.82 2.84 2.96 3.09 3.01 3.26 3.58 C Durables -1.01 1.70 2.83 4.05 4.95 5.94 6.85 8.00 9.82 13.22 Education and health 2.22 3.16 3.31 4.12 4.05 4.89 5.17 5.45 6.53 7.71 Personal services and entertainment 1.65 1.99 2.33 2.12 2.39 2.40 2.36 2.39 2.63 2.89 Social obligations 12.41 14.73 16.45 17.12 18.06 19.53 20.94 20.82 23.18 27.89 Locational group Food Homve produced 22.47 18.36 15.44 15.54 13.52 12.54 10.22 9.71 5.34 -1.17 Locally produced 21.40 19.47 18.11 '16.99 16.05 15.07 14.29 13.55 '12.75 9.88 Imported 23.53 20.11 18.43 17.08 16.00 14.27 13.12 12.02 9.62 5.18 Nonfood Locally produced 17.87 23.50 27.70 28.74 32.06 34.21 37.36 38.91 43.06 52.79 Imported 14.74 18.56 20.31 21.64 22.37 23.90 25.01 25.82 29.23 33.32 Nontradables 23.72 28.64 32.42 33.35 36.30 38.12 41.07 42.47 46.18 54.92 Average farm size (acres)a 2.14 2.33 3.02 3.12 3.33 4.14 4.08 4.00 4.50 5.64 Average family size 7.07 6.64 6.42 5.90 5.45 5.43 5.27 4.66 4.65 3.89 Per capita expenditure (M$) 150.00 197.00 225.00 255.00 289.00 327.00 369.00 419.00 514.00 820.00 aFarm area is the operated paddy area. NOTE: All household characteristic variables are evaluated at decile means. SOURCE: Hazell and 1oell (1983). TABLE 35 MARGINAL BUDGET SHARES BY PER CAPITA EXPENDITURE DECILE IN GUSAU, 1976/77 (in percents) Per Capita Expenditure Decile Group 1st 2nd 3rd 4th 5th 6th 7th 8th 9th 10th Canodity group 'Food, alcohol, and tobacco 77.88 76.74 76.75 76.89 76.45 76.26 76.18 75.78 76.15 75.89 Cereals and cereal products 50.99 44.40 42.07 39.89 37.86 35.40 34.56 31.49 30.45 26.48 Fruits, vegetables, and nuts 4.57 5.94 6.71 7.61 8.18 8.59 9.56 9.68 10.41 11.72 fMat and fish 7.22 9.44 11.44 11.03 11.32 11.44 12.70 12.72 12.80 13.58 Eggs and dairy products 7.32 8.50 7.57 8.77 9.28 10.73 8.35 11.00 11.39 12.28 Clothing and footwear 8.87 9.08 9.31 8.94 8.98 8.96 8.85 8.92 8.56 8.31 Consuner expendables 4.23 4.33 4.44 4.45 4.47 4.39 4.72 4.49 4.57 4.66 Housing 0.45 0.48 0.45 0.41 0.43 0.46 0.34 0.42 0.37 0.32 Transport 1.49 1.98 2.04 2.27 2.52 2.77 2.62 2.88 2.95 3.16 Durables 0.77 1.21 1.36 1.33 1.33 1.37 1.54 1.65 1.71 1.91 Education and health 1.41 1.57 1.57 1.53 1.57 1.59 1.57 1.64 1.64 1.68 Personal services and entertainment 1.34 1.21 1.09 1.08 1.04 0.99 1.13 1.07 0.99 1.00 Social cbligations 3.57 3.39 2.99 3.09 3.20 3.22 3.06 3.15 3.06 3.06 Locational group Food Home produced 59.15 53.55 50.05 50.82 50.00 50.08 46.16 46.60 46.18 43.82 Locally produced 14.49 18.62 21.73 20.63 20.92 20.47 23.62 23.01 23.59 25.22 Imported 4.23 4.57 4.97 5.44 5.53 5.70 6.39 6.18 6.39 6.85 Nonfood Locally produced 10.95 11.05 10.56 10.73 11.02 11.21 10.93 11.24 10.94 10.96 Imported 11.17 12.21 12.69 12.38 12.53 12.53 12.90 12.98 12.91 13.14 Nontradables 26.96 29.22 27.61 29.29 30.49 32.06 30.11 33.09 33.75 35.61 Average farm size (acres)a 8.52 10.98 9.03 7.94 10.08 8.21 7.63 9.02 10.23 9.82 Average fanily size 12.52 10.48 7.66 7.31 7.66 7.69 5.62 6.24 5.45 4.61 Per capita expenditure (M$) 42.00 62.00 73.00 84.00 96.00 107.00 120.00 140.00 163.00 221.00 aFarm area is the operated paddy area. NOTE: All household characteristic variables are evaluated at decile means. SOURCE: Hazell and Rell (1983). 105 Gusau. There is a sharp drop in the share allocated to cereals and cereal products, but this is offset by increasing marginal budget shares for fruits, vegetables, and nuts; meat and fish; and eggs and dairy product. As in Muda, the marginal budget share for home-produced foods declines between the bottom and top expenditure deciles. There is no significant change in the margi-nal budget shares for clothing and footwear or consumer expendables between expenditure deciles in either region. But in Muda the marginal budget shares for all other nonfoods increase significantly between the bottom and top expenditure deciles, especially for housing, durables, social obligations, and the aggregate of locally produced nonfoods and nontradables These demand patterns are absent in Guasu, and there are no significant changes in the marginal budget shares for any nonfood category between per capita expenditures deciles. Only nontradables show any increase across the income deciles, and this is because of an increase in the nontradable foods consumed. These results correlate strongly with the degree of access to market centers and urban areas in the two regions. Unlike Muda, which is densely populated and has an excellent transportation system, the villages in Gusau are typically much more removed from market centers and are faced with very limited means of transport. There may simply be little else to buy other than food without an arduous and time-consuming trip to the nearest town. Also, Hazell and Roell suggest that noncereal foods are considerable more expensive relative to cereals in Gusau than in Muda, and a larger share of the budget is required simply to enable the richer 106 households to diversify their diets. For example, the relative price of beef to sorghum or millet is about 10:1 in Gusau. In contrast, the relative price of fish (the main beef substitute) to rice is only 5:1 in Muda. King an-d Byerlee's study of rural household expenditure behavior in Sierra Leone provides a useful comparison to the Muda and Gusau results. Figure 10 summarizes their results on rural demand linkages by per capita expenditure decile. The size of the marginal budget share for rurally produced nonfoods is about the same in rural Sierra Leone as in Gusau-- about 12 percent for the average household. However, unlike Gusau, this marginal budget share increases between the lower and upper expenditure deciles. But this encouraging Muda-like pattern is tempered by the observation that hduseholds in the first three-income deciles buy -virtually no rurally produced nonfoods, and households in the top deci-le have a marginal budget share for rural nonfoods that is only half that of the average household in Muda. Access to rural nonfoods is clearly a problem, but the puzzle is why this is so much more pronounced for the poor. Table 36 provides additional details on the marginal budget shares from the King and Byerlee study of rural Sierra Leone. Two results are of some importance. First, the marginal budget share for rural nonfood goods is much smaller than the marginal budget share for rural services and ceremonies in all income classes; less than one-tenth as large. This confirms that the consumption linkages for rural services are much 107 FIGURE 10 Marginal budget shares of locally produced goods and services, by per capita expenditure decile. Sierra Leone. 1974/75 Marginal Budget Share t percent) 75 2so 0- 1st 2nd 4th 6th 8th iOth and and and and 3rd 5th 7th 9th Decile "I'r""l Rurally produced Rurally produced Home- produced nonfoods and purchased foods foods Source: R' P King and D Sverfee. Income Diestnbunon. Consurnmon Parrems and Consumpton Linkages in Rural Sierra Leone, African Rural Economy Paper 16 (East Lansing. Mich.. Michigan State Universitry. 1977). Table S 3 108 TABLE 36 MARGINAL PROPENSITIES TO CONSUME BY INCOME CLASS FOR COMMODITIES IN SIERRA LEONE Marginal Propensity to Consurre Incone Class Second Fourth Sixth Eighth Mean and and and and Expenditure Lowest Third Fifth Seventh Ninth Highest Origin Classification Level Decile Deciles Deciles Deciles Deciles Decile Rural subsistence food .445. .684 .588 .497 .436 .370 .286 Rural purchased food .216 .153 .178 .201 .218 .235 .257 All rural food .661 .837 .765 .698 .654 .605 .543 Rural nonfarm goods .011 .013 .012 .011 .011 .010 .009 Rural services and cererunial .112 -.036 .023 .080 .117 .158 .211 All rural products .784 ,814 .801 .789 .782 .773 .763 Rural town productsa .055 .034 .043 .051 .056 .062 .069 Urban productsb .020 .019 .019 .020 .020 .021 .021 ITported products .141 .133 .137 .140 .142 .144 .147 Total *1.000 1.000 1.000 1.000 1.000 1.000 1.000 aRural towns are those between 2,000 and 100,000 in population. Byerlee and King call them "srall urban" areas. Products include primarily transport, services and small industrial products. bUrban areas are all above 100,000 in population. SOURCE: King, R.P. and D. Byerlee (1977). 109 stronger than those for rural manufacturing, as posited elsewhere in this paper. Second, the marginal budget shares for products manufactured in small urban centers is about twice as large as the share fur products manufactured in large urban areas. Given that most services are provided at the local level, this supports the hypothesis that the consumption linkages are spatially focused on rural market towns rather than on larger urban areas or dispersed rural settlements. Income distribution also plays a potentially important role in determining the magnitude of consumption linkages. In situations where the marginal budget share for local nontradables differs by income group, the stimulatory effect of the consumption linkage on the local nonfarm economy will be affected by the way in which income is distributed. Based on the Muda results in Table 34, Haze.11 and Roell argued that the richer households, who also tend to have the larger farms, have the strongest demand links to the local economy, and hence the indirect benefits from agricultural growth will be larger if these households are the major beneficiaries of agricultural growth. They point out, however, that this conclusion might be tempered if a) the richer households had higher savings rates and their savings were not invested locally, b) the large farms were more mechanized and employed less labor per unit of land, or c) the region's export demand for foodgrains became inelastic. Similar conclusions are derived by Hazell and Ramasamy (1986) for North Arcot, though in this case the disparities in savings rates between small and 110 large farms are such as to neutralize the desirability of differences in expenditure behavior (Hazell, Ramasamy and Rajagopalan, 1987). These kinds of distributional considerations seem less relevant for Africa given the pattern of expenditure behavior reported for Gusau in Table 35. -Differences in the marginal budget share. for nontradables across income deciles are iufficiently small that t-hey might realistically be ignored. However, the Sierra Leone results in Figure 10 do suggest that income distributional considerations may become more important as village access to nonfoods becomes better. 6. A Caution In several cases, available evidence has suggested that African farm- nonfarm linkages are lower than those found in Asia. Backward production linkages are expected.to be lower because of an absence of irrigation inputs, lower levels of animal traction and lower population densities in Africa. Similarly, African marginal budget shares speft on nontradables average about 0.3, compared to the 0.4 measured in Muda. More alarming still, at the high end of the income distributions, they attain a maximum of 0.36 in Africa compared to 0.55 in Muda. Both average and highest decile marginal budget shares spent on nontradables appear to be over 30% higher in Asia than in Africa. Both lower African production and consumption linkages foreshadow estimates of lower growth multipliers. But bear in mind that available measures may bias the African rural linkages estimates downward because of a key feature of the African rural economy - the high share of nonmarketed goods and services in total consumption. Because they are not marketed, many of these goods and 111 services are not measured. The Gusau consumption study, for example, does not impute a value for consumption of home produced nonfood commodities. And none of the studies values nontraded household services - cooking, cleaning, laundry and child care. Because the Asian economies are more highly monetized, more of these services enter the measured consumption accountF, Thus, in the same way GNP comparisons exaggerate income gaps between rich and poor countries - because high female workforce participation leads to booming restaurant, laundry, house cleaning and childcare businesses - the Asia-Africa comparisons are likely to overstate the apparent linkage gap between the two.regions. Consider too the farm household investment data in Table 37. These data make a rare attempt at valuing not only physical inputs but also fixed investment produced by family labor. The data, from Nigeri.a, reveal that. over 80% of the value of investment comes in the-form of land clearing and improvement, road construction, tree planting and construction - all primarily produced by family labor. In contrast, agricultural equipment accounts for only 7% of total assets. Even outside of cash cropping areas, farming systems researchers point to promising investments in soil fertility and drainage, for example through land leveling and tied ridging (Ohm, Nagy and Sawadogo, 1985; Eicher and Baker, 1982). So in the future, service and labor linkages may well prove to be the largest of the backward linkages supported by African agriculture. Today, because they are not procured with cash, many of these investments simply escape measurement. But in the future, as the rural Africa becomes increasingly monetized, both commodity and labor flows will become more visible. And the currently observed gaps between Asia and African linkages may well diminish. 112 TABLE 37 FARM HOUSEHOLD GROSS CAPITAL FORMATION, ONDO STATE NIGERIA, 1981 Value per Percent Household of Total 1. Immobile Assetsa (Naira) Trees and tree crop planting 1,650 57 Land clearing 412 14 Roads and road construction 222 8 Utility structures 125 4 2. Equipment and supplies Bicycles and autobikes 201 7 Agricultural machinery 118 4 Agricultural tools and implements 76 3 Stored produce 58 2 Biological input supplies 31 1 Total Capital 2,893 100 a Valued at'imputed labor cost plus price of material inputs. SOURCE: Adesimi (1983). 113 C. Probable Magnitude of the Agriculture to RNFE Multipliers 1. Direct Estimation Data from Togo and Sierra Leone enable us to relate regional differences in rural nonfarm earnings to differences in agricultural income. These data, reproduced in Appendix Table A.2 and plotted in Figures 8 and 9, show a clear positive relationship between per capita agricultural income and per capita rural nonfarm earnings. That positive association squares well with our understanding of the changes taking place during structural transformation. It also reinforces the testimony from analysts who have observed higher levels of nonfarm activity attracted to prosperous agricultural regions, for example Berry (1975) in Nigeria and Freeman and Norcliffe (1985) in Kenya. By regressing rural nonfarm income on agricultural earnings, we can generate crude initial estimates of-the magnitude of agricultural growth linkages. At first blush, the estimated coefficient on agricultural income might be considered an upper bound projection of the agriculture- to-rural-nonfarm income multiplier; because some regional variation in nonfarm income may well be caused by forces independent of agriculture, for example government or resource based extraction and processing. For this reason, the regression estimates may bE high side estimates of agricultural growth multipliers. But other factors, such as spillovers and leakages across regions, could lead to the opposite result, so it is not possible to say with certainty that the following estimates represent maxima. 