Contents - Previous - Next
This is the old United Nations University website. Visit the new site at http://unu.edu
This article presents a theoretical framework for answering the question, Does women's income-producing work result in an improvement or a deterioration in the health and nutritional status of their children? The author maintains that the literature has not provided clear evidence that the children of working mothers have lower nutritional status than those of mothers who do not work outside the home. She notes that previous studies have not controlled either for the family's income level or for other important variables, such as the presence of an employed male and land tenure status.
Bennett argues that one of the key steps in understanding women's role in the determination of child health and nutritional status "may be a careful investigation of the process of resource allocation within the household. " She opposes the use of a single joint-utility function as an adequate representation of household decision-making dynamics. Her recommendations include conducting investigations from an interdisciplinary approach, including economics, health, nutrition, and anthropology, and she presents a framework for these investigations.
There is a deep and legitimate uncertainty about the relationship between women's role as mothers and their role as economic providers. In its most basic form this uncertainty reduces to the question of whether women's income-producing work-which is increasingly recognized as crucial to the survival of poor families-results in an improvement or a deterioration in the health and nutritional status of their children. Since the question has not been adequately answered, and perhaps cannot be answered in this simplistic form, the usual response has been for agencies to focus on one or the other of these two roles or to have separate divisions for each. One section is charged with health and nutrition or family planning, which tends to focus on women as mothers, and the other is charged with agricultural development, employment, or, more frequently, "women in development," which focuses on women as producers. At the policy level, it is clearly recognized that both roles are important, but, at the operational level, activities carried out by specialized agencies or divisions tend to omit one aspect or the other. As Nieves  has pointed out, this is often not very helpful to the women themselves, who have no choice about whether to work or not. For women in low-income families the issue is, precisely, how to balance work with their child-care responsibilities.
In recent years a number of authors have reviewed the available literature seeking to clarify the issues and highlight what is known about the relationships between women's work and their child-care responsibilities [1-6]. Many of the insights from this literature have been used in building the conceptual framework put forward in the final section. What is clear from all the reviews is that there is a great deal of partial and conflicting evidence. For example, numerous studies can be cited that show that the children of working mothers have lower nutritional status than those of mothers who do not work outside the home [7-10]. However, as the reviews point out, these studies have not controlled either for the family's income level or for other important variables such as the presence of an employed male and land-tenure status  Hence, the poor nutrition observed among the children of working mothers in many of these studies may well be due to the conditions of poverty that drove the mothers to work in the first place and the low wages such women are able to earn, rather than to the fact of their working per se.
Restatement of the problem: Women as "maximizing" individuals within the household
Several reviews [2, 4] have suggested that a broader framework should be used for beginning the search for the determinants of child health and nutrition. An examination of the trade-off between a woman's income-production time and her child-care time is inadequate. What is needed is an investigation of how this trade-off is itself conditioned by the economic and socio-cultural environment. It is equally important to set this trade-off within the household where it takes place.' It is critical to determine both what domestic services and income other members of the family and kin group are contributing and whether or not these contributions influence household decision-making patterns in ways that affect child health and nutritional status. There are a growing number of detailed household studies that include data on the time allocation and income contribution of individual family members, but most of these studies lack either the complementary data on child health and nutrition [ 11, 12] or sufficient data on decision-making or income control [13-18].
One of the key steps in understanding women's role in the determination of child health and nutritional status may be a careful investigation of the process of resource allocation within the household. We need to know gender-specific priorities for expenditure and the degree to which men and women control or influence various areas of decision-making. It is a difficult task, amenable to quantitative measurement and statistical analysis only to a limited degree. Nevertheless, the attempt is necessary if we are to move beyond the commonly accepted premise of the New Household Economics that a single joint-utility function can adequately represent the dynamics of household decision-making. To clarify what is at issue here, I will define what is meant by a joint-utility function and discuss its limitations in analyses of the household.
The joint-utility concept is based on a number of inaccurate and ethnocentric assumptions about the nature of the family. As both Rosaldo  and Yanagesako  have noted, these assumptions are consistent with a tendency in Western thought to idealize or sentimentalize the family as an "unchanging, nurturant, and altruistic core," which is somehow above the more instrumental and political modes of the public sphere. One of these ethnocentric assumptions is that all household members share the same priorities for the use of household time and resources. In the words of its own proponents, the assumption of a joint-utility function "means simply that all household members agree to certain management rules regarding the distribution of income within the house hold and the allocation of household member time' [21, 22].
