RESEARCH REPORT COMMERCIALIZATION OF AGRICULTURE UNDER POPULATION PRESSURE: EFFECTS ON PRODUCTION, CONSUMPTION, AND NUTRITION IN RWANDA Joachim von Braun Hartwig de Haen Juergen Blanken IFPRI Research Reports Publications Review Committee and Procedures Gunvant Desai, Chairman Romeo Bautista Dayanatha Jha Joachim von Braun Marc Nerlove Maarten Immink Richard Sabot Nurul Islam Barbara Rose (ex officio) All manuscripts submitted for publication as IFPRI Research Reports undergo extensive review. Prior to submission to the Publications Review Committee, manuscripts are circulated informally among the author's colleagues and presented in a formal seminar. Following submission to the Review Committee, four to five reviewers are selected. At least two reviewers are identified from outside IFPRI, one from inside IFPRI, and one from the Review Committee. Reviewers are chosen for their expertise in the manu- script's subject matter and methodology and, when applicable, their familiarity with the country setting. The author responds in writing to the reviewers' comments and resubmits the manuscript to the Review Committee after making any necessary revisions. The Review Committee then makes its recommendation on publication of the manuscript to the Director General of IFPRI. With the Director General's approval, the manuscript becomes part of the IFPRI Research Report series. IFPRI Board Gerry Helleiner Chairman, Canada Harris Mutio Mule Vice Chairman, Kenya Sjarifuddin Baharsjah Indonesia Anna Ferro-Luzzi Italy Ibrahim Saad Ahmed Hagrass Egypt Yujiro Hayami Japan James Charles Ingram Australia of Trustees Roberto Junguito Colombia Dharma Kumar India Theodore W. Schultz U.S.A. Leopoldo Solis Mexico M. Syeduzzaman Bangladesh Charles Valy Tuho Cote d'lvoire Just Faaland, Director Ex Officio, Norway COMMERCIALIZATION OF AGRICULTURE UNDER POPULATION PRESSURE: EFFECTS ON PRODUCTION, CONSUMPTION, AND NUTRITION IN RWANDA Joachim von Braun Hartwig de Haen Juergen Blanken Research Report 85 International Food Policy Research Institute Copyright 1991 International Food Policy Research Institute. All rights reserved. Sections of this report may be reproduced without the express permission of but with acknowledgment to the International Food Policy Research Institute. Library of Congress Cataloging- in-Publication Data Von Braun, Joachim, 1950- Commercialization of agriculture under popu- lation pressure : effects on production, consump- tion, and nutrition in Rwanda/Joachim von Braun, Hartwig de Haen, Juergen Blanken. p. cm.—(Research report/ International Food Policy Research Institute; 85) Includes bibliographical references. ISBN 0-89629-087-5 1. Agriculture and state—Rwanda. 2. Pro- duce trade—Government policy—Rwanda. 3. Ex- ports—Rwanda. 4. Food supply—Government policy—Rwanda. 5. Nutrition policy—Rwanda. 6. Rwanda—Population. I. Haen, Hartwig de, 1942- , n. Blanken, Juergen, 1957- , m . Title. IV. Series: Research report (International Food Policy Research Institute); 85. HD2127.5.Z8V66 1991 90-27251 338.1'867571—dc20 CEP CONTENTS Foreword 1. Summary 11 2. Research Objectives 14 3. The Forces Driving the Commer- cialization Process 16 4. Theoretical Foundation, Research Design, and Data Base 28 5. Production and Income Effects of the Commercialization Pro- cess 38 6. Consumption Relationships and Effects of Commercialization 64 7. Consumption-Nutrition-Health Links 85 8. Long-run Perspectives for Rural Development 96 9. Conclusions and Policy Implica- tions 109 Appendix 1: Survey Design 112 Appendix 2: Supplementary Tables 115 Bibliography 117 TABLES 1. Some basic features and long- term trends of Rwanda's econ- omy, 1965-85 17 2. Total arable land area, land use for export crops, and popula- tion density in Rwanda, 1961- 83 18 3. Product mix in different agro- ecological zones in and around the study area, 1984 21 4. Total and harvested tea areas, production of dry tea, and yields of dry tea per hectare in Rwanda, 1962-85 23 5. Harvested area and produc- tion of plantation and small- holder tea for Rubaya and Nyabihu factories, 1980-85 24 6. Household size and landhold- ings, and age of household head, 1985/86 38 7. Alternative concepts assessing household subsistence orienta- tion, by person-land ratio and total expenditure quartiles, 1985/86 39 8. Farm size, subsistence orienta- tion in agricultural production (concept 1), and main sources of market integration, 1985/86 40 9. Subsistence orientation of agri- culture (concept 1) and income sources, 1985/86 41 10. Subsistence orientation (concept 2) and income from other sources, 1985/86 41 11. Income by farm-size group, 1986 42 12. Farm size, source of land acqui- sition, and labor-land relation- ships in subsample households, 1986 44 13. Land allocation of subsample households, 1986 45 14. Average labor time allocation in person-days per adult family member, 1985/86 46 15. Allocation of total labor input by crop and by family and non- family labor, 1985/86 47 16. Distribution of family labor in- put by sex and age groups and by activities for different crops, 1985/86 48 17. Distribution of total nonfamily labor input and type of nonfam- ily labor, by activities for differ- ent crops, 1985/86 49 18. Family and total labor input for different crops and farm-size quartiles, 1985/86 50 19. Stated reasons for growing or not growing a particular crop, 1985/86 51 20. Average land and labor produc- tivity