114 Equations (1) and (2) display the results of two ordinary least square regressions of per capita nonfarm income (RNFY) on per capita farm earnings (AGY). The figures in parentheses are standard errors. Note that the Sierra Leone regression includes all regions except Scarcies, the outlier region with a high proportion of nonagriculturally related fishing income. (1) Togo RNFY = 5.8 + .43AGY R2=.56 (.22) (2) Sierra Leone RNFY = 4.2 + .14AGY R2=.31 (.09) The Togo results indicate that a $1 increase in agricultural income, other things equal, leads to a $.43 increase in regional nonfarm earnings. This estimate is strikingly close to the modeling results that follow. But the Sierra Leone estimate, even after removing the outlier region, is far lower. It suggests that a $1 increase in farm income generates only an additional $.14 in nonfarm earnings. Closer inspection of the survey methodology reveals the reason for this low figure. The Sierra Leone data exclude income from households primarily engaged in trading. Because commerce generates consistently the highest share in rural nonfarm income, the Sierra Leone data likely seriously underestimate total rural nonfarm earnings. For present purposes, one advantage of this omission is that it reduces the coverage of the nonfarm multiplier to principally production linkages. Supporting the notions advanced previously, this interpretation suggests that rural production linkages from agriculture 115 are in fact very low, at least in the predominantly hand hoe agriculture of Sierra Leone. 2. Modeling Agricultural Multipliers In view of the limited opportunities for direct estimation of growth multipliers, much of what we can learn must come from modeling. The simple model used here is borrowed from Hazell (1984). It allows estimation of regional multipliers based on three basic parametersi two production and one consumption, values of which can be obtained from farm management, rural nonfarm enterprise and consumption surveys. The great advantage of modeling, of course, is that it allows us to explore the implications of alternative types of agricultural growth on the ensuing nonfarm linkages. Moreover, it also allows a breakdown of the total multiplier into its production and consumption based components. a. The model For most rural regions in developing countries, the major output is food or cash crops, the greater part of which is exported out of the region at given prices. In return, these regions import from outside the region manufactured goods for investment, production and household consumption. Locally produce.d nonfood goods and services are also produced in the towns and larger villages which cater almost exclusively to local demand. Local nontradable goods often include some specialty agricultural commodities, such as fresh fruits, vegetables, and livestock products. In fact, as we saw in our analysis of household expenditure behavior, the latter may be the dominant nontradables in rural Sub-Saharan Africa. 116 A reasonable characterization of the supply structure of many rural regions is as follows. The supply of their major output--foodgrains or export crops--is typically fixed by the technology and, in Asia by land, in Africa by labor. On the other hand, the output of nontradables is elastic. In Asia this is because the supply of labor, which is the major input for nontradables like services, is also elastic. In Sub-Saharan Africa the nontradables are mostly income elastic agricultural products which can be produced at the expense of major foodgrain or export crops. Other nontradables, particularly services, are often provided in the dry season when labor is relatively abundant. If we further assume that the prices of all tradables (exports and imports) are given to the region, then we can construct a simple regional model to measure the multiplier effects. Let X.denote regional total gross output, which is comprised of tradable output T and nontradable output N. Then (1) X = T + N We assume that the output of tradables is fixed (2) T = T However, the output of nontradables is elastic, and therefore output is determined by the regional demand for nontradables. This is comprised of household expenditure CH on nontradables for consumption and investment purposes, and intermediate demands for nontradables Cp in regional production. We shall assume household expenditure can be approximated by a linear function of income 117 (3) CH C0 + OY where C is a constant and a is the marginal budget share for nontradables out of income (value added) Y. We also assume that intermediate demands for nontradables are proportional to regional gross output X, so that (4) Cp = aX Here a is the ratio of nontradable intermediates gross output. Regional demand for nontradables, which is equal to nontradable output, is then (5) N = CH + CP Finally, total household income in the region, or regional value added is (6) Y = VX Where v is the ratio of value added to gross output. Substituting equations (2) through (5) into (1) yields X = T + C + (a + 8v)X, or (7) X = (C + T)/(1 -a- Ov) Equation (7) determines the region's gross output given the output of tradables and key behavioral parameters of the regional economy. Suppose now that the output of tradables is increased through technological change in agriculture. What will be the multiplier impact on the region's income or value added? Using (6) and (7), the derivative (8) dY = v/(1 a - 0v) 118 measures the change in regional value added for a unit change in regional gross output. The value added multiplier is defined as -the change in regional value added for a unit change in the value added from agriculture. To obtain this multiplier we need to divide (8) by the ratio of value added to gross output for agricultural tradables. Since, by assumption, this ratio is also v, the value added multiplier is (9) 1 . dY = 1/(1 - a- Ov) v T In this model the value added multiplier happens to equal the gross output multiplier dX/dT. b. The size-of the multiplier The size of the value added multiplier is determined entirely by three key parameters in this simple model of a regional economy. These are the marginal budget share for nontradables in household expenditures (s), the ratio of nontradable intermediates to gross output in total production (a), and the ratio of value added to gross output in total production (v). If these three parameters are known, then we can calculate the size of the multiplier benefits deriving from agricultural growth in a rural region. Since a, 8 and v are all ratios, they must take on values between 0 and 1. Furthermore, since 1-v is the ratio of total intermediates to gross output, then a cannot exceed 1-v. In Table 38 we show the values of the regional value added multiplier for some selected values of a, and v. 119 Other things being equal, the regional value added multiplier increases with the ratio of value added to gross output (v). Poorer regions tend to have higher values of v because they use less intermediates in production. However, this favorable component to the multiplier is likely to be offset in poor regions by a lower marginal budget share for nontradables- (); Table 38 shows that the multiplier declines with 8. How the value of a might differ between regions is less clear. On the one hand, a higher ratio of value added to gross output should lead to a lower value of a. On the other hand, this may be offset by a greater reliance on nontradables as intermediates in poorer regions. By itself, larger values of a act to increase the value added multiplier. The Muda region studies by Bell, Hazell and Slade (1982) is a moderately prosperous rice growing region in Malaysia. Ir, 1973 the per capita income was US$227. Using data from Bell, Hazell and Slade, we estimate the relevant parameters for Muda to be v = 0.5, = 0.4 and = 0.25. Table 38 gives a multiplier of 1.818 for this combination of parameters. This is very close to the value added multiplier of 1.83 reported by Bell, Hazell and Slade after a much more careful analysis. Hazell, Ramasamy and Rajagopalan (forthcoming) have derived an identical multiplier for their North Arcot study region in south India using a similar modeling approach. In 1982-83 per capita income in north Arcot was US$100. While similar studies have yet to be undertaken in Sub-Saharan Africa, it is possible to provide rough orders of magnitude for the parameter values. The Gusau region of Northern Nigeria studied by Hazell 120 TABLE 38 VALUE OF THE REGIONAL VALUE-ADDED MULTIPLIER UNDER DIFFERENT ASSUMPTIONS ABOUT KEY PARAMETERS FOR A REGIONAL ECONOMY Percent Multiplier Value-Added Multipl.ier Due to Consumption Consumption Consumption v. a a Endo2enous Exogenous Linkagesa 0.3 0.3 0.35 1.786 1.538 31.5 0.3 0.3 0.525 2.597 2.105 30.8 0.3 0.5 0.35 2.0 1.538 46.2 0.3 0.5 0.525 3.077 2.105 46.8 0.4 0.3 0.3 1.724 1.429 40.7 0.4 0.3 0.45 2.326 1.818 38.3 0.4 0.4 0.3 1.852 1.429 49.6 0.4 0.4 0.45 2.564 1.818 47.7 0.4 0.5 0.3 2.0 1.429 57.1 0.4 0.5 0.45 2.857 1.818 56.0 0.5 0.3 0.25 1.667 -1.333 50.2 0.5 0.3 0.375 2.105 1.6 45.7 0.5 0.4 0.25 1.818 1.333 59.3 0.5 0.4 0.375 2.352 1.6 55.6 0.5 0.5 0.25 2.0 1.333 66.7 0,5 0.5 0.375 2.667 1.6 64.0 0.6 0.3 0.2 1.613 1.25 59.2 0.6 0.3 0.3 1.923 1.429 53.5 0.6 0.5 0.2 2.0 1.25 75.0 0.6 0.5 0.3 2.5 1.425 71.4 0.68 0.3 0.10 1.440 1.111 74.7 0.68 0.3 0.15 1.550 1.176 68.0 0.85 0.3 0.05 1.439 1.052 89.0 0.85 0.3 0.10 1.550 1.111 81.1 aCalculated as (M -M)/(M -1) x 100% where M and M are the multipliers calculated with c6nsumption specifies endogenously at exogenously, respectively. 121 and Roell (1983) is much poorer than Muda and North Arcot. In 1976 the region's per capita income was about half of that in Muda in 1973. Hazell and Roell give an estimated value for a of about 0.3. A similar value of 0 can be obtained for rural Sierra Leone using the data provided by King and Byerlee. The value added to gross output ratio (v) tends to be high in African agriculture--about 0.9 to 0.95--because few modern inputs are used (See Appendix Table A.4). Careful studies of a- range of rural nonagricultural activities in Botswana (Haggblade 1982, Section B. and Haggblade 1984, p. 315) and-Sierra Leone (Liedholm and Chuta 1985) suggest an average v value of about 0.6 (See Appendix Table A.5). Assuming that about 70% of rural value added derives from agriculture, then the average value of v lies in the range 0.82 to 0.86--say 0.85. Since a must be less than or eq*ual to 1-v = 0.15,.we shall assume a values in the range 0.05 to 0.1. Given these parameter values for rural regions in Sub-Saharan Africa, the multiplier would be about 1.5 (Table 38). When compared to the multiplier of 1.83 reported for the two Asian case studies, the indirect gains in nontradable value added are only 60% as large in Sub-Saharan Africa. These indicative multiplier depend critically on the assumption of a highly elastic supply of nontradables. If this supply is less elastic, perhaps because of labor shortages in Sub-Saharan Africa, then the multiplier will necessarily be smaller. c. Estates verses smallholders and the size of the multiplier It might be expected that the multiplier will be affected by the 122 relative importance of commercial farmers (estates) and smallholders in producing total agricultural output. Available data on cash crops grown by estates suggests v values in the range of 0.65 to 0.75 (Appendix Table A.4). In contrast, smallholders have v values of around 0.8 to 0.9 for cash crops and 0.9 to 0.95 for food crops. Estates also use modern inputs more intensively than smallholders, so it seems unlikely that their a values are much larger than those for smallholders despite their lower values of v. If, in the absence of better information, we also assume similar values of o for estate families and smallholders (large farmers had only marginally higher values of s in Gusau), then regions dominated by estates will necessarily have smaller multipliers. Our model can be used to predict how much smaller the multiplier will be. Let us compare two regions, both of which derive 70% of their total value added from agriculture but one is dominated by estates and the other by smallholders. As before, assume a v value of 0.6 for nonagriculture and an average a of 0.3. In the estate dominated region the v value for agriculture will be about 0.7, giving an average v value for the regional economy of 0.68. Then, given values of a in the range of 0.1 to 0.15, the multiplier will lie between 1.44 and 1.55 (Table 38). In the smallholder dominated region the v value for agriculture will be closer to 0.9, hence the average v for the region will be about 0.85. The value of a is then likely to fall in the range of 0.05 to 0.1, giving a regional multiplier of between 1.45 to 1.55. This happens to be the same as the multiplier for the estate dominated region suggesting that the relative importance of 123 smallholders and estates within a regional economy has little bearing on the size of the multiplier. A more significant difference would emerge if the 8 values also differed. For example, if est-te families were found to have more urbanized expenditure patterns and their a value were only 0.2, then the estate dominated region (v=0.68, 8 = 0.1) would have a multiplier of only 1.31. - d. The relative importance of consumption and inter-industry linkages in the multiplier We can also use our model to isolate the importance of the household consumption linkages relative to the inter-industry linkages in the multiplier. To do this we need a variant of the model in which household consumption is held constant. The derived multiplier arising from agricultural growth will then be due entirely to -inter-industry linkages. Replace equation (3) with (3)1 CH = C0 where CO is a constant. Regional gross output X becomes (7)' X = (C + T)/(1 - a). Calculating the value added multiplier as before, this is now (9)' 1 *dY =1/(1 - a) v dT Thus, when household consumption expenditure is held constant, the value of the regional multiplier is determined solely by a; the ratio of nontradable intermediates to gross output in total production. 124 The values of the multiplier obtained from (9)' for selected values of a are also reported in Table 38, where they can be compared with the full multiplier corresponding to endogenous consumption behavior. The difference between each matching pair of multipliers is the component of the full multiplier attributable to the consumption linkages. For example, in the first row of Table 38, the full multiplier leads to an indirect gain in regional income of US$0.786 for each US$1.0 increase in value added in agriculture. But when household consumption is held constant, the indirect gain is only US$0.538. The difference, US$0.786- US$0.538 = US$0.248, is therefore the indirect gain attributable to the household consumption linkages. This is equal to 31.5% of the total indirect gain. The share of the multiplier attributable to the -consumption linkages increases with a and v, but declines as a increases. For the Muda region with parameter values v = 0.5, = 0.4 and a - 0.25, 59.3% of the multiplier is attributable to the consumption linkages (this compares almost exactly with Bell, Hazell and Slade's estimate--page 179). Hazell, Ramasamy and Rajagopalan report that half of the multiplier in North Arcot was due to household consumption linkages. In contrast, the consumption linkages become relatively much more important in regions having the kinds of parameter values we previously associated with Sub-Saharan Africa. For example, with v = 0.85, s = 0.3 and a = 0.1, the consumption linkages account for 81% of the multiplier. This is not because the consumption linkages are more powerful in Africa than Asia (we have assumed a smaller value of a for Africa). Rather, it 125 reflects the much weaker inter-industry linkages associated with high values of v and low values of a. this reiterates the conclusions drawn from the Togo and Sierra Leone direct multiplier comparisons above. 