A number of scholars from various disciplines [3, 23-26] have called this concept into question, citing data from Africa and elsewhere. These data show, first, that the household income is not always pooled and, second, that men and women often have separate, culturally designated obligations to meet different sets of needs within and beyond the conjugal family. In particular, it has been observed that women are often responsible for providing all or part of the resources needed to support their children [23, 24, 26, 27]. These studies and my own work in Nepal indicate that what the household decides to do with its resources is the outcome not of spontaneous utopian agreement but rather of a loose set of entitlements subject to serious bargaining (whether covert or overt) on the part of individual members [28, 29].
This brings us to the two other assumptions implicit in the concept of the joint household utility function: that all household members have equal bargaining power to enforce their own definition of utility, and that all members benefit equally from the way resources are actually allocated. According to this view, any observed inequality in the distribution of household resources is interpreted as the most efficient reaction to the prevailing wage/price situation. i.e. as part of the household unit's efforts to maximize income.
However, Folbre's reanalysis of the time-allocation and nutrition data from the Laguna, Philippines' study presents an alternative and, I believe, much more compelling explanation for reported inequalities in household-distribution patterns . Among the more obvious disparities evident from the Laguna data are two facts: men have significantly more leisure time than women; and they consume 101% of their recommended daily allowance (RDA) for calories and 116% of their RDA for protein, while women's diets are markedly deficient, providing only 87% of their caloric RDA and 79% of their protein RDA. Folbre sees this phenomenon, and all inequitable household distribution, as arising out of structural asymmetries in the economic, social, and legal positions of men and women that give the two genders unequal bargaining power. Rather than seeing the household as a single "maximizing" unit, she sees it as a group of maximizing individuals, "in which individual family members co-operate with one another primarily to further their own personal interest." (This is not to deny that there will often be a high degree of overlap in the allocational priorities of household members; it is only to suggest that the overlap will rarely be 100% as assumed by the joint-utility function.) The ability to realize personal allocation priorities, which she calls "bargaining power," is seen to be affected by such factors as "the individual's contribution to the household income," his or her potential earnings outside the household, and even "support from extra-household coalitions struck by members of the same class, race or gender" .
From my own work among hill Hindus in Nepal, I would suggest that a number of additional social factors, such as support from the family of origin, freedom to divorce and remarry, polygamy, and individual personality, also affect women's bargaining power . Moreover, I would add that the concept of "maximization," to the extent that it is appropriate in trying to understand individual decision-making within the household, cannot be understood with reference to purely economic considerations.
Folbre's point and mine is that, instead of assuming an exogenously given joint-utility function or household distribution rule, we need to try to discover how the rule for household distribution varies in different cultural and economic settings and explicate the process through which household members establish and modify the rule.
A few caveats
Before sketching the beginnings of a conceptual framework that could be used as the basis for empirical research into some of these questions, I would like to emphasize that this subject demands an interdisciplinary approach, combining skills and perspectives from the fields of economics, health, and nutrition as well as anthropology.
The whole question of how to measure the set of household welfare "outputs," which I have grouped together as "child food consumption and health care," will need the attention of nutritionists and health experts. It is clear, moreover, that for a feeding practice such as breast-feeding there are many biological factors, not encompassed in this model, that influence the mother's decision on whether or not and how long a child is breast-fed. What this may mean is that this framework is more suitable for understanding the determinants of certain child care and feeding "outputs" than others.
Because these relationships are set out in quantifiable terms, what follows may seem very abstract and overly neat. The arrows and boxes in the figures may obscure the fact that some of the most important aspects of the central subject, which is the household decision-making process, cannot be quantified. Instead, these aspects will need to be captured through in-depth anthropological observation-what Geertz has called "thick description"-of the social bargaining process within the family . Statistical correlations mean little without a thorough knowledge of the context in which they occur. I would like to stress, therefore, that despite the quantitative nature of the model presented here, the approach envisioned is similar to that used for the study on the status of women in Nepal [11, 12] and to those currently being used for the Population Council's study of infant feeding practices in four countries  and the ILO's study on women's roles and demographic change in India . In all these approaches, survey research is combined with in-depth anthropological research involving extended residence in the community under study. Likewise, at the analysis stage both statistical methods and anthropological interpretation are called into play to complement and strengthen each other.