of different cropping sys- tems, subsample households, 1985/86 53 21. Aggregate agricultural produc- tion function estimates 55 22. Labor productivity in agricul- ture, by farm and household char- acteristics, 1985/86 56 23. Determinants of labor produc- tivity in agriculture 57 24. Off-farm work by age andsex in different time periods, 1985/86 59 25. Off-farm work by type of em- ployment for men, women, and children, 1985 and 1986 60 26. Off-farm work of men, women, and children, by type of em- ployment and farm-size quar- tiles, 1986 61 27. Regression analysis of alloca- tion of work time to off-farm work 62 28. Expenditure shares of foods and nonfoods, by expenditure quar- tiles, 1986 65 29. Annual expenditures on foods and nonfoods, by quartiles of total expenditure per capita, 1986 66 30. Monthly expenditures on foods and nonfoods per capita in dif- ferent seasons, 1986 67 31. Regression analysis of deter- minants of the degree of subsis- tence orientation in consump- tion 68 32. Model of determinants of food expenditures 70 33. Sources of calories per adult- equivalent per day, by expendi- ture quartiles, April-June 1986 73 34. Calorie consumption per per- son, by expenditure and farm- size quartiles, survey round 2, May-June 1986 74 35. Calories per adult-equivalent per day, by subsistence quartiles, April-June 1986 75 36. Time spent on household water and wood fetching, by calorie consumption group, 1985/86 77 37. Determinants of calorie con- sumption 79 38. Farm characteristics and pro- duction of displaced and other farmers, 1985/86 82 39. Off-farm work of displaced and other farm households, 1985/86 83 40. Expenditures and food consump- tion of displaced and other farm households, 1986 83 41. Prevalence of malnutrition among children aged 6-72 months in the sample popula- tion, 1985/86 86 42. Prevalence of malnutrition among children aged 6-72 months, by household calorie- consumption levels, 1985/86 87 43. Prevalence of malnutrition among children aged 6-72 months, by various socioeco- nomicand farm-household char- acteristics, March 1986 88 44. Multivariate analysis of deter- minants of nutritional status of children aged 6-72 months, 1985/86 91 45. Shares of different age groups in total sample population, 1985- 2005 100 46. Person-land ratio and consumer- worker ratio, 1985-2005 100 47. Distribution of sample popula- tion by person-land-ratio quar- tiles, 1985-2005 101 48. Shares of different crops and cropping systems in total farm size, yields, and labor use, by altitude and person-land-ratio group, 1985 102 49. Average production of major crops and total calorie produc- tion, 1985-2005 104 50. Average agricultural labor in- put, production of major crops, and total calorie production per consumer-equivalent, by person- land-ratio quartiles, 1985-2005 105 51. On- and off-farm labor alloca- tion and distribution, by per- son-land-ratio quartiles, 1985- 2005 107 52. Development of overall food- energy production in relation to minimum requirements at house- hold level, 1985-2005 107 53. Population density and land use for potatoes and tea in secteurs of Giciy e commune, 1985 113 54. Age-specific birth rates of Gis- enyi prefecture, 1983 115 55. Death rates by sex and age co- horts used for the demographic model 115 56. Coefficients used to calculate the number of adult-equivalent persons 116 57. Coefficients used to calculate the number of consumer-equiva- lents 116 ILLUSTRATIONS 1. Map of the study area 19 2. Commercialization at the house- hold level: determinants and con- sequences for income, consump- tion, and nutrition 29 3. Allocation of household time between home goods produc- tion, farming for the market, wage earning, and leisure 32 4. Resource allocation to market versus subsistence production under risk 34 5. Seasonal distribution of off-farm labor, by kind of occupation 60 FOREWORD Rapid population growth in agroecologies that are already under high population pressure poses a major challenge for development policy. It becomes an even greater challenge in complex agroecologies where little new technology for rapid agricultural expansion is available. The mountain zones of the Zaire-Nile Divide in Central Africa present an example of such a challenging environment where agriculture has encroached onto marginal zones, that is, water catchment areas and the last tropical forests of the area. This study by von Braun, de Haen, and Blanken highlights the potentials of agricultural development for the employment, income, and consumption of the poor, but also stresses that nonagricultural rural growth and employment expansion are key to improved food security and nutrition in this setting. The authors show that the delivery of public goods—health services, sanitation, and education—has to move ahead in order to maintain and improve the human capital foundation in this stressed environment. The study is based on detailed primary household data utilized in innovative ways to assess the household's and farmer's (and her husband's) behavior in the subsistence economy vis-a-vis options for specialization. A number of interesting policy findings emerge, such as the poor being too poor to capture the gains from efficient specialization because they need to take care of subsistence-based insurance against hunger. While generally favorable effects of commercialization of agriculture for nutrition are manifested