3. Macroeconomic Consequences and Simulations The foregoing analysis presumes that regional growth is an end in itself, and it gives no regard to spillover effects that might be induced elsewhere in the national economy. For example, imports into a rural region are leakages as far as regional growth is concerned. But if these goods are produced in other rural areas, or if they create jobs for the urban poor, they will still be desirable from the national viewpoint. Similarly, savings that are invested outside the region represent a loss to regional growth, but they are nevertheless valuable in furthering national economic growth. To measure the full indirect impact of agricultural growth in the national economy requires a more general equilibrium modeling approach. Byerlee has undertaken such work for Nigeria (Byerlee, 1973) and Sierra Leone (Byerlee, et.al., 1977). Using a macroeconomic simUlation model, he analyses the impact of government policies for expanding, alternatively, the production of export and food crops. For Nigeria, his results show that policies which expand food crop production increase value added in nonagriculture by Naira 0.23 for e3ch Naira 1.00 increase in value added in agriculture. In contrast, policies to increase export crop production lead to between 0.73 and 0.96 Naira of additional value added in nonagriculture for each Naira 1.00 increase in agricultural value added. 126 The results for Sierra Leone are more modest, probably reflecting the much weaker industrial base of that country. Increases in food crop production have a similar impact on the national economy as in Nigeria: Leone 0.29 of nonagricultural value added for each Leone 1.00 of additional value added in agriculture. But the linkages emanating from increased export crop production are weaker than in Nigeria - Leone 0.42 of value added in nonagricu1ture for each Leone 1.00 of additional value added in agriculture. These results are generally consistent with tha size of our regional multipliers. This would seem to suggest that any positive spillover effects arising beyond a region's boundaries are largely offset by negative spillover effects, arising perhaps from reduced labor and capital outflows from a growing region-. There are no directly comparable studies for Asia, but modeling studies of India (Rangarajan, 1982) and South Korea (Adelman, 1984) suggest much stronger linkage effects from agricultural growth on the national economy. 127 IV. CONSTRAINTS TO THE GROWTH OF RURAL NONFARM ENTERPRISES A. Demand The power of agricultural growth linkages depends heavily on whether or not dem-and currently constrains the output of rural nonfarm enterprises. If indeed demand constraints prevail,-second round responses to agriculturally induced growth occur at very low or no opportunity cost, and multipliers attain their full value. Demand for the goods and services of rural nonfarm enterprises (RNFE) depends heavily on the purchasing power of the surrounding farm population. Agriculture generates about 70% of rural income in many African countries (Table 3), thus providing a principal. source of rural spending pdwer. Because fiscal, tariff 'and pricing policies all critically affect farm incentives and farm income, the sum total of these policies affecting the terms of trade between agriculture and the rest of the economy government policy is a crucial determinant of effective demand for rural nonfarm products. A host of analysts have described the array of policies affecting African agriculture and in general conclude that_ agriculture has been heavily taxed relative to urban and manufacturing sectors. African governments, for example, typically tax export crops, at rates between 30 and 50% (World Bank, 1981). Sharpley's (1981) estimate for Kenya indicates that government has increasingly taxed agricultural income, reducing farmer purchasing power by substantial amounts. Thus, if demand does constrain RNFE, government agricultural policy must bear major responsibility. 128 - To argue that demand constrains rural nonfarm employment requires evidence of excess capacity in nonfarm activities. While capacity utilization is difficult to measure in rural areas.where seasonality and lack of electrical illumination prevent multiple shifts, two African studies have ventured estimates of excess capacity in nonfarm enterprises. Based on entrepreneurs' judgements about how -much output could be increased in the face of unlimited demand and material availability, Liedholm and Chuta (1976) estimate that rural nonfarm enterprises operate with 25 to 45 percent excess capacity. As Table 39 indicates, excess capacity of this magnitude appears in all major rural nonfarm manufacturing and repair enterprises, although surplus capacity seems greater in smaller rural localities. Steel's (1977) estimates for Ghana find excess capacity in the range of 34 to about 60 percent for three towns, two of which can be classified as rural towns. He, too, finds -sorme variation by activity, although he offers no disaggregation by size of locality. Because they refer to slack in only single shift operations, these figures are likely to be lower bound estimates of excess capacity. This considerable excess capacity is consistent with the notion that rural nonfarm activity is demand constrained. Of course, since excess capacity may also stem from supply side bottlenecks, it constitutes a necessary but not a sufficient condition for judging RNFE demand constrained. Since tailors and service enterprises, the most numerous of the RNFE, require customers to supply working materials, it seems likely that a substantial portion of the excess capacity is in fact caused by limited demand. Consequently, demand likely does constrain RNFE output, 129 TABLE 39 EXCESS CAPACITY IN AFRICAN RURAL NONFARM ENTERPRISES (percent) Blacksmith/ Baking/ Sierra Leone (1974) Tailoring Carpentry Metalwork Milling Aggregate (Locality size) Under 2,000 45 45 43 -- 2,000 - 20,000 34 33 47 41 -- 20,000 - 100,00 29 25 30 32 -- Over 100,000 24 22 -- 30 -- Overall 33 34 41 34 35 Ghana, 1974a 34 58 41 52 51 aThree urban areas, Accra (population 25,618) and Aburi (population 7,656). SOURCE: Sierra Leone: Liedholm & Chuta (1976). Ghana: Steel (1977). 130 and agriculturally driven linkage multipliers can, therefore, be expected fully inf the 1.5 range predicted by the modeling evidence above. B. Supply Constraints Entrepreneurs, when asked what principal problems constrain their activity, invariably emphasize the supply side.-- Capital shortages - expressed either as needs for cash, operating funds, liquidity, credit, or concern with the high price of material inputs - almost always emerges as their principal preoccupation. Table 40 documents this concern for a wide array of rural and urban small enterprises in Africa. Studies in other geographic regions echo the same refrain. But prominent analysts of rural and small-scale nonfarm enterprises doubt that working capital is as-big a constraint as entrepreneurs-contend (See Harper, 1977 & 1984; Kilby, 1962, 1982; Chuta and Liedholm 1979; Liedholm and Mead 1987). A variety of management inefficiencies may result in cash flow difficulties, making working capital constraints at least partially a symptom rather than a cause of operational business difficulties. So too may poor transport facilities and irregular input supply or output marketing networks surface as cash flow problems. While entrepreneur perceptions of cash flow difficulties very likely exaggerate their absolute importance, the rudimentary nature of transport, communication and institutional infrastructure does undoubtedly diminish the efficiency with which nascent rural financial.markets operate. Hence credit cannot be dismissed as a supply side concern. The practical issue then becomes whether working capital shortages should be addressed by 131 TABLE 40 PRINCIPAL BUSINESS PROBLEMS AS ASSESSED BY NONFARM ENTREPRENEURS (percent of entrepreneurs citing specific problems) Raw Transport Capital Customers Materials Infrastructure Prices Other 1. Rural Ente rises hana, ura Towns (1984) 61 3 5 -- -- 31 Kenya, Bungoma District 48 0 28 72 22 -- Central Province 77 35 20 -- -- -- Meru District 84 7 58 40 28 -- RIDC Clients (1972) 68 0 75 -- 84 Zambia, Six Provinces (1985) -- -- 53 -- -- -- 2. Urban. Srrall Entepises Accra, Ghana 42 18 36 -- -- -- Bamako, Mali 35a 36 a -- -- -- Djibouti 57a 11a - -- -- Kigali, Rwanda 67a 21 a -- -- -- Lome, Togo 22a 52 a . -- -- -- Nairobi, Kenya 29 39 .13 1 -- -- 'Nouakchott, Mauritania 45a 36 a -- Yaounde, Cameroon 49a 24 a -- -- -- a Includes lack of funds, cost of equiprent and raw materials. SOURCE: Ghana: Yankton (1985); Kenya, Meru and Bungoma Districts: Williams and McClintock (1981); Kenya, Central Province: Freeman and Norcliffe (1985); Kenya, RIDC: Child (1977); Zambia: Milim and Fisseha (1985); Accra: Steel (1977); Nairobi: House (1978), cited in Livingstone (1986); Bamako, Djibouti, Kigali, Lome, Nouakchott, Yaounde: ILO (1985x). 132 treating the symptom, through direct lending, or the ultimate cause of the cash flow difficulty by providing management assistance, infrastructure and institutional development, or some combination of all these strategies. The answer depends on the cost effectiveness of' the alternatives-, an issues that will be dealt with presently. Raw material shortages, possibly related to-working capital problems but also likely the result of poor road and transportation networks, also emerge as a common concern of rural nonfarm entrepreneurs. As one would expect, Table 40 indicates that this concern is far more prevalent in rural than in urban areas. Concerns with infrastructure and transport prices, both closely related to the raw material problems, also surface with some regularity. Entrepreneurs rarely identify management or technical deficiencies as constraining their operations. But, here too, the outside analysts and seasoned practitioners concur that there exists room for improvement on both scores in many small enterprises. Malcolm Harper, in particular, maintains that management consulting, on a very basic level, can generate significant increases in the profitability of rural nonfarm enterprises (Harper, 1977 & 1984; Harper and Soon, 1979). C. Existing Policy and Programs A welter of institutions and individuals have worked to influence the productivity and profitability rural nonfarm enterprises by addressing one or more of the major RNFE constraints. They have operated on two basic levels, working with both policies and projects. 133 1. Policies African governments maintain a wide array of policies that affect the operation of rural enterprises. These policies operate primarily through their differential impact on large and small firms; because most - although not all - rural firms are small. Investment codes, tariff, tax, and labor laws are of primary importance in shaping-the environment within which rurai firms compete with urban and foreign competitors. Virtually all African countries have adopted investment codes primarily to encourage investment in manufacturing. They provide an array of tax and tariff advantages as well, sometimes, as government guarantees of infrastructure provision to encourage the establishment of domestic manufacturing. A recent review of investment codes in Francophone West Africa indicates that common provisions include corporate income tax holidays ranging from 3 to 25 years, duty exemption on imports of equipment and raw materials for up to 15 years, as well as a variety of other country-specific incentives (Haggblade, 1984). All benefits are scaled up so larger capital investments receive higher incentivPs. In fact, most investment codes apply only to firms requiring above a certain threshold of investment. In Burkina, for example, only firms with fixed investment of over $63,000 qualify for major investment code advantages. While Botswana's code (their Financial Assistance Policy) remains a major exception, most other African investment codes effectively exclude the great majority of small rural and urban firms from consideration. The only explicit recognition of urban-rural differences embodied in these codes frequently lies in the provision of slightly higher levels of 134 preferences for businesses established outside of the capital or major cities. Sometimes these incentives are accompanied by provision of subsidized infrastructure in the form of serviced industrial land or industrial estates. But these efforts rarely succeed in enticing large firms out of the major cities (UNIDO, 1978). Investment codes and the related policy environment almost invariably favors the large, and hence the urban, enterprise; and they frequently do so by reducing the price of capital faced by these large firms. Drawn up with the large firms in mind, regulations exempt big manufacturers from import duty on all equipment, while small rural tailors, metal workers, transporters ahd service establishments face duties on their sewing machines, hand tools, welding equipment, and vehicles. Evidence from Nigeria, Sierra Leone, and Burkina suggests that for common small enterprise.activities, the duty rates faced frequently lie-.in the range of 30-50% raising their equipment costs by that amount and placing them at a competitive disadvantage compared to the large firms (Chuta and Sethuraman, 1984; Liedholm and Chuta, 1976; Haggblade, 1984). Foreign exchange guarantees and access to credit at artificially low rates of interest further accentuate the capital subsidies. Labor laws work in the other direction, raising up labor costs to large firms and providing small firms some measure of labor cost advantage. Because large firms ultimately do pay taxes after tax holidays have expired, tax laws, too, confer a cost advantage on small firms which more easily elude the fiscal net. 135 The net effect of policy induced factor price distortions has been measured for several African and Asian countries. The results, displayed in Table 41, indicate that the common impact of the array of promotional policies is to underprice capital an.d overprice labor for large firms, and vice versa for the small. Thus, they encourage more capital intensity in large firms and more labor using technology in small firms than would prevail in a neutral environment. The net effect of policies on the competitiveness of small vis-a-vis large firms is difficult to sort out. But in Sierra Leone, the one instance where data exist in sufficient detail to make an assessment, it is clear that the net impact of government policies has been to subsidize large scale enterprises and to penalize the small (Haggblade, Liedholm and Mead, 1986). Moreover, in this case, the large firms produce at negative social profitability, supported only by the extensive array of policy induced subsidies. 2. Programs a. Program profiles African governments, often aided by donors and'Private Voluntary Organizations (PVOs), have instituted a wide array of programs on behalf of rural, and even more commonly, on behalf of small enterprises. Common program components include credit, serviced industrial land and workshop buildings, assistance in project preparation, management and technical assistance. In some cases programs provided an integrated package of credit, project preparation, management and technical assistance. This integrated, package approach was particularly evident during the wave of 136 TABLE 41 THE DIFFERENTIAL IMPACT OF POLICY-INDUCED FACTOR PRICE DISTORTIONS ON LARGE AND SMALL NON-AGRICULTURAL ENTERPRISES: PERCENT DIFFERENCE IN LARGE FIRMS' 00ST RELATIVE TO SMALL FIRMS Percent Difference in Capital Cost Owing to:a Percent Percent Difference Difference in in Labor Trade Interest Total Wage/Capital Country Period Costs Regime Rate Taxes Capital Rental Rate Asia Rong Kong 1973 0 0 0 0 0 0 Pakistan 1961-64 0 -38 -44 +22 -60 +150 South Korea 1973 0 - 5 -35 +10 -30 + 43 Africa Ghana 1972 +25 -25 -42 +26 -41 +119 Sierra Leone 1976 +20 -25 -60 +20 b -65 +243 Tunisia 1972 +20 -30 -33 n.a. n.a. n.a. Latin Arerica Brazil 1968 +27 0 -33 n.a. n.a. n.a. aAll capital-related figures have been converted into the annual rental value of a unit of capital (or user costs) using a modification of capital recovery formula presented in Guisinger (1981, p. 329). b n.a. = data not available. SOURCE: Haggblade, Liedholm, and Mead (1986). 