Proposed framework for analysis
Focus on the internal dynamics of the family
In the framework proposed in figure 1, biological and environmental factors are bracketed, and child health and nutritional status are analysed as outcomes of the social economy of the household. Our procedure is first to document the income and services that flow into the household from each family member and then to examine the process through which these resources are allocated and have consequences for the welfare of each family member.
The final welfare outputs in which we are particularly interested are the health and nutritional status of children in the household, especially infants and children under five. While these outputs would be the focus of the analysis, it is recognized that there are many other welfare outputs crucial to the development of the child. These would include shelter, clothing, the child's own leisure and play time, education, support for special life-cycle rituals, and even the leisure time of the mother and father, which may enable them to be more creative and involved as parents. If data on these sorts of outputs are gathered along with data on such factors as the child's food consumption and care time, they can also be examined within the same framework.
Moreover, by expanding the data collected to include the dietary intake and health and nutritional status of the mother and her children, it would be possible to look at the factors affecting both these interrelated dependent variables within the same framework. It would then be possible to identify situations in which the child's welfare was being purchased at the expense of the mother's. In fact, it is suggested that, as far as possible, data should be gathered on welfare outputs, including expenditure per person, dietary intake, and anthropometric measurements of health and nutritional status for the following age/sex groups: (1) male and female infants 0-1 year old; (2) male and female children 1-5 years old; (3) male and female children 6-15 years old; (4) adult women who (a) are unmarried, with no children, (b) have children under five years old, (c) are of childbearing age, (d) are past child-bearing age; (5) adult men who (a) are heads of households, (b) are dependent adults. In addition, data on the leisure time of all household members, which is also an important welfare output, will be available from the time-allocation data.
FIG. 1. Tentative framework for analysis: the major family determinants
Figure 1 shows the four main determinants within the family of the health and nutritional status of the child. These are, first of all, the child-care time and the income to purchase material goods such as food, clothes, medicine, and shelter or home-produced equivalents. But there are other factors that determine how the first two inputs get used. These are the priorities and bargaining power of different family members, which determine the pattern of distribution of household resources, and the knowledge and skills of various family members, which determine how efficiently the time and income inputs are transformed into child welfare.
One of the main resources the family has is the time of its members. Time can be used either in the production of services for the family (i.e., domestic work like cooking, cleaning, laundry, and child care) or in the production of goods. This production time includes wage or salary work as well as the unpaid subsistence production of food and other goods on the family farm.
While it has always been relatively easy to measure the contribution to family welfare made by the wage earner who contributes cash income, this has not been the case with subsistence production. Often this crucial contribution has been overlooked in macroeconomic analysis for two reasons: it has been difficult to assess individual input into what, in many cultures, is viewed as pooled or communal family production; and much of the food produced is consumed by the family and is usually not counted as part of the national GNP. However, Acharya and Bennett  have developed a technique that uses time-allocation data and detailed household-production data to estimate the relative value of each household member's contribution to household subsistence production without resorting to imputed wages rates. With this technique, it is possible to look at the income inputs not only of the wage earners who work in the market economy but also of those who work without pay for the family and produce income in kind on the family farm.
This technique opens the way for disaggregating inputs, as shown in figure 2. Disaggregation permits us to begin to analyse the trade-offs among different members of the family for certain types of inputs. For example, it allows us to determine, in cases where the mother is working outside the home and cannot spend as much time on child care and food preparation, whether or not other family members make up the difference. Evidence from the Philippines  shows that such substitution takes place, but there has been uncertainty about whether the quality of the care given by other family members, especially siblings, is equivalent to that which would be provided by the mother. A separate assessment of various care-takers' knowledge of and skills in child care and food preparation would attempt to address this issue.
Disaggregation of inputs also allows us to explore the relationship between income earning and bargaining power or the ability to influence how household resources are used. Among the hypotheses that could be tested are the following: (1) increases in maternal income are positively associated with increased maternal influence over income allocation; (2) increases in maternal influence over income allocation lead to the devotion of more resources to the child, better dietary intake for the child, and improved child health and nutritional status.