by this study and preceding IFPRI studies on this topic, the present study also draws attention to the need for concern about land tenure when the stimulus for agricultural commercialization is given in a land-scarce environment. Just Faaland Washington, D.C. January 1991 ACKNOWLEDGMENTS This research was made possible by a grant to IFPRI from the German Agency for Technical Cooperation (GTZ) and by financial support from the Volkswagen Founda- tion to the Institute of Agricultural Economics of the University of Goettingen, which collaborated with IFPRI in this undertaking. Much of the extensive fieldwork for this study in Rwanda was supervised by Joanne Csete, who also contributed to the design and field-testing of questionnaires for the complex survey, especially in the areas of consumption and nutrition. The large data sets were handled in a most efficient way by Graciela Wiegand-Jahn, who also contributed greatly to the analysis and interpretation of results. The Ministry of Agriculture of the Republic of Rwanda and the GTZ project in Giciye provided institutional and logistical support. Bernard Mutwewingabo from the ministry was a helpful collaborator. Serge Rwamasirabo and Yvan de Jaeger from the Enquete Agricole project of the ministry provided valuable advice. We also benefited from a careful review of an earlier draft of this report by Joseph Laure of the Office de la Recherche Scientifique et Technique Owfre-Mer-Institute of Nutrition of Central Amer- ica and Panama (ORSTOM-INCAP), Tshikala Tshibaka, Dayanatha Jha, and anony- mous reviewers. We are indebted to the bourgmestre of Giciye commune for generously giving his support and office space. We are grateful for the efforts of the survey teams and for the patience of the household respondents who gave their valuable time. Joachim von Braun Hartwig de Haen Juergen Blanken SUMMARY Increased market integration of traditional agriculture is part of a development strategy oriented toward growth. Integration in the local, national, and international exchange economies promises gains through specialization. But it is the design of programs and policies and their actual implementation that determine whether or not the poor obtain a fair, or even a positive, share of gains from agricultural commercialization, directly or indirectly. Gains for the poor are not a priori assured, and numerous cases quoted in the literature—though not well documented and frequently methodologically flawed—point to adverse effects of increased market integration on the welfare of the poor, including their nutrition. This study, which is part of a larger effort at IFPRI, looks in detail into the effects of commercialization on production, income, employment, consumption, and nutrition. The study location is in Rwanda and is among Sub-Saharan Africa's most densely populated rural areas. The specific objectives of this research are, first, to assess the effects of the interaction between increased commercialization and population growth on production, household real incomes, family food consumption, expenditures for nonfood goods and services, and the nutritional status of the population, and second, to develop a long-term perspective, based on household analysis, of the implications of this change for agricultural, infrastructural, employment, and nutrition policies. The empirical analysis of the research is based upon a detailed survey of production, income and consumption, nutrition of individuals, and health in about 200 households. The survey was undertaken during 1985/86 in an area in the high-altitude zone of the Zaire-Nile Divide in northwestern Rwanda. The study area is very densely populated, and a high population growth rate (4.2 percent) increases the pressure. The study site also is undergoing agricultural commercialization induced especially by the introduction of tea production and the expansion of potato production for the market. While tea and potatoes play important roles in the overall commercialization of rural households via product and labor markets, other important forces identified by the survey are nonagri- cultural off-farm employment and home production activities, the latter especially referring to the brewing of sorghum beer as a more traditional form of commer- cialization. The economic analysis of tea production in the study area concludes that the crop is not adding to aggregate income there, because competition with other crops (cereals, pulses, roots, and tubers) is strong, and established tea factory capacities are underuti- lized, leading to high fixed costs per unit of output. One parastatal tea factory in the study area responded to these efficiency problems by externalizing costs: smallholders were expropriated and land was added to the factory-managed plantation to reach higher- capacity utilization. Such an aberration of commercialization and the finding of lack of comparative advantage of tea are departures from the generally successful introduction of tea into Rwanda's economy. 