AGRICULTURAL TECHNOLOGY, HEALTH AND NUTRITION

LINKAGES: SOME RECENT EVIDENCE FROM SUB-SAHARAN AFRICA

Tesfaye Teklu

Abstract: Investment in agricultural technology is crucial for countries in Sub-Saharan Africa in order for them to meet their growing demand for food at low cost. Current evidence provides support for the view that such investment is, indeed, profitable and does contribute to improved productivity. However, there is still a lack of empirical evidence derived from rigorously measuring the impact of technological change on household welfare, based on consumption, health, and nutrition outcomes. The few recent studies show that technological change improves income and food consumption. However, the impact on nutrition outcomes seems weak. This phenomenon is attributed to the weak relationship between income and food consumption, as well as between income and health expenditures. Given the strong link between morbidity and child nutrition in Africa, the weak link between income and health expenditures is a key limiting factor. As the review of the African case studies in this paper suggests, in order for technological change to have an appreciable effect, nutrition outcomes, investments in agricultural technology have to be accompanied by investments in health and environmental sanitation, better nutrition education, and, possibly policies that lower the trade off between employment and child care, especially for the primary child carer in technology-adopting households. Policymakers, however, need to be guided by more inter-disciplinary research to promote a greater understanding of how the links between agricultural technology and nutritional outcomes can be strengthened.

1. Introduction

Poverty in Sub-Saharan Africa is markedly manifested in a low level of food consumption, despite a large presence of food in rural household budgets. Improving food consumption goes a long way towards reducing food security risk. It is only when a country progresses on its growth path that the risk of food insecurity is reduced, since the country then has the ability to produce sufficient food or generate resources to import food to meet food requirements. If such growth enables a large segment of the population to participate in the growth process, the income gains are widely shared to enhance the food entitlement of the population. The gains are even stronger if markets are developed and operated at low transaction costs.

For most African countries, improving food security through agricultural growth is pivotal. Its most direct contribution is through increasing production of food staples largely for smallholder subsistence farmers whose food security is essentially a food production problem. For those who depend on the market place, agriculture's contribution to food security occurs directly through its growth effect on employment and income of the non-food producing farm population, and indirectly through stimulating growth in non-farm sectors and in reducing food prices.

Sub-Saharan Africa faces the prospect of a high rate of food consumption, especially since most of the countries have yet to achieve lower rates of population growth. In addition, the anticipated improvement in economic growth is likely to increase food demand because of the evidence of high-income elasticities of demand for food, especially for low-income countries. Given the limited prospect of meeting the future food gap through imports, the region has to accelerate growth in agriculture, especially food production.

With economic and environmental costs associated with land expansion increasing at the margin, rapid growth in technological change is critical to foster future agricultural growth. This calls for improved adoption and efficient utilization of existing technologies, as well as continuous generation of new and appropriate frontier technologies that enhance the productivity of the most limiting factors. Because of the heterogeneity of African production environments, a variety of new technologies have to be developed to correct the deficiency in soil nutrients and moisture, reduce rainfall-linked variability in production, and overcome seasonal shortage of labor.

Improvement in income through technology-led agricultural growth is necessary for better health and nutrition outcomes. In general, an increase in the level of income is associated with a decline in the incidence of child malnutrition (UN ACC/SCN 1992). However, this is not a sufficient condition for nutrition security. For agricultural growth to have a substantial impact on nutrition security, the increase in income-induced food intake has to be coupled with nonfood inputs such as health facilities and the provision of clean water sources (Kennedy and Bouis 1993).

Countries that have pursued policies to improve both food security and health conditions have been successful in the 1980s (e.g., Indonesia, Thailand, Chile, State of Tamil Nadu in India, and Zimbabwe). Some African countries have successfully increased food production through technological change of smallholder agriculture (e.g., Kenya, Malawi, and Zambia). However, the evidence does not show a parallel improvement in health and nutritional status.