137 industrial estate programs launched in the 1960'and early 1970's. Imported from India to Kenya in 1966, the initial urban industrial estate program (KIE, the Kenya Industrial Estate Programme) was expanded to rural areas in the form of Rural Industrial Development Centers (RIDCs). From Kenya, often-with Scandinavian and UNIDO funding, the industrial estates spread to Tanzania, Swaziland, Lesotho, Botswana, B-urkina and beyond. In addition to the highly integrated package of promotional services available in the nursery industrial estates, many agencies have implemented programs tailored to address one or more of the principal nonfarm enterprise constraints. Credit programs have clearly been the most numerous; on the order of 90% of all donor-financed small enterprise projects include a finance component (Devres, 1981). Implemented by a variety of governments, development banks, PVOs and church groups, these programs most often provide credit for equipment rather than raw materials and frequently at concessional interest rates. Donors have strongly supported the small enterprise credit schemes. The World Bank, between 1977 and 1985, has directed approximately 30% of its industrial lending through intermediaries to small enterprises; although, as in many such programs, "small" loans have remained fairly large - in the range of $20,000 to $50,000 - and primarily urban (Levitsky, 1985). But PVOs and church groups operate a welter of very small scale and often rural loan funds. Governments and PVOs have also initiated management and technical training. A number of countries have established vocational training institutes expressly to improve technical skills of potential nonfarm 138 entrepreneurs. Rural artisan, especially rural blacksmith programs, have been a popular variant of this genre, springing up in Botswana, Burkina, Rwanda, Tanzania and undoubtedly elsewhere as well. Several PVOs, notably VITA and Care, provide technical extension in areas such as mud oven bakeries, charcoal production, beekeeping, and wood and metal working. Management training and extension has been most strongly advocated by Malcolm Harper (1979, 1984) whose methods of barefoot management extension have been adopted by the Partnership for Productivity (PfP) in Kenya, Botswana, Malawi and Burkina. Technoserve, another PVO with wide experience in management extension, often engages in long-term management contracts with individual firms. Another variant of management training is centered on assistance with pre-project feasibility studies. Commonly funded by UNIDO, these services are provided in the great majority of African Ministries of Industry or their promotional arms. Finally, many countries have begun decentralizing industrial infrastructure by creating serviced industrial and commercial zones in regional towns in an effort to encourage decentral-ization of business activity. These projects, almost always donor funded, typically involve zoning, surveying, and installation of basic road, electrical, communication and water infrastructure to a designated area. Given the large array of institutions involved - governments, PVOs, trade associations, church groups and donors - it is not possible to maintain an up-to-date inventory of all promotional activities currently underway on behalf of African rural nonfarm enterprises. Nonetheless, 139 Table 42 attempts to identify the government promotion.al programs operating in a cross section of African countries. Some programs - as in Nigeria, Kenya, and Botswana - are explicitly rural, while others focus on small firms and therefore may partially spill their operations over into rural as well as urban areas. Several common propensities are evident in the current array of assistance programs. First, most effort and money goes to the provision of credit. And in a majority of cases, agencies limit lending to investment rather than working capital. In part this penchant for fixed capital lending stems from the bankers' desire for seizable collateral; in part it reflects a desire to directly expand business activity. Given the apparent excess capacity in many rural nonfarm enterprises and given the expressly stated need for working capital, this bias is unfortunate. Second, most assistance programs focus on manufacturing. 'Although there are some notable exceptions, many programs preclude services and commercial enterprises from participating as project beneficiaries. This proclivity stems largely from the notion that sectors other than manufacturing are ."unproductive". Commerce, in particular, suffers from the image many people have of small children and women purchasing single packs of cigarettes, matches or bunches of bananas and reselling them one by one. Perhaps it is the traders' misfortune that*these representatives of their craft are the ones most visible from the windows of consultants' hotels. But even in these cases, the mini traders perform useful services for which people are willing to pay. They buy in quantities larger than many consumers can afford and sell individual items. They save consumers 140 TABLE 42 SUMMARY OF RURAL AND SMALL NONFARM ENTERPRISE PROMOTION PROGRAMS IN SUBSAHARAN AFRICA Small Ent. Promotion Nursery SSE Extension Feasibility Technical Agencies Estates Credit Services Studies Training Botswana BEDU-U/R Yes Natl. TA,MA-U/R Yes Yes RIOs-R Dev.Bk. BAS-U/R Burkina OPEV-U Yes Natl. No No Yes Ch.Comm. Dev.Bk. Artisanat-U/R Burundi No No No No Yes -- Cameroon CNAPME No Yes -- -- -- Com.Bk. Ethopia HASIDA-U No Ag.Bank MA,TA-U/R -- -- (for coops) Ghana GEDC-U No GEDC No -- -- Kenya KIE-U Yes KIE-U TA,MA-U Yes -- RIDC-R Yes RIDC-R TA,MA-R Lesotho LEDU Yes Yes -- -- -- Liberia -- -- LDC-U -- -- -- LBIDI NIC Madagascar Nigeria IDCs-U/R Yes-U SIC-R/U MA,TA -- VIC VICs-U/R (IDCs) Rwanda -- No Dev.Bk.-U/R MKA-R Yes Yes Bk.Pop.-R Senegal -- -- -- -- Yes -- Swaziland SEDCO-U* Yes SEDCO? MA,TA-U -- -- Tanzania SEDO-U/R Yes? -- -- -- -- 141 Table 42 (CONT.) Abbreviations: General: U=urban, R=rural, TA=technical assistance, MA=management assistance, Com.Bk.=commercial banks. Botswana: BEDU=Botswana Enterprises Development Unit, RIOs=Rural Industrial Officers, BAS=Business Advisory Service. Burkina: OPEV=Organization pour la Promotion de 1'Entreprise Voltaique, Ch.Comm.=Chambre de Commerce, de l'Industries et de l'Artisanat. Cameroon: CNAPME=Centre National d'Assistance aux Petites et Moyennes Entreprises. Ethiopia: HASIDA= Ghana: GEDC=Ghana Enterprises Development Commission. Kenya: KIE=Kenya Industrial Estates, RIDC=Rural Industry Development Centres. Lesotho: LEDU=Lesotho Enterprise Development Unit. Liberia: LDC=Liberian Development Commission?, LBIDI=Liberian Bank for Industrial Development and Investment, NIC=National Investment Commission. Madagascar: Nigeria: IDC=Industrial Development Centers, VIC=Vocational Improvement Centers. Rwanda: Dev.Bk.=Banque Rwandaise de Developpement, Bk.Pop.=Banques Populaires. Senegal: SONAPI=Societe Nationale pour la Promotion de l'Investissement? Swaziland: SEDCO=Swaziland Enterprise Development Corp. Tanzania: SEDO=Small Enterprise Development Organization. SOURCE: ILO (1983b-h,j; 1984, 1985b,c); Steel (1979); Chuta and Sethuraman (1984), Kilby (1984), Kilby (1982, 1987), Hunter (1979), Livingstone (1977), and personal visits. 142 time and transport costs in congested cities. They save society resources and improve consumer welfare. Trade and services produce value added just as surely as manufacturing does. And the argument that manufacturing generates backward linkages not present in other sectors ignores the crucially important forward linkages and market opportuniti.es that can be opened up by trading and distribution. Given that all available empirical evidence - the cross section, time series and consumption data - project that the strongest rural nonfarm growth will come in services and in commerce, the common manufacturing-only stance seems destined to divert resources from activities with the highest growth potential. b. Program evaluations Over the past 5 to 10 years, a small but growing body of careful ex- post project evaluations has emerged shedding important light on the effectiveness of alternative forms of small and rural enterprise promotion. Many are Africa specific, for example, Goldmark, Rosengard and Mooney (1978); Lassen, Traore, Brown and Walton (1985), Haggblade (1982), Kilby (1982 and 1987), Hunter (1980), and Livingstone (1977). But many more come- from other geographic regions. Particularly important are the recent comparative studies by Blayney and Otero (1985), Farbman (1981), Ashe (1985), Kilby (1979), Kilby and Bangasser (1978), Kilby and d'Zmura (1985), Levitsky (1985), and Liedholm (1985). While the evaluation experience is not exclusively drawn from rural Africa, a number of lessons emerge with sufficient consistency that they lend confidence to those wishing to direct African efforts based on the collective wisdom of world- wide findings. 143 Liedholm and Mead (1987) provide the most succinct recent distillation of the findings of the small enterprise project evaluation literature. A first, and perhaps surprising finding, is that a significant share of the small and rural enterprise promotion schemes evaluated have generated not only positive but high B/C ratios and internal rates of return. Of course the sample of evaluations available in the literature may be biased in an optimistic direction given that agencies implementing costly failures rarely hire outside evaluators and even less frequently propagate unflattering reviews. Certainly the series of evaluations from Botswana shows a mixed record. Including reviews of four technical assistance projects and three infrastructure provision efforts, two had B/C ratios of 0, three were less than one, and only two exceeded one (Haggblade, 1982). Kilby and Bassanger's (1978) review of ten ILO-technical assistance projects found six of ten has B/C ratios below one. But aside from these two sources, the bulk of the reviews conclude that small and rural enterprise promotion has frequently been cost effective. Kilby and d'Zmura's (1985) review of seven PVO small enterprise credit schemes computed B/C ratios between 1.02 and 8, leading them to conclude that "These rates of return place microenterprise lending schemes among the most successful categories of all types of foreign aid programs." Likewise, Levitsky (1985) suggests that World Bank small enterprise lending has fared as well standard lending programs. Blayney and Otero (1985) conclude, on a similarly optimistic note, that "small and micro-enterprise projects can accomplish ... (income and employment creation) objectives effectively." 144 This evidence suggests several hypo,theses about the relative effectiveness of different categories of intervention. First, credit projects seem to generate the highest B/C ratios, while technical assistance projects have a very mixed record. And the infrastructure projects evaluated in Botswana all fared poorly, with B/C ratios below .4. Finally, we are not aware of any careful B/C evaluation of management assistance efforts. While it may indeed be possible to raise rural nonfarm income by improving management skills, we have no evidence on whzther it can be done cost effectively. Liedholm and Mead (1987), in addition, draw several important operational lessons from the array of currently available evidence. For credit projects, they suggest that working capital lending appears more successful than those projects lending for fixed capital investment, probably because they respond to the- more important credit need of the small firms. In addition, working capital loans flow necessarily to existing businesses; and the evidence points very clearly to higher success rates among projects working with existing firms as opposed to those attempting to suscitate new enterprises. Because existing firms have proven markets, access to raw materials, labor and production facilities already, working capital credit provides them with a "single missing link" that enablfs them to increase production and profit. Liedholm and Mead echo the conclusions of Kilby (1979) and Kilby and d'Zmura (1985), that those projects that focus on single missing ingredients have a far better chance of success than those that try to provide everything - credit, marketing, management, and technical 145 assistance - at once. The dismal record of the industrial estate programs underlines the simplicity and power of this notion, that interventions will be most viable if they can identify opportunities in which provision of a single missing component will enable a multitude of nonfarm enterprises to grow. Boomgard et al. (1986) have proposed a diagnostic research technique, the subsector approach, that. aims to identify key missing links for important or potentially important nonfarm employers. Box 3, below, illustrates how a subsector study can be used to identify high payoff interventions on behalf of small rural enterprises. The evaluation literature also leads to some clear suggestions for specific types of intervention, particularly in credit projects where we have the greatest experiences, but also for other forms of direct small ehterprise promotion. .The evaluations of infrastructure provision in Botswana indicate that the projects failed the test of cost effectiveness because of overspecification of design standards. The project provided facilities of a standard far in excess of rural entrepreneurs' needs, ability and willingness to pay. This lesson is worth remembering as efforts turn to specifying necessary productive infrastructure in rural towns. Given the mixed record of rural electrification programs alone in stimulating rural business activity (Fluitman, 1981), questions of sequencing and complimentarity will need to be explored. In these and other areas of RNFE promotion, those contemplating spncific interventions would do well to examine in detail the references cited above, as they delve into more.detail than is appropriate here. 146 Box 3. Botswana's Sorghum Beer Industry: An Application of Subsector Research Batswana consume about 20% of all cereal products in the form of sorghum beer, a thick pinkish brown beverage with a distinctively sour yogtirt-like taste. A mildly alcoholic drink, 2-3% by weight, sorghum beer is hailed for its b-vitamins and high caloric content. Policy makers consider the industry important for several reasons. It furnishes 80% of rural manufacturing income in a country where rural employment creation is a key government priority. About 14% of Botswana's labor force earns some income from home brewing; and because 85% of the population lives in rural areas, that employment is primarily rural. Moreover, brewing income accrues exclusively to women in a country where 30% of all households are female headed and known as a group to fall disproportionately at the low end of the income distribution (Kossoudji and Mueller, 1983). Women have historically supplied -this huge sorghum beer market, brewing in their back yards with simple equipment costing about $110. But, beginning in the mid 1960's, factory brewers from Zimbabwe began contesting these markets. Setting up breweries in Bots4ana's two major towns, they initially confined their sales efforts primarily to urban centers. But in the late 1970's, aggressive new management began to expand rapidly into rural areas. In view of the important i'ncome and equity concerns, the government commissioned a study of the probable impact of the expansion of factory brewed sorghum beer. .Unlike much previous research on Africa's rural nonfarm economy, which has concentrated primarily on the small enterprise segment of the economy, the Botswana study followed a subsector approach which examined dynamics of the whole production/distribution system in which the home brewers operated. Figure 11 describes schematically how home brewers fit into the network of input suppliers, urban factories, and distributors. The schematic shows the four principal production/distribution channels through which raw materials are transformed and distributed to consumers in the form of sorghum beer. Two channels center on home brewing and two on factory brewing. Table 43 summarizes the principal analysis of the subsector review. On efficiency gnounds, home brewing with (factory produced) trade malt (Channel 2) and factory brewing retailed through home retailers (Channel 3), emerge the lowest cost producers, valuing inputs at their opportunity cost. Home brewing with home-made malt is revealed to be inefficient, primarily because low quality home malted sorghum requires about 40% more grain adjunct to produce a given quantity of beer. On equity grounds, home brewing with trade malt clearly outdistances both factory brewing channels, providing twice as much employment per volume of output and generating 50% more income for 147 the poor and rural dwellers. The effect of increases in factory brewing has been to transfer income from rural to urban areas, from women to men, and from poor to rich. Two important policy conclusions emerged from this review. First, Channel 2, home brewing with trade malt, emerged as the optimal technology on both efficiency and equity grounds. Revelation of the gains to be had by encouraging the *use of trade malt among home brewers, induced extension staff to work with trade maltsters to promote its use among Channel 1 home brewers. Switching to trade malt would enable them to produce more consistent brews, at lower cost, and remain competitive, with consequently beneficial employment, efficiency and equity impacts. Second, the study found that some District Councils had been misinterpreting retailing legislation which - in contrast with many neighboring countries who have outlawed home retailing-of factory-brewed sorghum beer - made specific provision for home retailing in order that displaced home brewers could recoup lost brewing income by retailing factory brew. Discussions with errant Council staff rectified this problem, increae ng home brewer income at very low cost, in an intervention that could only have been identified by activity specific, diagnostic work such as that embodied in the subsector approach. 