11 Commercial potato production in the area is done in a former natural forest area (Gishwati forest), partly on licensed plots obtained from a reforestation project on a temporary basis and partly in an uncontrolled form without such entitlements. This potato production is concentrated in larger holdings—that is, in the context of this smallholder system, on farms with more than 1.5 hectares. The average farm size in the sample is 0.7 hectare per household. While central to current employment and household food availability, potato production in the forest area—a key water-catchment area— poses risks to sustainability of the production environment in the longer run. Agriculture in the study area is still very subsistence-oriented. On average, 67.5 percent of the value of agricultural production is consumed by the households, thus aggregate marketed surplus is 32.5 percent. Sorghum beer and potatoes account for about 30 percent each of agricultural sales, and the remaining 40 percent is from livestock, tobacco, tea, and occasional sales of subsistence crops such as sweet potatoes and sorghum. Yet a look at agriculture alone would give a biased impression of overall rural commercialization. The great majority of farm households have sizable nonagricultural incomes. Off-farm income is on average 57.5 percent of total income, and in the smallest farm-size quartile it reaches 80.1 percent. Off-farm employment is largely found by men, while women do most of the fteldwork. Women account for 74.1 percent of family labor input in agriculture. Only in single-crop potato production with modern inputs do men contribute an important share of labor input. A significant share of the work force on the tea plantations—mainly for plucking—consists of women (19 percent in 1985). So far, this is the only significant off-farm employment of women in the area. Subsistence orientation—that is, the share of own-produced food in total per capita food and nonfood consumption—is remarkably stable across different farm sizes and per capita income levels. The value of own-produced food in percentage of the total value of food and nonfood consumption is on average 47.8 percent. Multivariate analysis shows that, all else holding constant, this value shrinks by only 1.3 percent with a 10.0 percent increase in land scarcity (person-land ratio). Although options for gains from commercialization and specialization appear to exist, the poor farmers in this setting forgo them to a large extent because of the need for food insurance (subsistence). They are thus too poor to opt for an "efficient" production pattern. Yet from their household perspective, efficient resource utilization—including that for food insurance—needs to be factored in. The critical development issues in this very densely populated region relate to employment and labor productivity. Gross margins per labor day for major crops in 1985/ 86 ranged between US$0.44 for sorghum in mixed cultivation and US$0.71 for sweet potatoes and were about US$0.60-0.70 for potatoes and maize. Econometric estimates point to the interesting finding that increased land scarcity in the study area, due to the rapid population growth (4.2 percent a year in the 1980s), can still be substantially compensated for by intensification of labor and capital input per unit of land. The indigenous mechanisms for increasing labor productivity under increased land scarcity are found to be sizable: a 10 percent increase in the person-land ratio results in only a 3.6 percent decline in labor productivity. While this is encouraging, given the already extremely high person-land ratio in the area (5.5 adult-equivalent persons per hectare), it also stresses the increased need for technological change. 12 The major proportion of incremental income that households earn is spent on food. For an average household a 10 percent increase in income leads also to a 10 percent increase in the consumption value of food, and to an increase of 5 percent in calorie consumption. Richer households spend much more for a more diverse diet. Conse- quently, households in the top per capita income quartile spend 77 percent more per calorie than households in the poorest income quartile. Hunger, that is, calorie deficiency, is a problem in a large proportion of households. Calorie consumption has only a weak link to farm size but a strong one to income. In mid- 1986, 41 percent of the households consumed less than 80 percent of recommended requirement levels—a commonly used critical cutoff point below which calorie con- sumption levels are seriously deficient. Among the calorie-deficit households is a large share of those who were continuously in deficit over the survey period (about 60 percent of the deficit households). In these households in particular, the nutritional status of children deteriorates, and increased growth retardation and underweight are in fact diag- nosed for the children. In the total sample population in 1986,21.5 percent of all child- ren below 7 years of age are identified as stunted (below 90 percent of the height-for-age standard), and 12.3 percent are substantially underweight (below 80 percent of the weight-for-age standard). Scarcity of subsistence food, cash, and time impinges on the nutritional outcome. Time constraints of adults—that is, women in the most resource- poor households—lead to a shift of labor demand to children for the production of home goods, such as fuelwood collection and water fetching. Children fulfill these tasks to a much larger extent in calorie-deficient households than in others. Increasing household calorie consumption is important but alone does not solve the nutritional status problem. The effects of the health environment and household sanitation on children's anthropometric status