The main thrust of this paper is to identify the factors that are responsible for the weak relationship between agricultural technological change and nutrition. The paper briefly discusses the basic conceptual linkages between technological change and nutritional outcome in Section 2. It then synthesizes the recent evidence based on selected Sub-Saharan African countries with which the author is familiar. The final section states the conclusions and implications for policy and research.

2. Agricultural Technology, Health, And Nutrition: A Conceptual Framework

The adoption of improved technology (for example, applications of improved seed-fertilizer, animal traction, irrigation technology) sets in motion adjustments in household resources, especially labor and land, which are the primary factors of production and sources of income in poor agricultural households. The main paths of adjustment are presented in Figure 1.

Figure 1. Agricultural technology, health and nutritional linkages: A conceptual framework

There are at least three direct pathways though which nutritional outcomes can be affected: (1) income-food consumption-nutrition path; (2) time use-child care-nutrition path; and (3) income-health expenditure-morbidity-nutrition path.

The adoption of improved technology may trigger changes in crop mix, level and composition of employment, and productivity of labor. These changes, in turn, may translate into changes in the level and composition of income (link 1). The net income gain from technological change depends on net increase in labor productivity and whether there is forgone income due to switching of labor from other income generating activities and/or changing of crop composition. This forgone income is likely to be high in an environment where there is scarcity of agricultural labor, especially in the peak farm season, and shortage of land suitable for crop production. Since such forgone income is unlikely to be zero, the net income gain due to technological change will be less than the change in gross income that is uncorrected for forgone income.

A change in income level (or composition) induces a reallocation of household budgets (link 2). Depending on the household food expenditure elasticity, this might translate into a positive change in food expenditure (link 3). How much of the change in food expenditure actually results in an increase in the level of consumption (as measured by level of calorie consumption) depends on other considerations such as the desire for more variety in the diet, or personal taste (Lancaster 1966; Ladd and Suvannunt 1976; LaFrance 1983; Bouis 1995). Typically, consumers shift to high cost calories as their income level increases.

These chains of causal relationships (technology-income-food expenditure-food consumption) are usually expected to have a positive impact on nutritional status (link 5). Even households which do not adopt the new technology may benefit through either or both of the following ways. First, via a labor market effect, to the extent that such a technological change is accompanied by an outward shift in the market labor demand curve and increases in both employment and wage rates; second, via a food market, to the extent that the increase in production translates into a decline in food prices and the beneficiaries are net food buyers.

The second main pathway starts through the impact of technological change on household labor supply (link 6). Labor supply responses to technology-induced increased demand for labor may differ: (1) additional members of a household may enter into the work force; (2) those who are already in the work force may work longer hours and/or change the allocation of time across activities; and (3) family labor supply may remain unchanged but additional labor is drawn from non-family labor such as hired labor.

What happens to women's time, in particular, especially the time allocated to child care, depends on the relative valuation of time between income generation and child care (link 7). The competition of time between earning income and child care may become sharp for households which face an increased demand for their labor, but lack sufficient labor of their own or resources to hire labor. Adoption of a new type of technology that uses labor intensively could adversely affect health and nutrition outcomes in such households (link 8). That is, the potential gains from additional work and earning due to technological change may not be adequately realized as some household members experience increased energy expenditure which is not fully compensated by increase in consumption. In addition, increased time demands from new technology adoption might also lead to insufficient time devoted to child care.

The introduction of new technology can also impact on the incidence and extent of disease in various ways. First, net household income from technological change may be allocated to improve environmental sanitation (e.g., investment in better housing and water supply) and obtaining health services (link 9). Typically, the marginal budget share on these types of services tends to increase when incomes increase. Second, an increase in food consumption that is associated with an improvement in income may contribute to an increase in resistance to disease (link 10). Technological change may also positively or negatively affect the external health and sanitary environment (link 11). The latter has the potential to either neutralize or complement the private demand for health services (link 9) to improve health status. These intermediary outcomes, such as increased effective demand for health services, resistance to disease, and change in the external health and sanitary environment, have a direct bearing on the technology-income-morbidity relationship.