148 - FIGURE 11 SCHEATIC OF THE FOUR ALTERNATIVE CHANNELS FOR PRODUCING AND DISTRIBUTING BULK SORGHUM BEER Channel 1 Channel 2 Channel 3 Channel 4 HBHM HBTM FBHR FBRB Home brewing Home brewing Factory brewing Factory brewing with home- with trade with home with bar made malt malt retailing retailing Home brewer Home retailer Sorghum been bar I. Retailing Fully using N = 2,400 1N =430 integratedN .e t trade malt Vol= 9 ML Vol.= 16 ML home maltster- -Brewer N = 16,000 -retailer V Factory II. Brewing brewer . vN = 2. Vol.= 28 ML N = 40,000 Vol.= 52 Trade Municipal maltster maltster III. Malt Production = 0 (imported) N = number of enterprises Vol. = volume in million liters enterprise boundary - - division between functions within a firm firms using identical technology at a given horizontal level in a production/distribution system indicates flow of goods between firms Source: Haggblade (1984). 149 TABLE 43 CRITERIA FOR COMPARING BOTSWANA'S ALTERNATIVE INDIGENOUS BEERPRODUCTION/DISTRIBUTION CHANNELS' (1982 U.S. dollars) Channel 1 Channel 2 Channel 3 Channel 4 HBHM HBTM FBHR FBRB Home Brewing Home Brewing Factory Brewing Factory Brewing Home-Made Malt Trade Malt Home Retaining Bar Retailing 1. Efficiency a. economic cost per unit of value-added 1.20 1.14 1.14 1.36 b. cereal requirements (tons) 25,000 15,700 15,200 15,200 2. Income Distributionb (income percentiles) a. lowest 65 percent $ 5,570 $ 6,010 $ 4,240 $ 3,540 b. 66 - 95 percent 5,040 3,960 4,150 3,510 c. richest 5 percent 940 1,420 3,810 51100 $0 1390 T2 =1 3. Geographic Distribution ofb Income a. rural $ 9,820 $10,480 $ 6,610 $ 3,290 b. urban 1,730 910 5,590 8,860 TIT~31 =1T T E TI22 4. Sexual Distribution ofb Income a. female $ 9,240 $ 7,340 $ 4,790 $ 6,560 b. male 2,310 4,050 7.,410 5,590 $11,550 $11,390 $12,200 $12,150 5. Emloyment. a. full-time labor equivalentsc 5,750 5,485 2,355 1,453 b. total number of people .22,990 23,724 23,344 2,127 aProjections based on output of 100 million liters of bulk indigenous beer in each production/distribution channel. bThousands of dollars. CA full-time labor equivalent equals 50 5-day weeks at 8 hours per day or 2,000 person-hours of work. SOURCE: Haggblade (1987). 150 V. POLICY AND RESEARCH IMPLICATIONS Several themes have recurred throughout this review. Together, they point in clear directions for policy makers and practitioners interested in balanced rural growth and in maximizing the rural nonfarm spinoffs of agricultural advance. 1. Agriculture will be the engine of rural growth. Agricultural growth will be essential in launching successive rounds of rural growth. Consumption linkages appear particularly important, and agriculture will necessarily play a key role given that farmers constitute the largest rural consumer group by a considerable margin. Forward production linkages provide additional thrust, while backward linkages seem weakest among the inter-sectoral. stimuli fostered by an agriculturally centered growth strategy. Generating full advantage from agricultural growth multipliers will require increases in farmer productivity. Raising farm labor productivity not only permits the release of labor from agriculture to non-agricultural pursuits, it is also essential in boosting per capita income to levels that will enable consumer diversification from food into non-food items. 2. Policies matter in maximizing growth linkages. Government policies affect not only the magnitude of agricultural growth but also the ability of rural nonfarm enterprises to respond to the production- and consumption-linked opportunities that ensue. Historically, African policy environments have hindered both locomotive and box-car. Macro policies have taxed agriculture for generations, 151 diminishing farmer incentives, farm output, farm income and hence both consumption and production multipliers. Investment codes and related legislation have regularly, if unintentionally, discriminated against small, rural nonfarm firms. The codes have treated large firms more generously than the small, conferring substantial competitive advantages on the, large through tariff relief, subsidized credit, subsidized infrastructure, and access to foreign exchange at overvalued official rates. Of course, some policies have favored the small and the rural. Evidence from Kenya suggests that the recent imposition of a sales tax has led to a substantial jump in income and production by rural and urban informal sector firms at the expense of the large enterprises who cannot elude the fiscal net. So policies can work in both directions. But in the few cases where it has been possible to net out the impact of these countervailing preferences, the large firms emerge as distinctly favored by the sum total of the policy environment. Beyond the general policy environment, industry-specific measures may have important implications for key rural nonfarm activities. The legal' restrictions on home brewing in Zambian townships has clearly diminished employment in that country's largest nonfarm activity. Similarly, tariffs on hand tools and sewing machines have penalized rural producers in Nigeria and Sierra Leone. -These inhibiting measures, because they are frequently confined to individual enterprise categories, can many times only be unveiled through industry specific reviews of important nonfarm activities. The subsector approach provides a promising vehicle for 152 activity specific diagnostic policy research of this nature (Boomgard et.al .1986). 3. Among RNFE, services and commercial enterprises will grow fastest. The historical focus on manufacturing, by both policy makers and researchers, seems worthy of redirection. Spatial, time series and consumption data uniformly point, not to manufacturing, but rather to commerce and services as key growth sectors over the course of the rural structural transformation. Some manufacturing activities do grow as well, particularly food processing and preparation, tailoring, carpentry and metal work. But the exclusion of services and commerce from many promotional schemes on the grounds that they are "unproductive" is unfor'tunate. Services, in particular, are the activities best insulated from urban competition; hence consumption leakages are sma-llest in this sector. Service and commercial enterprises require credit, management and technical skills as well as access to key infrastructure if they are to attract a clientele commensurate with their potential. Certainly they should not be excluded from assistance programs. And although they present special problems for the researcher - because of mobility, size and location of operation - investigations of rural commerce and services would provide a useful supplement to the array of data now available of manufacturing and repair enterprises. 4. RNFE growth will be centered in rural towns. Rural towns emerge as focal points in the development of the rural nonfarm economy. Cross-section data systematically identify dramatically higher densities of nonfarm activity in rural towns than in dispersed rural settlements. By providing nonfarm enterprises with catchment areas 153 sufficient to support minimum profitable scales of operation, rural towns offer firms the potential to exploit economies of scale and scope. Prospects for sharing equipment as well as the emergence of repair and support facilities induce. enterprise establishment in rural towns. Moreover, the emergence of rural towns enables policy makers to provide necessary productive infrastructure at lower cost than would be possible under widely dispersed settlement patterns. 5. Infrastructure will be necessary for RNFE and rural' town growth. For rural towns to play their role in a balanced rural development process, it will be essential to assure adequate economic and social infrastructure to support nascent modern nonfarm activities. Physical infrastructure such as electricity, t'elegraph and telephone communications, water and roads will undoubtedly be important. But our -understanding of sequencing requirements and complimentarity among infrastructural inputs remains imperfect. Anecdotal evidence on the rise of secondary cities in the Third World indicates that ground transportation - roads or railroads - are necessary first infrastructural investments (Rondinelli and Ruddle, 1978; Rondinelli, 1983). But beyond that, evidence remains elusive. The spotty record of rural electrification programs alone in fostering industrialization (Fluitman, 1981) suggests a need to review evidence on complimentarities, sequencing and necessary additional conditions fcr generating productive impact from infrastructural investments. Further experience with both overbuilding and subsequent maintenance problems points to a need for collateral examination of levels of service demand, ability to pay and feasible 154 administrative mechanisms for financing rural infrastructure. Closely related is institutional infrastructure which will be equally essential in fostering the transition to a more specialized, productive rural economy. Efficient rural financial markets will be particularly important. It appears that improvements on this front will require integration of the existing informal credit markets with the formal banking system. Such integration can improve the efficiency of existing informal markets in several ways: a) by enabling transfers of investment funds within and across regions in response to differential opportunities; b) by allowing the infusion of funds across regions to avoid the "covariance problem" of regional coincidence of need to recall loans and inability to pay in individual regional economies driven by good or bad harvests; and c) by substituting legal for social enforcement of contracts, a necessity in graduating up to large scale fInancing where the stakes and risk of default become too high for social enforcement to be effective. Work with credit unions and rural savings banks - often based on centuries old rotating and socially regulated capital transfer mechanisms - appear particularly promising for integrating the rural capital markets with the formal banking system (See Miracle, Miracle and Cohen, 1980; Binswanger and Rosenzweig, 1986; Haggblade, 1978; von Pischke and Rouse, 1983; von Pischke-, Adams and Donald, 1983). Labor markets are also essential, and their increased efficiency will follow in the wake of improved communication and transport infrastructure. Economic and institutional infrastructure necessary to support rural factor markets will be essential in directing productive resources into 155 areas of highest return. And of course, improved communications and transport facilities can only increase competitiveness and efficiency of product markets as well. 6. Human capital investments will also be required. Because much current writing emphasizes the need for investments in infrastructure, it is easy to overlook the collateral need for investments in people. Since services will be among the most rapidly growing rural nonfarm activities, investment in human capital will likely be essential for realizing those potential gains. Services depend more on skilled people than on equipment and infrastructure. 7. Direct assistance programs can be cost effective in promoting RNFE. Many programs of direct assistance to RNFE, although by no means all, have generated B/C ratios .exceeding 1. Least effective in general are programs which aim to provide a complete package of financial, technical and management assistance to individual firms. The poor performance of nursery industrial estates furnish eloquent' testimony to this effect. Instead, programs that provide single missing ingredients seem to have fared far better, because they have identified situations in which all necessary components - market, enterpreneurship, management, technical skills, raw material procurement - except one are in place. Project provision of the one missing element enables businesspersons to exploit previously unattainable economic opportunities. Credit programs have been most popular. Among them, projects providing working, rather than fixed, capital credit have typically generated the highest rates of return. Lending programs maintain an 156 additional attraction in that they operate in an arena where past experience provides the most effective guidelines for operational success. The performance of technical and management assistance projects is not well documented. But evidence to date suggests a mixed record. Cost effective assistance will require careful following of the guidelines of the emerging small enterprise promotion evaluation -literature. 8 . Women will play an important role in the eicerging rural nonfarm economy. Many of the nonfarm activities that will grow most rapidly during structural transformation - activities such as food processing and preparation, tailoring, trading and many services - are female dominated. So too are many of the declining rural nonfarm occupations - basket making, mat making, ceramics and weaving. Consequently,.women will be key actors in the economic transition of Africa's rural economy. To facilitate their contribution to an accelerated rural transformation will require assistance agencies and governments explicitly recognizing the key role to be played by women. Among other actions, many assistance agencies will need to recruit more females as extension, banking and technical assistance staff. In addition, some key regulations will merit review, particularly those in many countries which limit financial transactions by married women. Numerous countries, for example, prohibit married women from*holding their own bank accounts. Few topics strike closer to the viscera of African audiences than that of women's rights; and since the social ramifications of removing existing discriminatory legislation will have to be borne by the African societies, only they can make the necessary decisions. As outsiders, however, we 157 would be remiss in not alerting them to the very real efficiency costs that will likely be paid if women are not permitted to participate with full effectiveness in rural economic growth. 