In this section we examine decisions about the allocation of resources within the household. We can think about allocation decisions as they pertain to consumption (cf., anthropological studies of food consumption), asset allocation, and investments in children, but we will focus in particular on allocation decisions regarding children.
I want first to go over some of the common elements and issues found in economic perspectives on these questions. Then I'll provide a verbal (i.e., nontechnical) review of a few representative studies and their findings. Subsequently, we'll look at evidence from Kinshasa on some related issues, including resource transfers across households.
Economic perspectives on resource allocation within the family
A common feature of economists' studies of resource allocation within the household is the presence of interdependent utility functions. That is, the utility (or income, or survival) of children enters directly into the parental utility function (cf., altruism).
Further, some researchers start with a family utility function (Becker's common preference model), while others allow for individual preferences (à la the bargaining model approach). These two approaches have somewhat differing empirical implications, and as Haveman and Wolfe (H & W) note, there is some recent evidence that supports the individual preference models.
The allocation of resources among children in the household clearly can take place according to at least two alternative types of decision rules. One criterion for such allocation decisions would be equality. There are two potential aspects here, though: equality of inputs (all children receive the same), or equality of outcomes (entailing compensatory distribution, with those who start with less in the way of initial endowments receiving more).
An alternative criterion would be to base allocation decisions on efficiency. In this scenario, resources would go to those children who offer the best prospects of gaining the most or providing the greatest return to parents for the "resource investment." This entails reinforcing distribution.
Outcomes for children (children's attainments) are seen in the economic view as dependent on the resources invested in the children. Also relevant as influencing factors are the number of siblings, the environment (neighborhood) in which the children grow up, changes in location, and family structure (and changes in family structure).
Figure 1 from Haveman and Wolfe, adapted from Leibowitz, provides a nice visual summary of the economic approach to home investments in children. The initial genetic endowment of a child is augmented by home investments of time and market goods so as to enhance ability [Ability = f1 (Genetic Factors, Home Investments)]. The ability, home investments, and family income all contribute to the child's final schooling level [Schooling = f2 (Ability, Home Investments, Family Income)].
A child's ability to generate income is seen as a reflection of all of the preceding factors plus post-school investments [i.e., Child's Income = f3 (Ability, Schooling, Home Investments, Family Income, and Postschool Investments)]. Note the multiple paths by which family income influences child outcomes; home investment also has multiple aspects to it. Note further that the three equations constitute a recursive system.
H & W note that this economic perspective, as well as perspectives on the determinants of children's attainments from other social sciences, emphasizes parental (family) circumstances and choices while neglecting choices made by society (or the government) and also neglecting choices made by the children themselves. They argue for a more comprehensive economic perspective incorporating all three elements.
In this view, government makes "the social investment in children" that determines the opportunities available to children and their parents. These opportunities, in turn, along with family-specific considerations, influence parental choices regarding the quantity and quality of resources devoted to children. Finally, given the opportunities available to them and the investments that have been made in them by their parents, children make utility-maximizing choices. Hence, government establishes the basic environment within which families and children make their choices.
Child survival, health, and human capital investments in children
Here we'll review three papers that raise issues relevant to this section. The first paper, by Rosenzweig and Schultz (R & S), is called "Market Opportunities, Genetic Endowments, and Intrafamily Resource Distribution: Child Survival in Rural India" (1982).
R & S treat child survival as an indicator of parental investment in children in a low-income country. Further, they note that in India and Pakistan, in contrast to elsewhere around the globe, mortality among children is higher for females than for males. They try to link this excess female mortality to economic behavior.
They begin with a family utility function, so as to be able to study intergenerational resource allocation as "an orderly optimizing process." The arguments (elements) of the utility function are parental consumption, the number of surviving male children, and the number of surviving female children.
Parents are able to affect child survival through the resources allocated to children. Children, in turn, are subject to gender-specific survival functions (what R & S called "biologically determined depreciation functions"): the more they consume, the greater the likelihood of their survival. At the same time, children not only consume; they also provide resources to the household via direct labor contributions or transfers.