The total effect of technological change on nutrition thus depends on what happens to these various links which operate through these three main pathways: (1) income-consumption-nutrition; (2) household time use-child care-nutrition; and (3) income-morbidity-nutrition. The following section gives a brief account of the existing empirical evidence from some Sub-Saharan African countries.

3. Agricultural Technology, Health And Nutrition: Some Evidence

The evidence in this section is drawn mainly from the studies in the Gambia (von Braun, Puetz and Webb 1989), Kenya (Kennedy 1989), Rwanda (von Braun, de Haen, and Blanken 1991), and Zambia (Kumar 1994; Holleman and Pinstrup-Andersen 1993). The studies (hereafter referred to "the African case studies") from the Gambia and Zambia are directly linked to the introduction of technological change in smallholder agriculture. The other case studies focus on the introduction of new cash crops. Since farmers who adopt new technology have a strong propensity to become commercially oriented, the two sources of growth are not quite separable in the post-production phase. This paper does not attempt to distinguish the effects of the two processes on food and nutrition security.

3.1 Impact on Employment, Income, Food Consumption and Nutrition

Technological change affects employment levels and labor productivity. These labor market outcomes influence production and, consequently, income. If income increases, then the additional income is distributed into expenditures and savings/investment. The increase in food expenditure is assumed to induce additional food consumption and to lead to an improvement in nutritional status.

3.1.1 Effect on Employment (link 1)

The effect on employment is crop- and technology-specific. For example, the land augmenting types of technology, which were adopted in the Gambia (irrigated technology of rice farms) and Zambia (improved seed, fertilizer and oxen-technology), led to an increase in demand for farm labor (von Braun, Puetz and Webb 1989; Kumar 1994). In the Gambia, irrigated rice lands allowed an increase in the planting area, which increased, in turn, women's workload in weeding and transplanting, even though they could not own land in the scheme. On the other hand, farm households that extensively adopted hybrid maize in Eastern Zambia used 53 percent more male labor, 20 percent more female labor and 80 percent more child labor as compared with households whose adoption of hybrid maize was low.

The increase in demand for farm labor was largely met through an increase in family labor. All categories of family labor (female adult, male adult, child labor) experienced an increase in level of employment, but the relative increase was much higher for male adult labor. For example, the share of male labor was relatively large in the case of hybrid maize as compared to male labor in local maize production (Kumar 1994; Celis and Holleman 1991). A household survey in the fertile Ufipa plateau of Rukwa region in Tanzania also showed that the increased profitability of crop production changed the gender specific division of crop activities, especially the increased participation of men (Beier et al. 1990)

Such reallocation of labor may be due to a greater demand by these types of technologies for particular activities, which are usually gender specific. For example, the introduction of mechanical technology for rice production in Sierra Leone increased the amount of time required for female labor to meet the increased demand for planting and harvesting, two activities usually carried out by female labor (Spencer and Byerlee 1976). However, the evidence from Tanzania shows that adult men are likely to replace females when they find farm work profitable. Not only were men more drawn to undertake the ox-drawn ploughs in Rukwa region, but they were increasingly engaged in weeding and harvesting, which were typically women's activities. Another example of men taking over women's crops when these become more profitable occurred in Nigeria, where a project promoting the marketing of rice, cassava, and melons was introduced (Burfisher and Horenstein 1985; Rogers 1983). .

On the other hand, there are studies that indicate an increasing share of women's time commitment after technological change. For example, a project that sought to increase productivity of nine major crops in Nigeria increased women's annual labor requirements by 17 percent while men's labor requirements increased by 6 percent. Large-scale mechanized cotton production in Burkina Faso doubled women's work hours in absolute terms (Jiggins 1986).