9. Research requires identifying priorities. This review, in sorting through the delicate and. spotty available evidence, has identified several key priorities for future research. In improving our confidence in estimates of growth multipliers, the key parameters to be measured are the marginal budget shares and expenditure elasticities for rurally produced nontradables. Those who analyze, and especially those who collect, rural consumption data would perform a vital service by tagging and highlighting the locational features of rural coisumption decisions. On the supply side of the rural nonfarm economy, it will be particularly important to start filling in descriptive and analytical profiles of the service and commercial sectors of the rural economy as a complement to the important, detailed work already undertaken on rural manufacturing and repair. Research in the future should also include both large and small, formal and informal firms within its purview. Past partial efforts leave us with nagging doubts about the sum total of the rural economy. Subsector research is a promising means of doing this. By looking at key activities one at a time and reviewing the entire network of resources flows, from raw material procurement through processing and distribution, the subsector approach integrates analysis of the relevant manufacturing, commercial and service segments of the economy. And by including large 158 and small firms together, it provides useful indications of their competitive or complementary relationships in alternative channels thereby providing insights into future dynamics in each commodity subsystem. Finally, we need more detailed review of the links between rural infrastructure and stimulation rural nonfarm activity. Questions of sequencing, complimentarities, substitutability, necessary collateral inputs, and effective methods of financing and maintenance remain obscured. 10. Magnitude of the multipliers. Based on the limited evidence available to date, we estimate Africa's rural agricultural growth multipliers to be on the order of 1.5. That is, a $1 increase in agricultural incomes will generate about an additional 50 cents in rural nonfarm goods and services. This initial estimate places the African multipliers at about 60% of what they appear to be in the few Asian countries for which we have estimates. Different types of agricultural growth - smallholder vs. largeholder, cash vs. food crops, tree crops vs. annuals,* mechanized vs. animal traction or hand hoe agriculture - may generate different multipliers. But initial estimates, based on simple modeling calculations, suggest surprisingly little variation. Contrasting small farmers and estates, and tree crops as opposed to annuals, we have found that agricultural systems with high rates of value added in total production are those with low backward linkage with the rural economy. Hence the multipliers vary little within the range tested. We must caution that farm management and especially consumption profiles of the very wealthiest farms remain 159 elusive. As data from these very high income farmers become available, the range of agricultural growth multipliers may well widen. But it remains to be seen whether presumably lower rural consumption linkages will be offset by potentially greater production links. Our preliminary hypothesis is that African multipliers are lower than those found in Asia because of a combination of -different policies and different natural environments. The nature of African rainfall patterns and geology of river basins preclude cost effective irrigation a scale as large as in Asia. Hence backward linkages into pump supply, canal construction and maintenance, all currently important in Asian countries, will simply be unavailable in Africa. Population density is also much lower in Africa, requiring larger geographic catchment areas than in Asia to support minimum viable scales business activity. This diminishes the -competitiveness of rural nonfarm producers contesting markets with.large urban suppliers. In addition, African consumption patterns seem less diversified into nonfoods than in Asia. But at this stage it is not possible to say whether the difference is due to different income levels, differing preferences for urban and imported goods, measurement error stemming from the large share of nonmonetized goods and service transfers in Africa, or simply the result of an array of existing fiscal, trade and pricing policies which, 'coupled with lower levels of productive infrastructure, induce lower second round supply responses in rural Africa. We have much still to learn. 11. Government will play a key role. But one lesson does emerge clearly. African governments have a major 160 role to play in facilitating the rural economic transformation and in maximizing rural growth linkages. To do so will require a mix of policy and investment decisions focused not only on agriculture but also on rural towns, since they are the laboratories in which rural nonfarm enterprises will incubate and grow. 161 APPENDIX TABLE A.1 Rural NONFARM EMPLOYMENT CORRELATIONS, BASIC DATA Population Area 2.Relevant 000's Rural GDP/ Ag.Share Ag.Inc/ 1.Natl. Rural Sq. Km. Density Capita in GDP Capita Benin, 1961 2,106,000 1,896,600 112.6 16.8 302 0.55 166 Cameroon, 1976 7,131,833 5,126,610 475.0 10.8 615 0.32 197 Chad, 1964 3,254,000 2,351,160 472.2 5.0 222 0.47 104 Ghana, 1960 6,726,815 5,477,296 239.0 22.9 1,436 0.41 589 Ivory Coast, Bouake Region, 1963 630,000 23.3 27.0 547 0.38 208 Malawi, 1977 5,547,460 5,076,802 94.3 53.8 197 0.44 87 Mali, 1976 6,394,918 5,318,089 1,240.0 4.3 100 0.38 38 Mauritania, 1977 1,500,000 713,769 1,031.0 0.7 457 0.30 137 Mozambique, 1980 12,100,000 10,134,606 802.0 12.6 195 0.44 86 Rwanda, 1978 4,653,224 4,443,125 26.0 170.9 237 0.46 109 Senegal, 1970/71 4,038,000 2,262,006 196.0 11.5 575 0.25 144 Sierra Leone, 1974 2,970,000 1,605,570 72.0 22.3 363 0.34 123 Tanzania, 1978 17,440,000 15,099,709 945.0 16.0 297 0.51 151 Togo, 1970 1,950,646 1,969,493 56.0 35.2 399 0.35 140 Zimbabwe, 1982 7,501,470 5,573,592 391.0 14.3 574 0.15 86 SOURCES: IMF International Financial Statistics Yearbook, 1986; World Bank World Development Reports, various years; Population censuses as per references in Table 16. 163 TABLE A.2 REGIONAL VARIATION IN FARM AND NONFARM INCOME, TOGO AND SIERRA LEONE Agricultural Nonfarm Income Per Income Per Country/Regions Capita Capita T090 (thousands of CFA francs) Centrale 36.7 18.9 Plateaux 33.0 24.5 Centre-Nord 27.1 18.1 Savanes 17.7 8.4 Kara 12.0 14.1 Sierra Leone (Leone) Ri verainGrasslands 95 17 East 93 14 South Plains 83 21 South Coast 82 17 Bolilands. 79 11 North Plain 69 15 North Plateau 50 10 Scarcies 42 30 SOURCES: ILO (1982); Matlon et.al. (1979). 1.64 TABLE A.3 REVIEW OF RURAL AFRICAN CONSUMPTION STUDIES Study Location Income Elasticity MBS ABS Adamu (1973) Nigeria 9 cons. categories --- compare by occupation 8 urban by occupation farmers vs. others Dah1 (1971) Kenya, RU 41 categories 41 categories Etukudo Nigeria, RU --- 14 categories (1978) Hazell and Gusau, 8 categories, Yes Yes Roell (1983) Northern includes Nigeria locational designations Humphrey and Malawi, plus 2 foods; 25 --- --- Oxley (1976) urban nonfoods and 3 ag. inv. compares 7 African studies King and Sierra 11 food categories, Yes Yes Byerlee Leone, R 16 nonfoods, with (1977, 1878) locational and factory intensity designations Massel (1969) Kenya 1963 9 foods, 9 foods, 6 nonfoods 6 nonfoods Malawi Malawi 8 categories (1984) Simmons Zaria, --- 4 categories, excludes (1976) N. Nigeria durables Sousa Togo, RU 8 categories by Y level (1970) 1 Chad, RU 8 categories by Y level 13 RU African 24 categories by Y level lo, ations Tanzania Tanzania-- 41 categories by (1972) RU + regions expenditure level also available for about 300 categories NOTES: R = rural; U = urban; MBS = marginal budget share; ABS = average share; Y = income. 165 TABLE A.4 ESTIMATED RANGE OF AGRICULTURE'S RURAL VALUE ADDED SHARE IN GROSS OUTPUT Gross OutLput Purchased _Inputs v 1. Annual Food Crops Mixed food crops N. Nigeria, 1974 99.73 11.16 .89 120.44 11.97 .90 148.97 10.07 .93 Millet/sorghum Uganda, 1972 147.00 2.0 .99 Maize Ghana, 1971 166.0 4.0 .98 Kenya, 1974 317.0 24.0 .92 Malawi, 1972 (modern) 85 33.0 .61 Rainfed rice Cameroon, 1976-7 152.0 9.0 .94 Ivory Coast, 1974-5 203.0 12.0 .94 Liberia, 1972 122.0 7.0 .94 Sierra Leone, 1971 112.0 7.0 .94 2. Annual Cash Crops Cotton, smallholders Malawi, 1972 (modern) 256.0 68.0 .73 Tanzania, 1976 114.0 1.0 .99 Cotton, large holdings Sudan, Gezira 1973 7,012 775 .89 3. Perennial Cash Crops Cacao, estates Cameroon, 1971 154.0 33.0 .79 Coffee, estates Kenya, 1975 1,325 404.0 .70 Kenya, 1976 1,325 477.0 .64 Coffee, smallholdings Kenya, Nyeri 1977 2,535 200.0 .92 Pineapple, smallholdings (contract) Ivory Coast, Bonoua 2,262 1,455.0 .36 Ivory Coast, Bonoua 2,715 1,401.0 .48 Pineapple, estates Cameroon, Nyombe 9,654 5,497 .43 166 Table A.4 (CONT.) 4. Livestock Dairy Farms, smallholdings Kenya, S. Baringo 1977 .80 9 .89 Kenya, Kaimbu 1977 311 46 .85 Kenya, Kericho 1977 175 47 .73 Dairy Farms, large holdings Kenya, Njoro 1977 82 16 .80 SOURCE: Matlon (1979); Ruthenburg (1980). 167 TABLE A.5 ESTIMATED RANGE OF NONFARM RURAL VALUE ADDED SHARE IN GROSS OUTPUT Gross Out ut Purchased Inputs v Bak ing Botswana, 1982 375 153 .41 Sierra Leone, 1975 5,015 2,288 .47 Botswana, 1982 156 70 .45 Sierra Leone, 1975 997 739 .74 Millina Botswana, 1982 480 440 .92 C ar2e n.try Sierra Leone, 1975 2,452 1,625 .66 Blacksmithing Sierra Leone, 1975 1,084 754 .70 Cement Block Manufacture botswana,1982 400 170 .43 Home Beer Brewing Botswana, 1982 107 51 .48 SOURCE: Liedholm and Chuta (1976); Haggblade (1982). 168 TABLE A.6 ESTIMATES OF MULTIPLIER PARAMETERS BASED ON REGIONAL INPUT-OUTPUT TABLES FOR TWO REGIONS OF IVORY COAST South East Region Bouake Region V - a v Food crops plus livestock 8.7 89.7 10.5 87.5 Cash crops (cocoa) 6.1 91.4 5.4 91.1 Agroprocessing 36.3 44.8 40.5 48 Wood industries 49.5 43.1 24.3 57.9 Vehicle repair 25.4 50.9 29.7 57.3 Other mechanical 17.4 66.7 6.4 85.3 and electrical Textiles 19.5 66.6. 9.9 56.5 Leather and shoes n.a. n.a. 2.5 94.6 Vegetable and animal oils 42.8 56 45.6 46.8 Transport 26.1 58.1 29.2 59.3 Trade 8.4 90 19.7 78.5 Average 10.5 85.5 15.7 77.6 If 8 = .3, multiplier is 1.57 1.64 SOURCE: Ivory Coast (1965a, 1967a). 169 - BIBLIOGRAPHY Aboagye, A.A. 1986. Informal.Sector Employment in Kenya: A Survey of Informal Sector Activities in Nairobi, Kisamu and Mombasa. Addis Ababa, Ethopia: International Labour office, Jobs and Skills Programme for Africa. Adamu, S.D. 1973. "Socio-Economic Characteristics in Consumer Expenditure Patterns: An Application of Analysis of Covariance," Economic Bulletin of Ghana. Second Series 3(2): 11-21. Adelman, Irma. 1984. "Beyond Export-Led Growth." World Development 12 (9): 937-49. . Adema, W. 1984. "General Report on the Future Role of the Agricultural Machinery Industry in Developing Countries," in Moens and Siepman eds. Development of the Agricultural Machinery Industry in Developing Countries. Wageningen, Netherlands: Pudoc. Adesimi, A.A. 1982. "Economic Analysis of Budgetary Decisions of Farm Families in Ogum State of Nigeria," Savings and Development 6(4): 393- 401. Adesimi, A.A. 1983. "Capital Formation in Traditional Agriculture: A Case Study of Farm Households in Ondo State of Nigeria," Savings and Development 7(2): 151-161. Adeyemo, Remi. 1983. "Credit as an Input in Marketing: A Study of the Nature and Use of Credit by Food Marketers in Anambra State of Nigeria," Savings and Development 7(1): 63-73. Allal, M. and Clinton E. 1982. Cottage Industries and Handicrafts: Some Guidelines for Employment Promotion. Geneva: 1LO. Ahmed, R. and Rustagi, N. 1984. "Agricultural Marketing and Price Incentives: A Comparative Study of African and Asian Countries." (Mimeo) Washington, D.C.: IFPRI. Aluko, S.A.; Oguntoye, 0.A.; and Afonja, A.0. 1972. Small Scale Industries: Western State of Nigeria. Ile-Ife: Industrial Research Unit, University of Ife. Aluko, S.A.; Oguntoye, 0.A.; and Afonja, A.0. 1973. Small Scale Industries: Midwestern, Kwara and Lagos State. Ile-Ife: Industrial Research Unit, University of Ife. Ancey, Gerard. 1971. Relations de vuisinage vile-capagne. Une analyse appliques in Bonake: sa couronne et su region (Cote d'Ivoire). Paris, ORSTOM, Memoires ORSTOM No. 70. 170 Anderson, Dennis. .1982. "Small Industry in Developing Countries: A Discussion of Issues." World Development (Vol. II), pp. 913-948. Anderson, Dennis and Khambata, Farida. 1981. "Small Enterprises and Development Policy in the Philippines: A Case Study." Washington, D.C.: The World Bank, Staff Working Paper No. 468. Anderson, Dennis and Khambata, Farida. 1982. "Financing Small-Scale Industry and Agriculture in Developing Countries: The Merits and Limitations of 'Commercial' Policies." Washington, D.C.: World Bank, Staff WGrking Paper No. 519. Anderson, Dennis and Leiserson, Mark. 1980. "Rural Non-Farm Employment in Developing Countries." Economic Development and Cultural Change. (January), pp. 227-248. Andrae Gunilla. 1981. Industry in Ghana: Production, Form, and Spatial Structure. Uppsala, Sweden: Scandinavian Institute of African Studies. .Anthony, K.R.M., Johnston, B.F., Jones, W.0. and Uchendu, V.C. 1979. Agricultural Change in Tropical Africa. Ithaca, Cornell University .Press. Anthony, K.R.M. and Uchenda, V.C. 1974. Agricultural Change in Geita' District, Tanzania. Nairobi: African Literature Bureau. Ashe, Jeffrey. 1985. The PISCES II Experience: Local Efforts in Micro7 Enterprise Development. 2 Vols. Washington, D. C.: U.S. Agency for International Development-i Austin, Vincent. iJ1. Rural Industrial Development: A Practical Handbook for Planners, Project Managers and Field Staff. London: Cassell.. Barbier, Y. 1973. Etude de Diagnostique Industriel dans la Region du Sud- Ouest. Abidjan, Autorite pour L'Amanagement de la Region du Sud- Ouest. Barker, C.E,; Bhagavan, M.R.; Mitschke-Collande, P.V.; and Wield, D.V. 1986. African Industrialisation: Technology and Change in Tanzania. Aldershot, England: Gower. Becker, Charles M.; deBodisco, Christopher N.; and Monison, Andrew R. 1986. Urban Africa in Macroeconomic and Microeconomic Perspective: Issues and Opinions. Washington, D.C.: World Bank, Water Supply and Urban Development Department, Discussion Paper No. 97. 171 Berry, Sara S. 1975, Cocoa, Custom and Socio-Economic Charge in Rural Western Nigeria. Oxford: Clarendon Press. Bell, Clive; Hazell, Peter and Slade, Roger. 1982. Project Evaluation in Re ional Perspective: A Stud of an Irri2ation 7ro 'e cf i Nort7Ewest aITY a Ff tire: e Johns Hkins 0nVersTf ress. Benin, Republique Populaire du. 1980. ComRtes de la Nation 1975-78. Ministere du Plan de la Satistique et de l'Analyse Economique, Institut National de la Satistique et de L'Analyse Economique. Berry, A. and Sabot, R.H. 1978. "Labor Market Performance in Developing Countries: A Survey." World Development (Nov./Dec.):1199-1242. Binswanger, Rans P. 1983. "Agricultural Growth and Rural Nonfarm Activities." Finance and Development (June):38-40. Binswanger, Hans P. 1986. "Agricultural Mechanization: A Comparative Historical Perspective." The World Bank Research Observer. 1 (1): 27-56. Binswanger, Hans P. and Donovan, Graeme. 1986. "Agricultural Mechanization: Issues and Policies." Washington, D.C.: World Bank, 'Agriculture and Rural Development Department, Report No. 6470. Binswanger, Hans P. and Rosenzweig, Mark R. 1986. "Credit Markets, Wealth and Endowments in Rural South Asia." Discussion Paper No. ARU 59. Research Unit, Agricultura and Rural Development Department, Operational Policy Staff. Washington, D.C.: World Bank. Blayney, Robert G. and Otero, Maria. 