are very substantial. Doubling household calorie consumption from 1,500 to 3,000 calories per adult-equivalent—an extreme change indeed—would reduce stunting by about a quarter of a standard deviation (or 17 percent of the Z-score mean), whereas a worm cure would have the same effect, and a clean latrine would have twice this impact on nutritional status. This underlines the role of improved health services, accessible to the poor, in nutritional improvement. Richer households spend substantially more on health care than the poor (calorie-deficient households) can afford. Long-term simulations with a demographic model stress the important role of the area's rapid population growth in the rural transformation process during the next two decades. The rapidly rising person-land ratio is expected to further increase intensifica- tion of food crop production with higher labor inputs per unit of land. A substantial absolute and relative expansion of sweet potato production stands out as a result and points to the importance of an increased focus on technological change in this crop. In a rapidly increasing share of rural households, self-sufficiency in staple foods will drop very fast within the coming decade. The person-land ratio in the area will increase from 5.5 adult-equivalent persons per hectare in 1985 to 12.0 per hectare in 2005. Most dramatic is the labor-supply expansion for nonagricultural employment that will, even under cautious assumptions in the simulations, more than double from its already high level. 13 RESEARCH OBJECTIVES Commercialization of the rural sector is considered a cornerstone of successful eco- nomic development. It allows increased participation of individuals and households in the domestic and international exchange economy. Through realization of comparative advantages, it is supposed to benefit not only individual rural families but also the agricultural sector and the whole economy. Commercialization may have many facets in this context. Generally speaking, it describes an individual's or a household's economic transactions with others. These may be both in cash and in kind, the latter playing a considerable role in many traditional communities. Transactions may relate to agricultural produce, indicating that a certain proportion of a farm's output is not produced for subsistence but for sale. They may also relate to inputs, indicating that a farm's production technology depends to a certain extent on external inputs. Finally, a household may also be commercialized by earning off-farm income, mostly from labor employment outside the household, but possibly also from capital investment. Evidently, all these transactions will not only enable a rise in a family's or an individual's income; they may also improve the nutritional situation, provided there is a preference for better nutrition and provided the individuals are able to express these preferences and get access to increased food supplies. While the first condition—high preference for food—will be mostly fulfilled, the latter—realization of preferences and market access—may not in cases of market failure or policy failure. In fact, a survey of research findings concerning the effect of commercialization in agriculture on the alleviation of poverty, on the distribution of food, and on the nutritional status of vulnerable groups has shown mixed results (von Braun and Kennedy 1986). Insufficient food consumption to meet nutritional requirements is closely related to poverty, and a significant portion of increasing incomes among the poor would be expected to be spent on more food. If low-income fanners and landless laborers capture at least part of the economic surplus generated by shifts from subsistence to cash crop production, and if a portion of these people are malnourished, one would expect that the nutritional status would improve. Conceptually, these relationships are, however, not straightforward, and a review of the literature showed that results from past studies do not provide satisfactory answers (von Braun and Kennedy 1986). As will be shown below, complex farm and rural household production-consumption relationships need to be understood and quantified in order to address these relationships properly for policy conclusions.1 This research has two objectives. First, to assess the effects of increased commer- cialization on production, household real incomes, family food consumption, expendi- tures for nonfood goods and services, and the nutritional status of the population in an 'Studies into the issue by Kennedy and Cogill (1987) in Kenya, von Braun, Puetz, and Webb (1989) in The Gambia, and von Braun, Hotchkiss, and Immink (1989) in Guatemala show generally positive net effects of commercialization for nutrition, or at least no adverse effects. 