Additional employment also accrues to non-family labor, both exchange and hired labor, but less in proportion as compared to family labor. For example, the study in Zambia shows that households using a high level of technology employ 37 times more hired labor and 11 times more exchange labor than households using low level technology (Celis and Holleman 1991). The evidence from the Gambia also shows an increased demand for hired labor on irrigated rice farms (von Braun, Puetz, and Webb 1989).

In short, the new agricultural technologies adopted in these African case countries contributed to an increase in demand for labor. The increase was relatively larger for family labor. But the evidence is not conclusive on allocation of the additional work within households. There is also little information in these studies on what happened to rural wages due to a technology-induced shift in labor demand..

3.1.2 Effect on Production (link 1)

Farmers tend to reallocate land to either crops using new technology or new cash crops. Such reallocation of land may not necessarily be accompanied by a decline in the proportion of land allocated to staple crops. For example, the difference in percentages of land allocated to staple food crops between new technology adopting and non-adopting farmers was only 3.2 points in the Gambia survey. The sugar cane producers in Kenya even increased the area allocated to grow maize through bringing fallow land into crop production.

Where there is a decline in area allocated to food crop production, it does not also necessarily translate into a decline in food production. Where yield level is high, farmers can meet their subsistence needs without committing a large area of land. The adoption of yield-increasing technology can reduce the land area needed for subsistence production.

The extent to which farmers choose to switch to non-staple commercial crops is not governed only by profitability considerations, but also by how much farmers weigh food security risk (due to uncertainties arising from production, off-farm employment, and the market). The evidence from the Africa case studies shows that farmers deliberately maintain food crops along with cash crops where a premium on food security is high. Where food security risks are low, farmers tend to specialize. Thus, the emphasis on a "food first' strategy becomes less of a priority in areas where there is a great deal of market integration.

3.1.3 Effect on Income (link 1)

In general, farmers adopting new technology experience an increase in income level due to an increase in factor productivity and a decline in unit costs of production. The Africa case studies (Gambia, Rwanda, and Kenya) show increases in income ranging from 17 percent to 25 percent (von Braun and Kennedy 1994). A study based on survey data from Eastern Zambia indicates that the net return per area from hybrid maize is five times greater than from local maize (Jha, Hojjati and Vosti 1991).

Such an increase in income may be accompanied by a change in relative distribution of income. This may be observed, for example, in the distribution of income between adopting and non-adopting areas or between adopting and non-adopting households. In addition, the adoption of new technology may induce a shift in income control and distribution of income even within households. For example, the emerging evidence indicates that men tend to assume greater income control as farming becomes more profitable. The Gambia study, for example, shows that, despite the project's policy of targeting irrigated technology to women-controlled rice farms, the control of income shifted to men. To the extent that such a shift in income control from women to men adversely affects distribution of food allocation within households, it may also adversely affect consumption and nutritional outcomes..

3.1.4 Effect on Food Expenditure (through links 2 and 3)

In most of the African case studies, the level of food expenditure increased with income. The recent survey of food demand patterns in Sub-Saharan Africa shows that, on average, food expenditure elasticities range between 0.4 and 0.6 (Teklu 1996). That is, a 10 percent increase in household expenditure leads to a 4 to 6 percent increase in food expenditure. However, the increase is more pronounced in low-income settings (on average, food expenditure elasticities range between 0.80 to 1.0). For example, the proportion of the incremental income spent on food is high in the Gambia and Rwanda (nearly 1.0).

Typically, as income increases, households purchase more expensive calories. This is evident from the different case studies which show lower calorie elasticities than expenditure elasticities. For example, Holleman and Pinstrup-Andersen (1993) estimated from the Zambia survey that the average "quality" elasticity (the difference between expenditure and calorie elasticities) is 0.16. The differences are even larger in the other case studies; for example, 0.46 for Gambia, and 0.50 for Rwanda (von Braun and Kennedy 1994). These studies confirm that "quality" elasticities are positively related with income. That is, an increase in food expenditure is not fully transmitted as an increase in food consumption.