1985. "Small and Micro-Enterprises: Contributions to Development and Future Directions for AIDS Support." Washington, D.C.: Employment and Enterprise Development Division. Bureau of Science and Technology. USAID. Boomgard, James J. 1983. "The Economics of Small-Scale Furniture Production and Distribution in Thailand." Ph.D.Dissertation, Michigan State University. Boomgard, James J.; Davies, Stephen P.; Haggblade, Steve, and Mead, Donald C. 1986. "Subsection Analysis: Its Nature, Conduct and Potential Contribution to Small Enterprise Development." East Lansing: Michigan State Univeirsity, Department of Agricultural Economics. MSU International Development Working Paper No. 26. Botswana, Government of. 1972. Report of the Population Census, 1971. Gaborone: Central Statistics Office. 172 Botswana, Government of. 1976. The Rural Income Distribution Survey in Botswana 1974/75. Gabowne, Botswana: Central Statistics Office, Ministry of Finance and Development Planning. Botswana,-Government of, 1982. "Financial Assistance Policy, 1982." Government Paper No.1 of 1982. Gaborone: Ministry of Finance and Development Planning. Bouman, F.J.A. 1977. "Indigenous Savings and Credit Societies in the Third World: A Message." Savings and Development 1(4):184-214. Byerlee, Derek. 1973. "Indirect Employment and Income Distribution Effects of Agricultural Development Strategies: A Simulation Approach Applied to Nigeria." East Lansing: Michigan State University, Department of Agricultural Economics. African Rural Employment Paper No. 9. Byerlee, Derek. 1980. "Rural Labor Markets in West Africa with Emphasis on the Semi Arid Tropics." in 1CRISAT, "Proceedings of the International Workshop on Socioeconomic Constraints to Development of Semi-Arid Tropical Agri'ulture, 19-23 February 1979, Hyderbad, India. pp.348-356. Byerlee, Derek and Eicher, Carl K. 1974. "Rural Employment, Migration and Economic Development: Theoretical Issues and Empirical Evidence from Africa," in Agricultural Policy in Developing Countries. N. Islam, ed. New York: Macmillan, pp. 273-313. Bylerlee, Derek; Eicher, Carl; Liedholm, Carl and Spencer, Dunstan, S.C. 1977. "Rural Employment in Tropical Africa: Summary .of-Findings." African Rural Economy Working Paper No. 20. East Lansing, MI: Michigan State University. Caselli, Clara. 1984. "Aspects of Financial Evolution in the African Countries," Savings and Development 8(3): 217-224. Chernichovsky, Dov; Lucas, Robert E.B. and Mueller, Eva. 1985. "The Household Economy of Rural Botswana: An African Case." Washington D.C.: World Bank. Staff Working Paper No. 715. Child, Frank C. 1977. "Small-Scale Rural Industry in Kenya." Los Angeles: University of California. African Studies Center Occasional Paper No. 17. Child, Frank C. and Kaneda, Hiromitsu. 1975. "Links tothe Green Revolution: A Study of Small-Scale Agriculturally Related Industry in the Pakistan Punjab." Economic Development and Cultural Change 23:249-275. 173 Chuta, Enyinna and L,idholm, Carl. 1979. "Rural Non-Farm Employment: A Review of the State of the Art." East Lansing: Michigan State University, Department of Agricultural Economics. MSU Rural Development Paper No. 4. Chuta, Enyinna and Liedholm, Carl. 1982. "Employment Growth and Change in Sierra Leone Small-Scale Industry, 1974-80." International Labour Review 121(1):101-113. Chuta, Enyinna and Liedholm, Carl. 1985. Employment and Growth in Small- Scale Industry. New York: St. Martins Press. Chuta, Enyinna and Sethuraman, S.V. 1984. Rural Small-Scale Industries and Employment in Africa and Asia. Geneva: ILO. Cisse, Daniel. 1969. Problems de la Formation de 1'Epangne Interne en Afrigue Occidentale. Paris: Editions Presence Africaine. Clark, Mari H. 1985. "Household Economic Strategies and Support Networks of the Poor in Kenya: A Literature Review." Discussion Paper No. UOD-19. Water Supply and Urban Development Department. Washington, D.C.: World Bank. Cleave, John H. 1974. African Farmers:' Labour Use in the Development of Small-holder Agriculture. New Yorx: Praeger. Colclough, Christopher and Fallon, Peter. 1983. "Rural Poverty in Botswana: Dimensions, Causes and Constraints." in Dharam Ghai and Samir Radwan ed. Agrarian Policies and Rural Poverty in Africa. Geneva: ILO. Collier, Paul and Lal, Deepak. 1980. "Poverty and Growth in Kenya." Washington, D.C.: World Bank. Staff Working Paper No. 389. Collier, Paul and Lal, Deepak. 1986. Labour and Poverty in Kenya, 1900- 1980. Oxford: Clarendon Press. Collier, Paul; Radwain, Samir, and Wangwe, Samuel; with Albert Wagner. 1986. Labour and Poverty in Rural Tanzania. Oxford: Clarendon Press. Congo, Republique Democratique de. 1968. Enguete sur les entreprises industrielles, agricoles, commerciales et de service. Kinshasa, Zaire: Ministere de 1'Economie Nationale. Connel, J. and.Lipton,M. 1976. Assessing Village Labor Situations in Developing Countries. New Delhi: Oxford University Press. 174 Cote d'Ivoire, Republique de. 1965a. Etude Regionale de Bouake 1962-1964: la_Synthese generale. Abidjan, Ivory Coast: Ministere du Plan. Cote d'Ivoire, Republique de. 1965b. Etude regionale de Bouake 1962-64: L'economie. Abidgan, Ivory Coast: Ministere du Plan. Cote d'Ivoire, R.ublique de. 1967a. Region du Sud-Est, Etude Socio- economi,que: le commerce et les transports. Paris: Societe d'Etudes pour le Development Economique et Social. Cote d'lvoire, Republique de. 1967b. Region du Sud-Est, Etude Socio- economigue: les budgets familiaux. Paris: Societe d'Etudes pour le Development Economique et Social. Cote d'Ivoire, Republique de. 1967c. Region du Sud-Est, Etude Socio- economique: les Comptes Economiques. Paris: Societe d'Etudes pour le Development Economique et Social. Cote d'Ivoire, Republique de. Ministere du Plan. 1973. "Elaboration d'un Scenario Referential du Developpement Ivoien: L'image Base 1970, Economie, Valeurs de Base, Evolutions Anterieures et Tendances. Paris: SETEF. Cote d'Ivoire, Republique de. 1976. Cote d'Ivoire: Donnees Statistiques sur les Activites Economiques, Culturelles et Sociales. Abidjan, Ivory Coast, Ministere de la Cooperation, Direction des Programmes, Sous-Direction des Etudes Economiques et de la Planification. Cote d'Ivoire, Republique de. 19 . Plan Quinquennal de Developpeinent Economique, Social et Culturel 1976-1980. Volume II. Abidijan, Ministere du Plan. Curtis, Donald. 1973. "Cash Brewing in a Rural Economy." Botswana Notes and Records 5:17-25. d'Agostino, Victoire. 1986. "Sub-Saharan Africa: Household Expenditure Behavior: An Annotated Bibliography.".(Mimeo) Washington, D.C.: World Bank. Dahl, Hans-Erik. 1971. Household Demand for Consumer Goods in Kenya 1963- 65 with Projections for 1966-73. Bergen, Norway: The C. Michelsen Institute, Department of Humanities. Dahomey, Republique du. 1962a. Donnees de Base sur la Situation Demographique au Dahomey en 1961. Paris: Ministere de la Cooperation. Dahomey, Republique du. 1962b. Plan de Development Economigue et Social du Dahomey. 175 Dahomey, Republique au. 1964a. Enquete Demographique au Dahomey, 1961: Resultats, Definitifs. Paris: Ministere de la Cooperation. Dahomey, Republique du. 1964b. Enquete Demographique au Dahomey, 1961, Resultats Definitifs. Paris: Republique Francaise, Ministere de la Cooperation, INSEE. Daines, Samuel R. 1979. Agribusiness and Rural Enterprise Project Analysis Manual. Washington, D..: Practical'Concepts, Inc. Daines, Samuel R. 1979. Agribusiness and Rural Enter prise Proj'ect Analysis Readings. Washington, D.C.: Practical concepts, Inc. DeLancey, Mark W. 1978. "Savi-ngs and Credit Institutions in Rural West Africa: Introduction." Rural Africana 2(Fall):1-8. DeLancey, Virginia. 1978. "Women at the Cameroon Development Corporation: How their Money Works." Rural Africana 2(Fall):9-34. Delgado, Christopher L. 1979. "The Southern Fulani Farming System in Upper Volta: A Model for the Integration of Crop and Livestock Production in the West African Savannah." East Lansing: Michigan State University,.Department of Agricultural Economics. African Rural Economy Paper No. 20. Delgado, Christopher L. 1984. "The Role of Science and Policy in Alleviating Long-Run Food Production Problems in Africa." (Mimeo) Paper presented atthe 67th meeting of the Board for International.Food and Development (BIFAD), Washington, D.C., Dec. 5. Deolalikar, Anuradha B. 1986. "Rural Industralization, The Experiences of India and Thailand: A Literature Study." (Mimeo) paper prepared for the Agriculture and Rural Development. Department of the World Bank. Development Alternatives, Inc. n.d. "Evaluating Small-Scale Enterprise Promotion: State of the Art Methodology and Future Alternatives." Washington, D.C.: DAI. Devres, Inc. 1981. Small Scale Enterprise (SSE) Development: A Summary of Descriptions and Evaluations of SSE Projects and an Assessment of Approaches Used in SSE Development by AID, the World ank and Other FidTgA5gencles. Washington, D.C.: Devres. Due, Jean M. and Mindenda, Timothy. 1985. "Women's Contributions to Farming Systems and Household Income in Zambia." East Lansing: Michigan State University. Office of Women in International Development. Working Paper No. 85. 176 Eckert., Jery B. with Nobe, Kenneth C.; Walters, Forrest E; and Wykstra Ronald A. 1982. Toward the Year 2000: Strategies for Lesotho's Agriculture. LASA Research Report No. 10. Masaru, Lesotho: Ministry f Agriculture, Lesotho Agricultural Sector Analysis Project. Eicher, Carl K. and Baker, Doyle C. 1982. "Research on Agricultural Development in Sub-Saharan Africa: A Critical Survey." East Lansing: Michigan State University, Department of Agricultural Economics, MSU. International Development Paper No. 1. Elliot, Charles. 1983. "Equity and Growth: An Unresolved Conflict in Zambian Rural Development Policy." in Dharam Ghai and Samir Radwan ed. Agrarian Policies and Rural Poverty in Africa. Geneva: ILO. Erasmus, Charles T. 1956. "Culture, Structure and Process: The Occurrence and Disappearance of Reciprocal Farm Labor." Southwestern Journal of Anthrop2ogy 12: 444-469. Etukudo, Abel. 1978. Rural and Urban Household Incomes and Expenditures in Nigeria. Research Bulletin No. 78/06. Lagos, Nigeria: University E7Tgo, Human Resources Research Unit. Farbman, M. 1981. The Pisces Studies: Assisting the Smallest Economic Activities of Urban Poor. Washington, D.C.: USAID. Fluitman, Fred. 1983. "The Socio-Economic Impact of Rural Electrification in Developing Countries: A Review of Evidence." Geneva: World Employment Programme Working Paper, WEP 2-22/WP. 126. Food and Agriculture Organization. 1986. African Agriculture: The Next . 25 Years. Rome, FAQ. Fransman, Martin. ed. 1982. Industry and Accumulation in Africa. London: Heinemann. Freeman, Donald B. and Norcliffe, G.B. 1985. "Rural Enterprise in Kenya: Development and Spatial Organization of the Nonfarm Sector." * University of Chicago, Department of Geography. Research Paper No. 214. Ghai, Dharam and Redwan, Samir. ed. 1983a. Agrarian Policies and Rural Povertyin Africa. Geneva: ILO. Ghai, Dharam and Redwan, Samir. 1983b. "Growth and Inequality: Rural Development in Malawi, 1964-78." in Dharam Ghai and Samir Redwan ed. Agrarian Policies and Rural Poverty in Africa. Geneva: ILO. 177 Ghana, Census Office. 1964. 1960 Popoulation Census of Ghana. Accra, Ghana. Gibb, Arthur. 1974. "Agricultural Modernization, Non-Farm Employment and Low Level Urbanization: A Case Study of a Central Luzon Sub-Region." Ph.D. Thesis, University of Michigan. Goldmark, Susan; Deschamps, Jean-Jacques; Recinnos, Joseph; and Glover, Beatrice. 1982. An Impact Evaluation of The Industrial Bank of Peru's Rural Development Fund. Washington, D.C.: Development Alternatives, Inc. Goldmark, Susan; Mooney, Timothy and Rosengard, Jay. 1982. Aid to Entrepreneus: An Evaluation of the Partnership for Productivity Project in Upper Volta. Washington, D.C.: Development Alternatives, Inc. Goldmark, Susan and Rosengard, Jay. 1983. "Credit to Indonesian Entrepreneurs: An Assessment of the Badan Kredit Kecamatan Program. Washington, D.C.: Development Alternatives Inc. Goldmark, Susan and Rosengard, Jay. 1985. A Manual to Evaluate Small- Scale-En,terrise Develo2ment,Projects. Washington, D.C.: U.S. Agency for International Development, Program Design and Evaluation Methods, Report No. 6. Guha, Sunil. 1974. "The Contribution of Nonfarm Activities to Rural Employment Promotion: Experience in Iran, India and Syria." International Labour Review. 109 (3): 235-250. Haggblade, Steve. 1978. "Africanization from Below: The Evolution of Cameroonian Savings Societies into Western-Style Banks." Rural Africana 2(Fall):35-56. Haggblade, Steve. 1982. Rural Industrial Officer's Handbook. Gabowne, Botswana: Ministry of Commerce and Industry. Haggblade,, Steve. 1984a. "Private Sector Assessment: Synthesis Report for Burkina Faso." Ouagadougou, Burkina: U.S.A.I.D. Haggblade, Steve. 1984b. "The Shebeen Queen." Ph.D. Dissertation, Michigan State University. Haggblade, Steve. 1987. "Vertical Issues in Choice of Technique Studies: A Case Study of Africa's Indigenous Beer Industry."Economic Development and Cultural Change (July, forthcoming). 178 Haggblade, Steve; Liedholm, Carl and Mead, Donald, C. 1986. "The Effect of Policy and Policy Reform on Non-Agricultural Enterprises and Employment in Developing Countries: A Review of Past Experiences." East Lansing: Michigan State University, Department of Agricultural Economics, MSU. International Development Working Paper No. 27. Hamer, Andrew Marshall. 1985. "Decentralized Urban Development: Stylized Facts and Policy Implications." Washington, D.C.: World Bank, Water Supply and Urban Development Department, Discussion Paper No. UDD-73. Hamer Andrew Marshall. 1986. "Urban Sub-Saharan Africa in Macroeconomic Perspective: Selected Issues and Options." Washington, D.C.: World Bank, Water Supply and Urban Development Department, Discussion Paper No. UDD-96. Harper, Malcolm. 1977. Consultancy for Small Business. London: Intermediate Technology Publications. Harper, Malcolm. 1984. Small Business in The Third World: Guidelines for Practical Assistance. New York: John Wily & Sons. Harper, Malcolm and Soon, Tan Thiam. 1979. Small Enterprises in Developing Countries: Case Studies and Conclusions. London: Intermediate Technology Publications. Haugerud, Angelique. 1984. "Household Dynamics and. Rural Political Economy Among Emba Farmers in the Kenya Highlands." Ph.D. Dissertation, Northwestern University. Hawbaker, George and Turner, Howard H. 1972. "Developing Small Industries: A Case Study of AID Assistance in Nigeria, 1962-1971," Washington, D.C,: A.I.D. Hazell, Peter B.R. 1984. "Rural Growth Linkages and Rural Development Strategy." (Mimeo) Washington, D.C.: International Food Policy Research Institute. Paper prepared for the Fourth European Congress of Agricultural Economics, Kiel, Germany, September 3-7, 1984. Hazell, Peter B.R.; Ramasamy, C. and Rajagopalan, V. 1987. "An Analysis of the Growth Linkage Effects of Agricultural Agricultural Growth on the Regional Economy." in Peter Hazell and C. Ramasamy ed. Technological Change and Rural Welfare: A Study of the High Yielding ce arie ies in Tamil Nadu, India. Washington, D.C.: IFPRI, forthcoming. Hazell, Peter and Ramasamy, C. 1986. "Household Expenditure Patterns andthe Growth of the Nonfarm Economy." (Mimeo) Paper presented at the IFPRI/TNAU Workshop on Growth Linkages, Oatacamund, India, Feb. 14-16, 1986. 179 Hazell, Peter, B.R. and Roell, Alisa. 1983. "Rural Growth Linkages: - Household Expendit,ure Patterns in Malaysia and Nigeria." Washington, D.C.: International Food Policy Research Institute. Research Report No. 41. Hedlund, Hans and Landahl, Muts. 1983. Migration and Change, Rural Zambia. Uppsala, Sweden: The Scandinavian Institute of African Studies. Hill, Polly. 1977. Rural Housa: A Village and Setting. Cambridge, England: Cambridge University Press. Hill, Polly. 