14 environment under severe population pressure. Second, to describe the process by which household food consumption and nutritional status are affected by increased commer- cialization, identifying the most important elements of the process and estimating how each element is influenced by the change. The analysis should form a basis for evaluating the alternative short- and long-term options for the design of policies and programs to cope with possible income and nutrition problems in the process of transformation from semisubsistence to more commercialized agriculture. This study—undertaken in a very densely populated location in northwestern Rwanda—forms part of a larger research effort at IFPRI on the policy questions posed by the commercialization process in traditional agriculture. While an integrative study of several studies is also currently being undertaken for more generalizable conclusions, each of the studies offers in-depth insight into the causes and consequences of commer- cialization in different socioeconomic and ecological environments. This study is located in an area that is unusual on a number of counts—it has an extremely high population density, an extremely high population growth, and an extremely low degree of urbanization, and is extremely landlocked in the center of Africa. The authors argue that it is particularly revealing to study such an "extreme" case in order to derive insights for future policy directions. In this case, this argument applies especially to the changes induced by the high level of rural population density and its rapid growth. There is an urgent need to improve the understanding of the development process in densely populated areas under increased population pressure, such as the study region, where there is little new agricultural production technology. Also, the interaction between agriculture and the rest of the rural economy especially requires further research. It can be hypothesized that a higher population density makes possible more rapid attainment of gains from specialization and the emergence of a rural service sector. Moreover, high population density may limit the number of households able to survive from agriculture alone, thus forcing some into nonfarm activities to supplement income (Haggblade and Hazell 1987). A central research issue for this study location under population pressure is the sustainability of agricultural production systems along with the efficient use of the resource base. A long-term view of options and alternatives is required for this issue (Tisdell 1988), Much effort has been made to develop sustainable agricultural produc- tion systems in Rwanda (Kotschi, Pfeiffer, and Grosser 1982), yet their attractiveness to small farmers has remained rather limited. To understand and overcome adoption problems, experimental work in the field of sustainable systems should start at the farm level and include assessment of short- versus long-run costs and benefits to farmers and the economy (Adelhelm and Kotschi 1985). While the present study acknowledges this critical requirement, it adds the dimensions of household production-consumption relationships and farm-nonfarm sector links, which are of central importance to agricul- tural technology adoption in subsistence farming. 15 THE FORCES DRIVING THE COMMERCIALIZATION PROCESS Underlying Structures and Policies Rwanda is among the most densely populated countries in the world, and its population density expressed in terms of inhabitants per square kilometer is very high (about 246 persons per square kilometer in 1987). The country is landlocked in the center of Africa, with long road transport routes across neighboring Burundi and Tanzania or Uganda and Kenya to the nearest seaports—Mombasa, Tanga, or Dar es Salaam. The related high transportation costs are almost prohibitive for the integration of bulky commodities into the international exchange economy. For instance, transport costs for cereals from Rwanda to seaports (for example, Mombasa) exceed normal f.o.b. world market prices. Thus, export and import parity prices of cereals in Rwanda establish a range between negative and more than twice the long-term average of the c.i.f. East Africa coast price. Failure of the domestic market to get access to international food markets is a reality for this landlocked country, as crises in neighboring countries occasionally block the trade routes. Consequently, Rwandan food policy emphasizes high national self-sufficiency and increasing regional market integration (Rwanda, Ministry of Planning 1983). The Rwandan economy is only weakly integrated into the international exchange economy. The value of merchandise exports represents only 8 percent of gross domestic product (GDP). For a landlocked country, an extremely low proportion of exports goes to neighboring regions: 81 percent of all exports found their way to the industrial market economies in 1985 (Table 1) (Sinamenye 1986). The economy as a whole appears to be quite subsistence-oriented, which is further suggested by the extremely low degree of urbanization (only 5 percent of the population lives in urban centers). The share of agriculture in GDP, however, was substantially reduced during 1965-85, from 75 percent to 45 percent. This change in the share of agriculture clearly indicates a substantial increase in internal diversification and specialization, including rural nonagricultural activities. In comparison with other African countries, Rwanda has not incurred very high debts and was therefore not much affected by the debt crisis of the 1980s. Debt service in percentage of total exports stood at only 4.3 percent in 1985, a low burden for the economy. In income per capita (gross national product [GNP]), Rwanda ranks among the 25 poorest developing countries. Life expectancy at birth is even below the average found in low-income economies (for selected indicators, see Table 1). So far, Rwanda's integration into the international exchange economy has