The form (cash or in-kind) in which income accrues to the household also has an impact on food expenditure patterns. But the evidence so far is mixed (von Braun and Kennedy 1994). In particular, the cash content of income had little effect on the budget share of food in the Gambia and Kenya, but it had a significant impact on food shares in Rwanda and Zambia.

Similarly, the control of income matters. The evidence from the Gambia, Rwanda and Kenya suggests that control of income by women has a greater impact on the share of income allocated to food. Other studies also confirm that women's income is more likely to be spent on food (Haddad and Hoddinott 1995; Haddad and Pena 1994; Dwyer and Bruce 1988; Fapohunda 1988; and Guyer 1980). In addition, it has also been shown that compared to men's income, there is a stronger association between women's incomes and improvements in children's health and nutritional status (Tripp 1982; Haddad and Hoddinott 1995).¹ To the extent that the introduction of new technology tends to be associated with a decline in women's control in household income, it can have a depressing effect on food expenditures.

There are two opposing tendencies working on the impact of technological change on food expenditure. An increase in the level of income contributes to an increase in food expenditure, but the impact is diminished if it is accompanied by a shift in control of income from women to men. However, the evidence so far shows that the direct effect of increasing the level of income tends to have a much stronger effect on food demand than these two other factors (form and control of income).

3.1.5 Effect on Calorie Consumption (link 4)

Similar patterns emerge with respect to the conversion of incremental food expenditure into consumption. Household calorie consumption increases with an increase in income, but the increase is lower than the increase in food expenditure. The leakage between food expenditure and calorie consumption is often associated with consumer preference for food characteristics, which often tend to be more expensive but not necessarily calorie-intensive.

In addition, the effect of increasing income in the form of cash and the loss of income control by women may also depress the potential increase in calorie consumption. However, the current evidence indicates the direct effect of income on calorie consumption more than compensates for these depressing effects. The existing evidence shows that a plausible interval for estimates of calorie elasticity with respect to income ranges between 0.5 and 0.1, depending on what level of income the elasticity is evaluated since the elasticity tends to fall with increase in income level.

The survey method used to generate calorie-income elasticities also matters since, as shown in Bouis, Haddad and Kennedy (1992), estimated calorie-income elasticities and demand patterns for specific foods in Kenya and the Philippines displayed wide differences, even though the same household samples were used. In particular, in both areas, food expenditure data resulted in upwardly biased figures, while 24-hour recall data yielded lower estimates.

3.1.6 Effect on Child Nutrition (link 5)

The links connecting technology-income-calorie consumption (through links 1 to 5) have direct bearing on child nutritional outcome. In general, the nutritional status of children improves with increased income, food expenditure and calorie consumption.

As a whole, the existing body of evidence supports the hypothesis that technological change has a significant impact through the income-food security pathway. However, there is less unanimity on how strong the link is. For example, household energy consumption is a significant factor in determining weight-for-age in the Gambia and Rwanda. But the evidence from Kenya does not show such a strong relationship, particularly in the short term, especially in cases where both sanitation and the health environment are poor. .

3.2 Impact on Non-food Expenditure, Morbidity and Nutrition

Studies on determinants of child nutritional status indicate that child morbidity, rather than income, has a strong influence on nutrition outcome (Kennedy and Bouis 1993; and Alderman and Garcia 1993). On the other hand, the evidence from the Africa case studies on the technology-income-morbidity relationship suggest a weak income-morbidity relationship (link 9). That is, an increase in income does not translate into an appreciable reduction in the incidence and extent of morbidity.