1963. The Migrant Cocoa Farmers of Southern Ghana: A Study in Rural Capitalism. Cambridge, England: Cambridge University Press. Ho, Sam P.S. 1980. Small-Scale Enterprises in Korea and Taiwan." Washington, D.C.: World Bank. Staff Working Paper No. 384. Hoselitz, Bert F. 1959. "Small Industry in Underdeveloped Areas." Journal of Economic History 19. Hoselitz, Bert F. (ed.) 1968. The Role of Small Industry in the Process of Economic Growth. The Hogue: Mouton. House, William J., Kubagumbe, Dennis and Green, Terry. 1977. "The Potential for Income and Employment Generation in Kenya's Urban Informal Sector: A Proposed Survey." Working Paper No.310. Nairobi, Kenya: Institute of Development Studies. House, William J. and Killick, Tony. 1983. "Social Justice and Development Policy in Kenya's Rural Economy." in Dharam Ghai and Samir Radwan eds. Agrarian Policies and Rural Poverty i*n Africa. Geneya: ILO. Houthakker, H.S. 1957. "An International Comparison of Household Expenditure Patterns Commemorating the Centenary of Engel's Law," Econometrica ( ): 532-551. Howe, Charles W. 1966. "The Use of Sample Household Expenditure Surveys in Economic Planning in*East Africa." Oxford Institute of Economics and Statistics Bulletin 28: 199-209. Hull, Galen Spencer. 1986. A Small Business Agenda: Trends in a Global Economy. New York: University Press of America. Humphrey, David H. and Oxley, Howard S. 1976. "Expenditure and Household Size Elasticities in Malawi: Urban-Rural Comparisons, Demand Projections and a Summary of East African Findings," Journal of Development Studies 12(2): 252-269. 180 Hunt, Diana. 1978. "Chayanov's Model of Peasant Household Allocation and its Relevance to Mbere Division, Eastern Kenya," Journal of Development Studies. 15(1): 59-86. Hunt, Robert W. 1983. "The Evaluation of Small Enterprise Programs and Projects: Issues in Business and Community Development." Washington, D.C.: U.S. Agency for International Development. Evaluation Special Study No. 13. Hunter, John. 1979. "The Successufl and the Unsuccessful Enterprise: An Analysis of Fifty Small Business Enterprises Supported by the Botswana Enterprises Development Unit." Gaborone, Botswana: Institute of Development Management. Hymer, Stephen and Resnick, Stephen. 1969. "A Model of an Agrarian Economy with Nonagricultural Activities," American Economic Review, 50: 493-506. Igbeka, J.C. 1983. "Selecting and Adapting Farm Machinery to Rural Conditions." Agricultural Mechanization in Asia, Africa and Latin America 14(3):45-51. Illy, Hans F. 1978. "How to Build on the Germs of Failure: Credit Cooperatives in French Cameroon." Rural Africana 2(Fall):57-68. Ingham, Barbara M. 1983. "Ghama Cocoa Farmers' Income Expenditure Relationships," Journal of Development Studies 9(3): 365-372. International Labour Office. 1972. Employment, Incomes and Equality: A Strategy for increasing Productive Employment in Kenya. Geneva: ILU. I.L.O. 1975. Report on ILO/ECA/YWCA/SIDA. "Workshop on Participation of Women in Handicrafts and Small Industries." Kitwe, Zambia. pp. 9-20. December, 1974. Geneva: ILO. I.L.O. 1977a. Narrowing the-Gaps: Planning for Basic Needs and Productive Employment in Zambia. Addis Ababa: ILO. Job and Skills Programme for Africa. - I.L.O. 1977b. Reducing Dependence: A Strategy for Productive Employment and Develo ment in SwaziFand. Addis Ababa: ILO. Jobs and Skills- rograme f-o-F-Uri1ca.* I.L.O. 1979a. Emploi et Development an Cameroun: Perspectives Sectorielles. Addis Ababa: ILO. Job and Skills Programme for ~T i ca. 181 I.L.O. 1979b. Options for a Dependent Economy: Development, Employment and Equity Problems in Lesotho. Addis Ababa: ILO. Jobs and Skills Programme for Africa. I.L.O. 1979c. Poverty and Employment in Rural Areas of Developing Countries. Geneva: ILO. Advisory Committee on Rural Development, inth Sesision, 27 November - 6 December 1979. I.L.O. 1980. Dynamigue de 1'Emploi dans un Systeme Sahelien: Le Niger. Addis Ababa: ILO. Job and Skills Programme for Africa. I.L.O. 1981a. Basic Needs In An Economy Under Pressure: Findings and Recommendations of an ILO/JASPA Basic Needs Mission to Zambia. Addis Ababa: ILO. Job and Skills Programme for Africa. I.L.O. 1981b. Employment Problems of Rural Women in Kenya. Addis Ababa: I.L.O. Job and Skills Programme for Africa, Report of a Technical Mission Submitted to the Government of Kenya. I.L.O. 1981c. Ensuring Equitable Growth: A Strategy for Increasing Employment,- EquTy and Basic Needs Satisfaction in Sierra. Leone. Addis Ababa:1L. Job and Sills Programme for Africa. I.L.O. 1981d. First Things First: Meeting The Basic Needs of the People ofNji2ria. Addis Ababa: 1LO Job and Skills Programme for Africa. I.L.O. 1982a. Basic Needs in Danger: A Basic Needs Oriented Development Strategy for Tanzania. Addis Ababa: ILO. Job and Skills Programme for Africa. I.L.O. 1982b. Disparites de revenus entre les villes et les campagnes au Cameroun. Job and Skills Program for Africa (JASPA). Addis Abeba: ILO. I.L.O. 1982c. Disparites de revenus entre les villes et les campagnes au Mali. Job and Skills Program for Africa (JASPA). Addis Abeba: ILO. I.L.O. 1982d. Disparites de revenus entre les villes et les campagnes au Senegal. Job and Skills Program for Africa (JASPA). Addis Abeba: ILO. I.L.O. 1982e. Disparites de rievenus entre les villes et les campagnes au Togo. Job and Skills Program for Africa (JASPA). Addis Abeba: ILO. I.L.O. 1982f. Disparites de revenus entre les villes et les campagnes en Cote d'Ivoire. Job and Skills Program for Africa (JASPA). Addis Abeba: ILO. I.L.O. 1982g. Disparites de revenus entre les villes et les campagnes en Haute-Volta. Job and Skills Program for Africa (JASPA). Addis Abeba: T L~Ui 182 I.L.O. 1982h. Disparites de revenus entre les villes et les campagnes en Republigue Populaire du Congo. Job and Skills Program for Africa (JASPA. Addis Abeba: ILU. I.L.O. 1982i. Disparites de revenus entre les villes et les campagnes en Afrique noire francophone: rapport de synthese. Job and Skills Program for Africa (JASPA). Addis Abeba: ILO. I.L.O. 1982j. Rural-Urban Gap and Income Distribution (A Comparative Sub- Regional Study): The Case of Ghana. Addis Ababa: ILO. Job and Skills Programme for Africa. I.L.O. 1982k. Rural-Urban Gap and Income Distribution (A Comparative Sub- Regia e C 0f Kenya. Addis Ababa: 1LO. Job and Skills-?Programme for Africa. I.L.O. 19821. Rural-Urban Gap and Incomhe Distribution (A Comparative Sub- Regional Study): The Case of Lesotho. Addis Ababa: ILO. Job and Skills Programme for Africa. I.L.O. 1982m. Rural-Urban Gap and Income Distribution (A Comparative Sub- Regional Study): The Case of Liberia. Addis Ababa:. ILO. Job and Skills Programme fo.r Africa. I.L.O. 1982n. Rural-Urban Gap and Income Distribution (A Comparative'Sub- Regional Study): The Case of Nigeria. Addis Ababa: ILO. Job and Skills Programme for Africa. I.L.O. 1982o. Rural-Urban Gap and Income Distribution (A Comparative Sub- Regional Study): he Case of Siena Leone. Addis Ababa: ILO. Job and Skil s Programme for Africa. I.L.O. 1982p. Rural-Urban Gap and Income Distribution (A Comparative Sub- Regional Study): The Case of Tanzania. Addis Ababa: ILO. Job and Skills Programme for Africa. I.L.O. 1982q. Rural-Urban Gap and Income Distribution (A Comparative Sub- Regional Study): The Case of Zambia. Addis Ababa: ILO. Job and Skills Programme for Africa. I.L.O. 1983a. Elements pour une Strategie d'Emploi dans les Zones Rurales en Togo. Addis Ababa: ILO. Job and Skills Programme for Africa. I.L.O. 1983b. Patterns of Industrialization and Impact on Employment and Incomes in African Countries: The Case of Ethiopia. Addis Ababa: ILu. Job and Skills Programme for ATrica. 183 I.L.O. 1983c. Patterns of Industrialization and Impact on Employment and Incomes in African Countries: The Case of Ghana. Addis Ababa: LO. Job and Skills Programme for Africa. I.L.O. 1983d. Patterns of Industrialization and Impact on Employment and Incomes in African Countries: The Case of Kenya. Addis Ababa: ILO. Job and Skills Programme for Africa. I.L.O. 1983e. Patterns of Industrialization and Impact on Employment and Incomes in African Countries: The Case of Liberia. Addis Ababa: ILO. Job and Ski lls Programme for Af-rica. I.L.O. 1983f. Patterns of Industrialization and Impact on Employment and Incomes in African Countries: The Case of Nigeria. Addis Ababa: ILO. Job and Skills Programme for Africa. I.L.O. 1983g. Patterns of Industrialization and Impact on Employment and Incomes in African Countries: The Case of Sierra Leone. Addis Ababa: ILO. Job and Skills Programme for Africa. I.L.P. 1983h. Politgue et Strategie d'Industrialisation en Afrique: Effects sur L'emploi et la distribution des revenus: Cas en Burundi. Addis Ababa: ILO. Job and Skills Programme for Africa. I.L.O. 1983i. Promotion of Employment and Incomes for the RrlPo,or2 IncludingRural Women, hrough Non-Farm Activmi 1es. Advry Committee on Rural Development, Tenth Session. -7eneva. 22 November to 1 December 1983. Geneva: ILO. I.L.O. 1983j. Politique et Strategie d'Industrialisation en Afrique: Effets sur 1'Emploi et la Distribution des Revenus: Cas du Rwanda. Addis Ababa: ILO. Job and Skills Programme for Africa. I.L.O. 1984. Politique et Strategie d'Industrialisation en Afrigue: Effets sur 1'Emploi et la Distribution des Revenus: Cas du Madagascar. Addis Ababa: ILO. Job and Skills Programme for Africa. I.L.O. 1985a. Informal Sector in Africa: Jobs and Skills Programme for Africa. Addis Ababa: LO. I.L.O. 1985b. Politique et Strategie d'Industrialisation en Afrique: Eflets sur I'Emploi et la Distribution des Revenus: Cas du Senegal. Addis Ababa: ILO. Job and Skills Programme for Africa. I.L.O. 1985c. Politique et Strategie d'Industrialisation en Afrique: Effets sur 1'Emploi et la Distribution des Revenus: Cas du Togo. Addis Ababa: ILO. Job and Skills Programme for Africa. 184 I.L.O. 1985d. Labor Use and Productivity and Technolo ical Ch an n African Smalhol er riculte he Caseof Gama. Addis Ababa: 1LO. Job and Skills Programme for Africa. I.L.O. 1985e. Labor Use and Productivity and Technological Change in African Smallholder Agriculture: The Case of Ghana. Addisbaba: ILO. Job and Skills Programme for Africa. I.L.O. 1985g. Labor Use and Productivity and Technological Change in African Smallholder Agriculture: The Case of Sudan. Addis Ababa: ILO. Job and Skills Programme for Africa. I.L.O. 1985g. Labor Use and Productivity and Technological Change in African Smallholder Agriculture: The Case of ZUbanda. Addis Ababa: ILO. Job and Skills Programme for Africa. ItL.O. 1985h. Labor Use and Productivity and Technological_ Cange in African Smal]holder-Agriculture: The Case of Zimbabwe. Addis Ababa: ILU. jot) andU7k1 Iis Programme for Af-rica. I.L.O. 1986. Bibliography of Published Research of the World Employment Programme. 6th ed. Institute of Development Management. 1978. Things Fall Apart: An Analysis of the Problems of 13 Small Botswana Enterprises. Gaborone, Botswana: Institute of Deyelopment Management. Jaeger, William K. 1986. Agricultural Mechanization: The Economics of Animal Draft Power in West Africa. Boulder, Colorado: Westview Press. Johnston, Bruce F. 1970. "Agriculture and Structural Transformation in Developing Countries: A Survey of Research," Journal of Economic Literature 8(2):369-404. Johnston, Bruce F. 1981. "Farm Equipment Innovations and Rural Industrialization in Eastern Africa: An Overview." Geneva: International Labour Office. World Employment Programme Research Working Paper, WEP 2-22/W.P. 80. Johnston, Bruce F. and Kilby, Peter. 1975. Agriculture and Structural Transformation: Economic Strategies in Late-Develoing Countries. London: Oxford University Press. Johnston, Bruce F. and Mellor, John W. 1961. "The Role of Agriculture in Economic Development, " American Economic Review 51(4): 566-593. 185 JQnes, Christine. 1986. "Household Expenditure Patterns: A Review of the Evidence from Sub-Saharan Africa." (Mimeo) Washington, D.C.: World Bank, Special Office for Africa. Jules-Rosette, Bennetta. 1982. "Women's Work in the Informal Sector: A Zambian Case Study." East Lansing: Michigan State University. Office Of Women in International Development. Working Paper No. 3. Kelley, Allen C. 1981. "Demographic Impacts on Demand Patterns in the Low Income Setting," Economic Development and Cultbral Change (1): 1-16. Kenya, Republic of. 1978. Statistical Abstract 1978. Nairobi, Kenya: Ministry of Economic Planning and Community Affairs, Central Bureau of Statistics. Kenya, Republic of. 1981. The Integrated Rural Surveys, 1976-1979. Basic Report. Nairobi, Kenya: Ministry of Economic Planning and Development, Central Bureau of Statistics. Kilby, Peter. 1962. The Development of Small Industry in Eastern Nigeria. Lagos, Nigeria: US,AID. Kilby, Peter. 1965. African Enterprise: The Nigerian Bread Industry. Stanford, CA: Hoover Institution. Kilby, Peter. 1969. Industrialization in an Open Lconomy: Nigeria 1945- 1966. Cambridge: Cambridge University Press. Kilby, Peter. 1979. "Evaluating Technical Assistance." World Development 7(3). Kilby, Peter. 1982. "Small Scale Industry in Kenya." East Lansing: Michigan State University, Department of Agricultural Economics. MSU Rural Development Series Working Paper No. 20. Kilby, Peter. 1987. "Small-Scale Manufacturing in Kenya," (mimeo) Washington, D.C.: Smithsonian Institution, The Wilson Center. Kilby, Peter and Bangasser, P. 1978. "Assessing Technical Cooperation: The Case of Rural Industry." International Labour Review. 117: Kilby, Peter and D'Zmura, David. 1985. "Searching for Benefits." Washington, D.C.: U.S. AID. Evaluation Special Study No. 28. Kilby, Peter and Liedholm, Carl. 1986. "The Role of Nonfarm Activities in the Rural Economy." Employment and Enterprise Development Policy Analysis Discussion Paper #7. Cambridge, MA: Harvard Institute for International Development. 186 Kilby, Peter; Liedholm, Carl and Meyer, Richard. 1984. "Working Capital and Rural Nonfarm Enterprises." in D. Adams, 0. Graham and J.D. Von Pischke eds. Undermining Rural Development with Cheap Credit. Boulder, Colorado: Westview Press. King, Kenneth, 1977. The African Artisan: Education and The Informal Sector in Kenya. London: Heinemann. King, Robert P. and Byerlee, Derek. 1977. "Income. Distribution, Consumption Patterns and Consumption Linkages in Rural Sierra Leone." East Lansing: Michigan State University, Department of Agricultural Economics, African Rural Economy Paper No.16. King, Robert P. and Byerlee, Derek. 1978. "Factor Intensities and Locational Linkages of Rural Consumption Patterns in Rural Sierra Leone." American Journal of Agricultural Economics (May): 197-206. Kitching, G.N. 1977. Economic and Social Inequality in Rural East Africa: The Present as a Clue to the Past. Monograph No. 1. Swansea: University of Swwnsea,Centre for Devlopment Studies. Kossoudji, Sherrie and Mueller, Eva. 1983. "The Economic and Demographic Status of Female-Headed Households in Rural. Botswana." Economic Development and Cultural Change 31(4):831-860. Kreamer, Ross G. 1986. "Gari Processing in Ghana: A Study of Entrepreneurship and Technical Change in Tropical Africa." Cornell International Agrticultural Economics Study, AE Research 86-30. Ithaca, New York: Cornell University, Department of Agricultural Economics. Krishna, Raj. 1976. Rural Unemployment: A Survey of Concepts and Estimates for India. Staff Working Paper No.234. Washington, D.C.: W=ord bank. Lang, Mahlon; Cantrell, Ronald P.; Ohm Herbert, W. and Sawadogo, Sibiri. 1984. Farming Systems Research Unit in Upper Volta, 1983 Annual Report. West Lafayette, Indiana; Purdue University, International Programs in Agriculture. Lassen, Cheryl; Traore, Richard; Brown, Alex and Walton, Joshua. 1985. "Credit and Enterprise Development Training That Reach The Small Producer Majority in Burkina Faso: Midterm Evaluation of Association pour la Productivite/Burkina Faso." Washington, D.C.: Partnership for Productivity.