been largely based on agricultural exports, specifically coffee and, more recently, tea. In 1985, coffee composed 65 percent and tea 15 percent of total exports from Rwanda. Yet export crops and nonfood cash crops cover only a minor share—5.6 percent in 1983—of total arable land in the country. At the aggregate, food availability per capita appears to have 16 Table 1—Some basic features and long-term trends of Rwanda's economy, 1965-85 Item Population (millions) Population growth (percent/year) Urban population (percent) Gross national product (US$/capita, in 1985 dollars) Growth of GNP per capita (percent/year) Share of agriculture in GDP (percent) Exports (merchandise) (percent of GDP) Share of total exports to industrial market economies (percent) Debt service (percent of exports) Share of coffee and tea in total exports (percent) Official development assistance (percent of GNP) Calorie supply (per capita/day) Infant mortality rate (per 1,000 children below 1 year) Life expectancy at birth (years) 1965 3.2 ... 3.0 195 ... 75.0 n.a. 96.0 n.a. n.a. n.a. 1,665 141 n.a. 1985 6.1 ... 5.0 280 ... 45.0 7.6 81.0 4.3 79.8 10.7 1,919 127 48 1965-85 Average ... 3.1 L8 ... ... ... ... ... ... Sources: African Development Bank, Selected Statistics on Regional Member Countries (Abidjan: ADB, 1987); World Bank, World Development Report (Washington, D.C.: World Bank, 1987); Inter- national Monetary Fund, International Financial Statistics (Washington, D.C.: IMF, 1987). Note: n.a. means not available. increased over recent decades.2 This growth was largely a result of the expansion of land area into marginal zones (Delepierre 1985). The population density in terms of persons per hectare of cropland has increased by 30 percent since the early 1960s and stood at more than eight persons per hectare in the mid-1980s (Table 2). With a population growth of 3.3 percent a year in the 1980s, the already very limited land base becomes more and more a constraint to agricultural growth and income generation. The obvious way out of this dilemma appears to be a combination of policies that will lead to reduced population growth, increased land productivity through techno- logical change in agriculture, conversion of land resources, and increased exploitation of the potentials of specialization in the rural economy. Investments in rural infrastructure, education, and technological change in agriculture should be the key inputs. Providing improved understanding of the potentials and the constraints in fostering this process is a main task of this research. The diversity of the country and of the region cannot be captured by a case study for just one location. However, the microlevel processes of commercialization studied at the 2It should be noted that such assessments are made on the basis of rather weak food production and trade statistics (African Development Bank 1987; World Bank 1987; International Monetary Fund 1987). 17 Table 2—Total arable land area, land use density in Rwanda, 1961-83 Item Arable land (1,000 hectares) Export crops and nonfood cash crops (percent of arable land) Population density (persons/hectare of arable land) 1961 521 4.6 6.2 for export crops, and population 1973 610 5.1 6.5 1983 750 5.6 8.1 Sources: Computed from FAO Production Yearbooks and African Development Bank statistics. location may be generalized under certain conditions. These processes and the responses of the households to them are the main interests of this study. The Study Area The research for the present study was conducted in the commune of Giciye, which is situated in the prefecture (district) of Gisenyi in the northwestern part of Rwanda (Figure 1). Some survey work also was done in the neighboring commune, Karago. Thefollowing description focuses on the main study area in Giciye commune? The total area of Giciye commune is estimated at 185 square kilometers, approxi- mately 120 square kilometers of which are suitable for agricultural production, the remainder being part of the former natural forest of Gishwati. Before the communes are described in more detail, some explanation should be given to better understand what a commune is in Rwanda. Since the administrative reform of 1974, a commune can be regarded as the lowest governmental and administrative unit, with a bourgmestre as the official government representative. The bourgmestres are nominated by the country's president. The commune is further subdivided into secteurs, which again are subdivided into cellules. A cellule normally consists of some 100 families. Agroecological Zone Characteristics Following the classification of Rwanda into the agroecological zones given by Gotanegre, Sirven, and Prioul (1974) and Jones and Egli (1984), the study area belongs to the agroecological zone of the Central Zaire-Nile Divide that passes through Rwanda from north to south, with the agroecological zones of the Kivu Lakeshore to the west and the High Plateaus to the east. The main characteristics of the Central Zaire-Nile Divide are relatively high altitudes, normally exceeding 2,000 meters, low average annual tem- perature of approximately 15°C, and abundant precipitation averaging 1,300 millimeters a year (Jones and Egli 1984). In the study area, altitude increases rapidly from 1,500 meters in the east, at the border of the High Plateaus region, to approximately 3,000 meters at the summit of the Gishwati forest, with most of the communal area situated at more than 2,000 meters. 