One plausible explanation often advanced to explain such a weak income-morbidity relationship is the low share of incremental income that is often allocated to obtaining health services and improving the household environment (for example, through better sanitation and having a clean water supply). Typically, direct household expenditures on health care are between 3 to 5 percent of total household expenditures (Pinstrup-Andersen 1987). The Africa case studies show that the average budget shares on these expenditure components do not exceed 10 percent of household income. Thus, private expenditures on health and sanitation do not seem to contribute significantly to improvement in health and nutritional outcome. One plausible explanation is that households may need health and sanitation improvements in their external environment (link 11) before internal improvements can actually work.

In African settings where the external health and sanitary environment is poor, such low private effort may be insufficient to offset the adverse effect of technological change on health and nutritional outcomes. It may even be worse in situations where the introduction of new technology exacerbates the spread and prevalence of disease.

3.3 Impact on Women's Time, Child Care and Nutrition

The common concern is that adoption of technology leads to an increase in demand for women's labor and, hence, to a reallocation of women's time away from child nurturing activities. Earlier time allocation studies from a few countries regions such as the Gambia, Northern Nigeria, and Tanzania indicate that women's time constraints during the peak agricultural season may interfere with breastfeeding (Carloni 1984). In a study that considered agricultural work patterns between women with and without children in Kenya, findings show that women with infants spend less time on agricultural production and self-care than women without infants. However, during the peak agricultural season, although, women devoted less time to child care they arranged for their older daughters to care for their younger siblings (Paolisso, Baksh and Thomas 1989).

The above studies do not focus on technology adopting households. The Africa case studies tested whether adoption of new technology would adversely affect time for child care (as measured by age of child weaning and feeding practices) and child nutritional status (von Braun and Kennedy 1994). The tests do not find significant differences between technology adopting and non-adopting households in terms of child feeding practices since there is no evidence of early child weaning or early introduction of solid foods. Hence, the few existing studies that do examine the effects of new technology adoption on child care do not find any difference between technology adopting and non-adopting households.

In addition, a lot has been written on the increased demands on women's time by improved technology adoption, but very few studies have looked into its associated impact on women's own nutritional status. Additional work due to technological change may lead to an increased level of physical activity (or energy requirement) that may, if it remains uncompensated, adversely affect women's weight. In the case of pregnant women, this may contribute to low birth weight. However, the existing evidence does not conclusively show adverse adult nutrition as measured by BMI or increased incidence of illness due to technology adoption or commercialization of agriculture. This is the case even in the irrigated environment of Gambia, where the length of time of illness between participants (15.2%) and non-participants (15.9%) was not significantly different.

4. Conclusions And Implications For Policy And Research

The reviewed African case studies have made a significant contribution to research on agriculture technology beyond the usual convention of focusing only on the evaluation of a new technology's profitability. While such a measure is essential, it is nevertheless insufficient for understanding the impact of technology on household welfare, as measured by consumption, health and nutrition outcomes. What is clear from the evidence presented so far is that technology adoption can affect household welfare through at least three different pathways (income-food consumption, income-nonfood expenditures-morbidity, and time allocation patterns-child care-nutrition). Thus, the design of appropriate technology necessitates the evaluation of how these pathways can independently and jointly impact on nutritional outcomes. The decision to intervene through a particular pathway needs to consider both independent and joint effects through the various links.

The current evidence on Sub-Saharan Africa suggests that the effect of technological change on child nutritional status is positive but weak, especially in the short-run (Table 1). The main explanations are twofold. First, the current evidence does not show a strong income-calorie relationship. This weak relationship is attributed to greater propensity of households to purchase more expensive sources of calories as their income level increases. Since there is a stronger association between women's incomes and improvements in children's health and nutritional status, a shift in income control away from women's control may also depress the incremental effect of income on food consumption.

Table 1. Prevalence of stunting, wasting, and malnutrition among preschool children, participant and nonparticipant households^a

SOURCE: J. Von Braun and E. Kennedy, Agricultural Commercialization, Economic Development and Nutrition (Baltimore, Maryland: Johns Hopkins University Press, 1994).