In figure 2 the welfare outputs are also disaggregated, first, as to what is received by each individual family member and, second, by the stages of the consumption or utilization process. If possible, data would be collected on the allocation of both cash and kind resources to each family member. This first stage of output is seen as the result of the total amount of income the family has (determinant 2 in fig. 2) and the influence of various members on the way that income is distributed (determinant 3).
FIG. 2. Tentative framework for analysis: disaggregation
As mentioned earlier, however, before this income and raw material can be consumed, two other inputs are necessary: time for child care, food preparation, and other domestic work (determinant 1) and the knowledge and skills (determinant 4) needed to transform income into a nutritious meal or adequate health care. The kinds of data that would be gathered to measure this second stage of welfare output would be (1) the protein/calorie dietary intake of each member, (2) feeding patterns (such as initiation, frequency, and duration of breast-feeding, age for supplementary foods, and frequency of meals), (3) health-care behaviour (such as treatment of diarrhoea, and expenditure on medical care), and (4) total child-care time received. Thus, the second-stage welfare output is conceptualized as the result of the income and raw material available per person plus the knowledge and time inputs of each family member.
Factors outside the family, such as the availability of foods and health care, would also need to be taken into consideration here, although they are not shown in the model. Also, in considering the issue of breast-feeding as part of the health care and feeding behaviour "output," it would be necessary to consider, in addition to the mother's time and knowledge (and that of other senior females in the household), her dietary intake and biological factors such as her health and age. These types of biological and environmental factors outside our model would be expected to have a proportionately greater effect on the final-stage welfare output, the health and nutritional status of each family member. Although this output would be measured, it would be expected to be a less sensitive indicator of what we are examining: the internal dynamics of the household economy as they affect the child.
Focus on the mother
Our model does, however, include demographic factors within the family and socio-cultural and economic factors outside it as these affect each member's ability to contribute to child welfare along the four major dimensions of time, income, influence on income allocation, and knowledge.
To begin to refine this vastly over-simplified schema and clarify some of these relationships, it will be useful to attempt to trace in more detail the forces affecting the mother's inputs into the health and nutritional status of the child. It must be remembered, however, that these inputs are complemented by those of other members of the family and in some cultures by the wider kin or neighbourhood-support group as well.
Figure 3 shows in more detail some of the kinds of cultural, demographic, and economic data that will need to be examined. The allocation of the mother's time is broken down into three major work categories, with leisure as the residual category, which is considered a welfare output. The first work category is time spent in domestic activities. The portion of this time devoted to preparation of food, caring for children, and assisting in health care is seen as one of the mother's direct inputs to child welfare. The second category is the time she spends working as unpaid labour on the family farm enterprise. This labour results in her contribution to household subsistence production, which is one of the components of maternal income shown in the figure.
The third time category encompasses work in the market sector.5 For our purposes, one of the most important aspects of this kind of work is that it produces income-either from wages or as the profit from own-account entrepreneurial activities. Whether the mother keeps this income to spend herself or contributes it to the family pool, it can be identified as her own earnings. Like the mother's contribution to subsistence income, the effect of her market income on child health and nutrition will be important to measure. It is important that these two components (or potential components, since many mothers do not earn wage income) be distinguished. Our hypothesis is that the second type of identifiable market income would have a greater effect than a women's subsistence income on (1) the degree of her influence over the allocation of pooled income and (2) her control over her own individual earnings-both of which will be measured separately.
We have hypothesized earlier that the greater the extent of a woman's influence over the allocation of income (whether pooled or individual), the better the child's dietary intake and nutritional status will be. Certainly, an assumption of this nature has been made in much of the relevant literature, but not much has been done to test it. Such is the strength of our unexamined belief in maternal altruism as a cultural universal. The model presented here allows us to look separately at both the degree and the direction of the mother's influence on expenditure of the household income on the child.
The forces that determine the degree of the mother's influence over resource distribution (i.e., her "bargaining power") are extremely complex. Economic, demographic, and socio-cultural factors affect this variable indirectly through their influence on the mother's allocation of time and, hence, her ability to contibute to household income. However, certain socio-cultural factors (such as gender ideology, support from family of origin, degree of knowledge of husband's earnings, and gender-specific obligations for child support) also have a direct effect on her control over the disposal of household resources.
