All the choice models considered in Chapters 6 and 7 thus far are models of unitary decision making. The defining characteristic of such models is that they treat the household as if it were a single actor, which seeks to maximize a single utility function.
We can easily incorporate concern for intrahousehold distribution (of food consump- tion, home time, and other goods) into unitary models of household decision making by assuming that the unitary decision maker maximizes a utility function in which food con- sumption, home time, and schooling enjoyed by female members, and food consumption, home time, and schooling enjoyed by male members, enter as distinct consumption items. After incorporating gender in this way:
Unitary models of household decision making point to roles that income, time endowments, prices, and preferences might play in determining the extent to which households provide female members with less food, home time, or schooling than they provide for male members.
Unitary models suggest, for example, that rising incomes can reduce gender gaps in con- sumption if households treat female consumption as a more strongly normal good than male consumption. Some evidence suggests that this is the case. For example, women and girls in southern India receive less nourishment than men and boys in the dry season, when income is especially low, but their consumption rises enough to equal men’s and boys’ consumption in the harvest season when incomes are higher (Behrman, 1988). Thus, rising incomes may be expected to bring some reduction in gender gaps. More generally, in most of the 32 countries included in a study of infant mortality, the ratio of female infant mortality rates to male rates was higher among poor families than among wealthy families (King and Mason, 2001).
Unitary models also suggest that increases in household time endowments—arising, for example, out of infrastructure projects that build wells or power sources close to homes—may lower gender gaps in schooling if households treat female schooling time as a more strongly normal good than male schooling time. The observation that women and girls bear most of the large time burden associated with carrying water in the developing world suggests significant potential to reduce gender gaps in schooling rates by improving access to water,
though direct estimates of the effects of such water projects on schooling are lacking (King and Mason, 2001).
Just as female consumption may be more sensitive to income and time endowment changes than male consumption in unitary models, so also it may be more sensitive to reductions in the cost of obtaining goods and services. Thus, general reductions in the prices of food, health ser- vices, and education can also decrease gender gaps. Several studies offer empirical evidence of such differences. Behrman and Deolalikar (1990), for example,find confirmation that some food price reductions would raise female nutrient intakes relative to male intakes, though the channels through which this happens are somewhat surprising. Theyfind that reductions in the prices of foods that are relative luxury goods in rural south India (including rice and milk) reduce nutrient intakes for men (presumably by inducing a substitution from staple foods toward luxury foods that provide fewer nutrients per rupee spent), but they leave the nutrient intakes for women and girls largely unchanged or in some cases increase them.
Unitary models also suggest the potential to raise nutrition, schooling, and health care for girls relative to boys by reducing the costs of goods and services for girls while holding the costs for boys constant. Successful girls’ scholarship programs bear out the significance of such efforts in education. For example, a secondary school scholarship program for girls in Bangladesh appears successful in raising girls’ enrollment (King and Mason, 2001).
Finally, in the unitary household model, gender differences in consumption and time use are determined in part by the unitary decision makers’ preferences.4This raises the possibility that advertising campaigns and education can alter gender gaps, even when income and prices remain constant.
Unitary models yield useful insights about gender and development, but:
Unitary theories suffer from three weaknesses as guides for research on gender in development.
One weakness is conceptual. Unitary theories assume that a household maximizes a single utility function. To apply this theory to a real-world household with multiple members, we must be willing to assume either that a single household head makes all the decisions for the household unilaterally or that all household members think alike and make unanimous decisions. A model that allows more nuanced possibilities, in which women may disagree with men (in the sense that their utility functions differ), and exercise at least some voice or autonomy, would be more compelling.
The other two weaknesses relate to unitary household model predictions that are contra- dicted by empirical evidence. First, to a unitary decision maker all income is equally useful for expanding the household budget and facilitating the purchase of goods and services, regardless of which family members bring the income into the household. Thus in unitary household models, household consumption decisions depend only on households’ total income and not separately on the incomes of men and women (see problem 5). The unitary decision maker is said to pool income belonging to all household members before making choices about consumption.
Several empirical observations suggest, instead, that households do not treat men’s and women’s income as identical. For example, Thomas (1990) finds that nonlabor income in the hands of Brazilian women has bigger impacts on family health, and especially on child survival probabilities, than nonlabor income in the hands of Brazilian men, and Duflo (2003) finds that pension income received by women in South Africa improved the nutritional status of girls in their households, whereas pension income received by men did not. Such evidence suggests that channeling development program benefits to women rather than men can increase program
4
For a more comprehensive review of what unitary models predict regarding intrahousehold allocations, see Strauss and Beegle (1996).
impacts on household nutrition and other investments in the human capital of children. Unitary models offer no guidance for deeper study of this possibility.
Unitary household models also predict that households will put resources to efficient use. This means that households allocate land, nonlabor income, and other resources across uses in such a way that it is impossible to raise one member’s well-being without reducing the well-being of another. Unitary models predict efficiency as long as the unitary decision maker gives at least some positive weight to each member’s individual well-being when assessing overall household utility. Under such circumstances, if household resource use were inefficient, household utility would not be maximized. Inefficiency (by definition) implies that one member’s well-being could be raised without reducing other member’s well-being and thus that household utility (which is a positive function of all members’ well-being) could be increased. If household decisions are indeed efficient, then while gender biases can leave female members worse off than male members within households, they do not prevent households from making the most of their resources and exiting poverty.
Several studies suggest, however, that use of household resources may be inefficient in at least some contexts. For example, Udry (1996) offers provocative evidence of inefficient allocation of agricultural inputs across male and female plots of land among rural households in Burkina Faso (see Box 7.3 later). Dercon and Krishnan (2000) offer somewhat more subtle evidence of inefficiency within households in southern Ethiopia. They point out that intra- household efficiency requires that household members “insure” each other, in the sense that they share the burden of coping with income shocks, even when the shocks affect only one of them directly. That is, husbands and wives who have not been hit by shocks should provide transfers to spouses who have been hit by shocks. The authors demonstrate that in southern Ethiopia men do not provide their wives with much insurance against health shocks, instead leaving women to bear the burden on their own. This is inefficient, because the men could have made themselves better off without leaving women worse off if they had provided women with insurance (preventing women from having to bear the burden of health shocks alone) while requiring women to“pay” them for the insurance by allowing men to consume a larger average share of household resources.