^aParticipating households are either new technology adopters or cash crop producers.

^bLess than 90 percent below height-for-age standard.

^cLess than 80 percent below weight-for-age standard.

^dLess than 80 percent below weight-for-height standard.

Second, the income effect of technological change on child morbidity is weak. The failure of increased incomes to have a significant impact on morbidity (which has been established as a major determinant of child nutrition) can be attributed to the small share of additional income allocated by households to health services and improvements in sanitation. However, it must also be pointed out that the lack of supply of health services, rather than demand, might be the most constraining factor for the utilization of health services.

The current evidence on the effects of new technology on child nutritional status via women's time use is not well established. In addition, very few studies have looked into the impact of technological change on women's own nutritional status. The prime emphasis of future research has to focus on the impact of technology on household time allocation patterns and its subsequent effect on the nutritional status of both women and children. This seems to be the link which has been least explored by existing studies.

However, further research work is needed even in the technology-income-consumption link in order to establish conclusive evidence, especially in the context of low-income African countries. The implication that technological change has a low impact on food consumption and, consequently, a weak linkage effect on child growth has to be confirmed through more empirical evidence. There are also related questions that need to be addressed. First, how much does the shift in control of income amplify or depress the effect of income on consumption? Second, how much can the leakage between expenditure and consumption be ascribed to demand for micronutrients? Finally, what is the feedback effect of improved consumption and nutrition on labor productivity and wages?

References

Alderman, H., and M. Garcia. 1993. Poverty, household food security, and nutrition in rural Pakistan. Research Report 96. Washington, D.C.: International Food Policy Research Institute.

Beier, M., C. M. Falkenberg, N. Mjema, and A. Zewdie. 1990. The Southern highlands development and the food security needs of Tanzania: The case of Rukwa Region. A study paper prepared for GTZ, Division 426.

Bouis, H. 1995. A food demand system based on demand characteristics: If there is curvature in the Slutsky matrix, what do the curves look like and why? Journal of Development Economics.

Bouis, H., L. Haddad, and E. Kennedy. 1992. Does it matter how we survey demand for food? Evidence from Kenya and the Philippines. Food Policy 17, no. 5: 349-360.

Braun, J. von, D. Puetz, and P. Webb. 1989. Irrigation technology and commercialization of rice in the Gambia: Effects on income and nutrition. Washington, D.C.: International Food Policy Research Institute.

Braun, J. von, H. de Haen, and J. Blanken. 1991. Commercialization of agriculture under population pressure: Effects on production, consumption, and nutrition in Rwanda. Washington, D.C.: International Food Policy Research Institute.

Braun, J. von, and E. Kennedy, eds. 1994. Agricultural commercialization, economic development, and nutrition. Baltimore, Maryland: Johns Hopkins University Press.

Burfisher, M. E., and N. R. Horenstein. 1985. Sex roles in the Nigerian TIV farm household. West Hartford, Connecticut: Kumarian Press.

Carloni, A. 1984. The impact of maternal employment and income on the nutritional status of children in rural areas in developing countries. Geneva: United Nations Administrative Committee on Coordination - Subcommittee on Nutrition.

Celis, R., and C. Holleman. 1991. The effects of adopting technology in the household use of labor. In Adopting improved farm technology: A study of smallholder farmers in Eastern Province, Zambia, edited by R. Celis, J. Milimo and S. Wanmali. Washington, D.C.: International Food Policy Research Institute.

Dwyer, D., and J. Bruce, eds. 1988. A home divided. Stanford, California: Stanford University Press.

Fapohunda, E. 1988. The non-pooling household: A challenge to theory. In A home divided, edited by D. Dwyer and J. Bruce. Stanford, California: Stanford University Press.

Guyer, J. 1980. Household budgets and women's incomes. Working Paper 28. Boston, Mass.: African Studies Center, Boston University.