FIG. 3. Focus on maternal inputs
The direction of the mother's influence however, is based on her own priorities, which will be, at least in part, socially conditioned and which can be observed and measured separately. Dwyer has pointed out that women's allocational priorities are often geared toward survival strategies, while men's priorities may focus on mobility . A woman's priorities may be conditioned by a need to protect against or prepare for divorce or desertion, especially in cultures where land ownership is in male hands. She may have to compete with a co-wife to secure family resources for her own children. The need to keep natal kin and neighbourhood support networks active by fulfilling certain social obligations may also be important to her. And, of course, societies differ as to whether women can express their individual priorities openly or must pursue them covertly.
The "knowledge" input would be measured with reference to certain specific practices regarding breast feeding, weaning age, supplementary feeding, response to diarrhoea and others, determined by nutritionists and physicians to be key to a child's survival. This variable is seen as conditioned primarily by socio-cultural factors, particularly the traditional child-care and feeding practices that prevail in the community. However, demographic variables, such as the age and educational status of the mother will obviously affect the degree to which the individual has absorbed these traditional beliefs and practices and the degree to which he or she has access to, and is responsive to, alternative ways of doing things. In some situations, economic variables, such as labour-force participation, will affect exposure to new ideas.
Potential programme and policy relevance
The proposed research would do much to clarify the relationship between a woman's role as a mother and her role as a producer. In particular, it should help to identify the social and economic conditions under which women's income-producing work results in healthier, better-nourished children and those conditions under which it fails to do so. Also, because this model does not define women's productive work narrowly in terms of labour-force participation, it will permit a much more accurate assessment of the actual extent of women's contribution to the income variable. It will also allow us to examine the differential effects of various kinds of income-producing work on the child's welfare. But perhaps the most important potential contribution this kind of study could make in terms of support to the design of programme interventions would be to provide a clearer grasp of the factors that determine decision-making patterns at the household level as they relate to child health and nutrition.
There are a number of specific questions that can be addressed within the proposed framework:
Statistical analysis of some of the issues may have important global policy implications and may also be of theoretical interest as well. However, the most useful information in terms of concrete project-design and the formulation of country-level policy by governments in the developing world is likely to be the insight gained from in-depth observation of how men, women, and children in poor families work out their individual and collective strategies for survival.
l. Nieves I. A balancing act: strategies to cope with work and motherhood in developing countries. Paper prepared for roundtable conference on The interface between poor women's nurturing and productive roles, International Center for Research on Women, Washington, DC, 1981.
2. Carloni AS. The impact of maternal employment and income on the nutritional status of children in rural areas of developing countries. Rome: FAO, 1983 (in draft).
3. Clark C. The microeconomic approach to the relation of women's market work and child welfare. Paper presented at roundtable conference, International Center for Research on Women, Washington, DC, 1982.
4. Dwyer DH. Women and income in the third world: implications for policy. Population Council International Programs Working Paper no. 18. New York: Population Council, 1983.
5. Engle PL. Maternal care, maternal substitutes, and children's welfare in the developed and developing countries. Paper prepared for roundtable conference on The interface between poor women's nurturing roles and productive responsibilities, International Center for Research on Women, Washington, DC, 1981.
6. Safilios-Rothschild C. The role of the family: a neglected aspect of rural poverty. World Bank Development Report. Washington, DC: World Bank, 1980.
7. Blau DM. Investment in child nutrition and women's allocation of time in developing countries. Center Discussion Paper no. 371. New Haven, Conn, USA: Economic Growth Center, Yale University, 1980.
8. Hart G. Women's participation in the labor force: implications for employment and health/nutrition programs. Ithaca, NY, USA: Cornell University, 1975.
9. Popkin BM, Solon F. Income, time, the working mother and child nutriture. Environ Child Health 1976; Aug: 15666.
10. Popkin BM, Bisgrove EZ. Urbanization and nutrition in low-income countries. Food Nutr Bull 1988;1Q(1):3-23.
11. Acharya M, Bennett L. Women and the subsistence sector: economic participation and household decision making in Nepal, World Bank Staff Working Paper no. 526. Washington, DC: World Bank, 1983.