3The following description of the study area is mainly based on the 1984 annual report of Giciye commune. 18 Figure 1—Map of the study area • Commune office — Commune Secteur 1 1 1 1 1 1 1 Forest boundary — • — — Main road Secondary road River Note: The underlined secteur names are those of the four secteurs selected for the study. During 1983-85, annual precipitation averaged 1,236 millimeters, which is rather low because 1984 was a drought year with rainfall of only 948 millimeters or 68 percent of the amount in normal years (1,380 millimeters). However, because of the prevailing tempera- ture and moisture regimes, the risk of crop failure due to changing climatic conditions may be considered generally low. Since this is an equatorial region, the distribution of rainfall follows a bimodal pattern. Four climatic seasons can be distinguished: a long rainy season from mid-February to late May; a long dry season from the end of May to mid-September; a short rainy season from mid-September to early or mid-December; and a short dry season until mid-February. Another important characteristic of this agroecological zone is its mountainous relief dominated by very steep slopes of up to 30°-40°, causing severe surface soil erosion. In fact, soil erosion due to the deforestation and transformation of former forest and pastureland into permanent arable farmland will be one of the most important problems of future development of farming systems in the study area. Soils are generally very poor and acid, mainly derived from metamorphic rocks like granites, schists, and gneisses. The analysis of a soil sample of 90 farm plots undertaken for this study revealed a high degree of soil acidity and a general deficit of phosphorus. In some higher-altitude areas of the commune, soils are found to be aluminum toxic, thus 19 prohibiting normal crop production, with the exception of tea, which can stand this level of aluminum concentration. The output mix of farms in the various agroecological zones shows a high degree of variety—a major problem for setting commodity priorities for agricultural innovations. This is exemplified in Table 3, which shows the crop production per farm in the two prefectures, Ruhengeri and Gisenyi, each broken down by agroecological zone. The differences in importance of bananas, potatoes, and coarse grains for the various agroecological zones are particularly pronounced. Population According to the population census, the total population of Giciye commune was 52,236 at the end of 1984. The average annual increase between 1978 and 1984 was 4.2 percent. The composition by age groups reveals a high population share (46.2 percent) for the group below 18 years of age and only 5.2 percent for that above 60 years of age. In 1984 the composition by ethnic groups in the commune was Hutu (98.8 percent) and Tutsi (1.0 percent), the remainder being Twa, the pygmy aborigines. Average popu- lation density was 282 persons per square kilometer in 1984, or 435 persons when only the area suitable for agricultural production is taken into account. However, population density varies considerably among the different administrative sectors of the commune, from approximately 270 persons per square kilometer in some higher-altitude secteurs in the west to more than 900 persons per square kilometer on the volcanic soils at the border of the High Plateaus in the east. Markets and Social Infrastructure There is only one main road (route principale) that crosses the communal area of Giciye (Figure 1). This road descends for a distance of 24 kilometers from Giciye to Mukamira, an important marketplace situated at the paved road that links the provincial capitals of Gisenyi and Ruhengeri to the capital, Kigali. The area is equipped with two health centers in the Karago commune, and one in Giciye attached to a hospital. Another hospital is located in the neighboring Kabaya commune. In Giciye, there are 19 primary schools spread throughout the commune area. In Karago, there are 14 primary schools and 2 secondary schools. In both communes, there is a centre communal de ddveloppement et deformation permanente (CCDFP), with branch centers for some secteurs. The main activities of the CCDFP are in the fields of literacy campaigns, improvement of agricultural practices, hygiene, and family planning. The centers work in close collaboration with the local cooperatives. Giciye commune has three markets of different size and importance: a market takes place twice a week in Gasiza, Jomba, and Vunga (Figure 1). Kabaya is another impor- tant marketplace situated at the border of Giciye and Gaseke communes in the south of the study area. Using the typology of Rwandan agricultural markets developed by the national agricultural research organization (I'Institut des Sciences Agronomiques du Rwanda [ISAR]), Jomba and Gasiza may be classified as marches paysans—that is, relatively small rural markets of only local importance—where farmers themselves constitute the majority of sellers and buyers. The number of merchants and middlemen organizing the 20 £ •a 1 f V -a I -§ es 1 8 't I a f I a IS r» ô >-i ^ o v ^ l o r - * oi oo cs o -H' Vt ob F- ts vo <7> © vo © SO x V S O \ IS rt > 2f? P5f~ fN fO VO O\ CO OS i i j d Ml I i -3 I r- r~ co r- inmf- 00 00 "I 00 OO CO in O' O' -J O O >-! O i o O t - en NO O • »n io rn ci