Haddad, L., and J. Hoddinott. 1995. Does female income share influence household expenditure patterns? Evidence from the Côte d'Ivoire. Oxford Bulletin of Economics and Statistics 57, no. 1.

Haddad, L., and C. Pena. 1994. Differential effects of male and female incomes on household food security: A review. Report to the Statistical Office of the United Nations. Washington, D.C.: International Food Policy Research Institute.

Holleman, C. F., and P. Pinstrup-Andersen. 1993. The impact of commercialization on household food security: A case study of small-scale farmers in Zambia. Washington, D. C.: International Food Policy Research Institute.

Jha, D., B. Hojjati, and S. Vosti. 1991. The use of improved agricultural technology in Eastern Province. In Adopting improved farm technology: A study of smallholder farmers in Eastern Province, Zambia, edited by R. Celis, J. Milimo and S. Wanmali. Washington, D.C.: International Food Policy Research Institute.

Jiggins, J. 1986. Gender-related impacts and the work of the international agricultural centers. Consultative Group on International Agricultural Research (CGIAR) Study Working Paper No. 17. Washington, D.C.: World Bank.

Kennedy, E. 1989. The effects of sugarcane production on food security, health, and nutrition in Kenya: A longitudinal analysis. Research Report 78. Washington, D.C.: International Food Policy Research Institute.

Kennedy, E., and R. Oniang'o. 1993. Determinants of household and preschooler Vitamin A consumption in Southwestern Kenya. Journal of Nutrition 123.

Kennedy, E., and H. Bouis. 1993. Linkages between agriculture and nutrition: Implications for policy and research. Occasional Paper. Washington, D.C.: International Food Policy Research Institute.

Kumar, S. K. 1994. Adoption of hybrid maize in Zambia: Effects on gender roles, food consumption and nutrition. Research Report No. 100. Washington, D.C.: International Food Policy Research Institute.

Ladd, G., and V. Suvannunt. 1976. A model of consumer goods characteristics. American Journal of Agricultural Economics 58, no. 3.

LaFrance, J. T. 1983. The economics of nutrient content and consumer demand for food. Ph.D. thesis, University of California, Berkeley, California.

Lancaster, K. J. 1966. A new approach to consumer theory. Journal of Political Economy, 74.

Paolisso, M., M. Baksh, and J. C. Thomas. 1989. Women's agricultural work, child care and infant diarrhea in rural Kenya. In Women, work, and child welfare in the Third World, edited by J. Leslie and M. Paolisso. Boulder, Colorado: Westview Press.

Pinstrup-Andersen, Per. 1987. Macroeconomic adjustment and human nutrition: Available evidence and research needs. Food and Nutrition Bulletin 9, no. 1.

Rogers, B. 1983. The internal dynamics of households: A critical factor in development policy. Medford, Massachusetts: Tufts University School of Nutrition.

Spencer, D., and D. Byerlee. 1976. Technical change, labor use, and small farmer development: Evidence from Sierra Leone. American Journal of Agricultural Economics 58, no. 5.

Teklu, T. 1996. Food demand studies in Sub-Saharan Africa: A survey of empirical evidence. Food Policy 21, no. 6.

Tripp, R. 1982. Farmers and traders: Some economic determinants of nutritional status in northern Ghana. Food and Nutrition 8, no. 1.

United Nations Administrative Committee on Coordination. Subcommittee on Nutrition (UN ACC/SCN). 1992. Second report on the world nutrition situation. Vol. 1, Global and regional results. Geneva.

¹ One explanation offered is that, in Africa, women are mainly responsible for the provision of food in the household while men have other expenditure responsibilities like housing and education. Another explanation suggests that male and female income flows are different, with the latter coming in more frequent and smaller amounts thereby increasing the propensity for it to be spent on household daily subsistence needs (Tripp 1982; Kennedy and Oniang'o 1993).

* Department of Economics, Western Michigan University, Kalamazoo, MI 49008.