12. Acharya M, Bennett L. The rural women of Nepal. In: The status of women. Vol. 2, part 9, Field studies. Kathmandu, Nepal: Centre for Economic Development and Administration, Tribhuran University, 1981.
13. Cain M, Khanam SR, Nahar S. Class, patriarchy and women's work in Bangladesh. Pop Devel Rev 1979; 5(3) :405-38.
14. DaVanzo J, Lee DLP. The compatibility of childcare with market and non-market activities: preliminary evidence from Malaysia. In: Buvinic M, Lycette MA, and McGreery WP, eds. Women and poverty in the third world. Baltimore, Md, USA: Johns Hopkins University Press, 1983.
15. Kumar SK. The role of household economy in child nutrition at low incomes: a case study in Kerala. Department of Agricultural Economics Occasional Paper no. 95. Ithaca, NY, USA: Cornell University, 1978.
16. Mueller E. Time use in rural Botswana. Ann Arbor, Mich, USA: Population Studies Center, University of Michigan, 1979.
17. Nag M, White B, Creighton P. An anthropological approach to the study of the economic value of children in Java and Nepal. Curr Anthrop 1978;19(2).
18. King-Quizon E. Time allocation and home production in Philippine rural households. Phil Econ J 1978;17(1,2).
19. Rosaldo MZ. The use and abuse of anthropology: reflections on feminism and cross-cultural understanding. Signs: J Women Cult Soc 1980;5(3):389-417.
20. Yanagesako, SJ. Family and household: the analysis of domestic groups. Ann Rev Anthrop 1979;8: 161-205.
21. Evenson RE, Popkin B, King-Quizon E. Nutrition, work, and demographic behavior in rural Philippine households. New Haven, Conn, USA: Economic Growth Center, Yale University, 1979.
22. Folbre N. Household production in the Philippines: a non-neoclassical approach. Econ Devel Cult Change 1984;32(2):303-30.
23. Fapohunda ER. Characteristics of women workers in Lagos: data for reconsideration by labour market theorists. La b Soc 1978 ;3(2): 158-71.
24. Guyer JI. Dynamic approaches to domestic budgeting: cases and methods from Africa. Paper prepared for seminar on Women and income control, Third World Conference, New York, 1982.
25. Kumar SK. A framework for tracing policy effects on intrahousehold food distribution. Washington, DC: International Food Policy Research Institute, 1983.
26. Roldan M. Intrahousehold patterns of money allocation and women's subordination: a case study of domestic outworkers in Mexico City. Paper presented at Population Council seminar on Women, income and policy, 1982.
27. Peluso NL. Survival strategies of rural women traders, or Women's place is in the market: four case studies from northwestern Slemen in the special region of Yogjakarta. Geneva: Rural Employment Policy Research Programme, International Labour Organisation, 1980.
28. Bennett L. Dangerous wives and sacred sisters: social and symbolic roles of high caste women in Nepal. New York: Columbia University Press, 1983.
29. Bennett L. The Parbatiya women of Bakundol. In: The status of women in Nepal. Vol. 2, part 7, Field studies. Kathmandu, Nepal: Centre for Economic Development and Administration, Tribuhuvan University, 1981.
30. Geertz C. The interpretation of culture. New York: Basic Books, 1973.
31. Van Esterik P. Integrating ethnographic and survey research: a review of the ethnographic component of a study of infant feeding practices in developing countries. Working Paper no. 17. New York: Population Council, 1983.
32. Nag M, Anker R, Khan ME. A guide to anthropological study of women's roles and demographic change in India. Working Paper no. 115. Geneva: Population and Labour Policies Programme, International Labour Organisation, 1982.
33. Goldschmidt-Clermont L. Output-related evaluation of unpaid household work. A challenge for time use studies. Home Econ Research J 1983;12(2)127-32.
34. Nieves I. The interface between poor women's nurturing roles and productive responsibilities. Paper prepared for second annual Women, Work and Public Policy Workshop. Cambridge, Mass, USA: Institute for International Development, Harvard University, 1982.
35. Saleha B, Greeley M. Women, employment and agriculture: notes from a Bangladesh case study. Brighton, Sussex, England: Institute of Development Studies, 1980.
Contents - Previous - Next