R & S derive the equilibrium conditions of their formal model, and their analysis focuses on the condition that says that the shadow price of the resources devoted to consumption of children of sex i (i=m,f) will be lower the higher are the sex-specific monetary contributions of surviving children of sex i. That is, the net costs to parents of feeding children of a given sex are lower if children of that sex are expected to provide higher incomes to the household than children of the other sex.
Resource allocation to children is not observed directly. However, child survival and labor market opportunities for each sex are observable. R & S note that (so long as surviving male and female children are not strong complements in the utility function) their model suggests that parental resource allocation behavior will favor children who have higher expected earnings opportunities as adults (typically, boys).
This translates into their working hypothesis that sex differences in child survival rates will be related to the relative returns to survival, with parents selectively allocating resources to children in response to variation in sex differences in their expected earnings opportunities as adults.
Based on extensive data analyses, they conclude that "intrahousehold allocation of resources is highly responsive to market signals... where women's expected employment in the labor market is relatively high, [female survival is higher suggesting that] female children evidently receive a larger share of household resources relative to male children."
An extension of the work by Rosenzweig and Schultz is provided in a 1990 paper by Pitt, Rosenzweig, and Hassan (PRH) entitled "Productivity, Health, and Inequality in the Intrahousehold Distribution of Food in Low-Income Countries." PRH motivate their analysis by referring to the presence of distinct gender differences in food consumption in low-income countries, especially in South and West Asia.
From R & S, we already have the notion that such gender-based inequality may reflect unequal labor market opportunities between men and women, with the monetary return from allocating food to females smaller than the return from allocating to food to males. PRH examine the relationship between the internal distribution of food within the household and labor market activities in a model that ties together food consumption, health, and labor market productivity.
Their model allows for individual heterogeneity in initial health endowments. Note that this implicitly raises a question touched on earlier: in the presence of heterogeneity across children, do parents compensate for differences in initial endowments, or do they complement/reinforce those differences in endowments? Further, PRH do not explore the reasons for unequal labor market opportunities by gender -- these differences are simply taken as given.
They begin by expressing health status (hik, the health status of individual i in class k, where classes are differentiated by age and gender) as a function of three variables: food consumption, the effort expended in work activity, and the initial health endowment. More food improves health status, while more work effort reduces it.
Labor market outcomes (wages) for individuals in a given class depend (positively) on effort and health status, and the returns to effort increase with health status. Hence, food consumption increases productivity in the model via improving health status. Household utility depends positively on the health status and consumption of each household member, and negatively on the work effort of each member.
When the model is solved for its equilibrium condition for allocation of food to individual i of class k, we get a complex expression that boils down to:
The interpretation of this expression is straightforward. To maximize utility from food consumption, individual i in class k should consume food up to the point at which the additional utility from food consumption (obtained through food consumption directly, MUc, and indirectly through improved health status, MUh * MPch) is just equal to the net marginal cost of food (the price of food minus the additional income generated due to food's impact on improving health status and hence earnings -- MPhw * MPch).
The equilibrium condition for the assignment of work effort to individual i of class k has the following form:
To maximize utility from expenditure of work effort, individual i of class k should provide increased effort up to the point at which the additional disutility from effort (both directly, MUe, and indirectly via deteriorated health status, MUh * MPeh) is just equal to the net monetary gain from increased effort (the higher income associated with greater effort, MPew, offset by the indirect adverse effect of increased effort on earnings via deteriorated health status, MPhw * MPeh).
As PRH note, the net marginal cost of allocating food to person i is lower the greater is the extent to which health increases work efficiency (MPhw). If class differences in tasks performed in the labor market result in differences in MPhw, this will result in differential food allocation across classes.
Their empirical analyses provide strong support for the notion that increased food consumption contributes to health status. There is also weak support for the idea that increased work effort results in a deterioration of health status.
They also find significant reinforcement with respect to calories. That is, individuals with greater weight-for-height consume more calories. This reinforcement exists for men but not for women. Among those aged 6-12, there is reinforcement for both males and females. PRH view this reinforcement as linked to activity levels and gender differences in activity.
Finally, they find evidence of a pecuniary return to health and effort, with adult males with higher endowments more likely to engage in very energy-intensive work. This is not the case among adult females. Overall, then, greater weight-for-height increases health directly, and indirectly increases health via increased consumption, while reducing health via increased effort.
The final study that we'll consider here is by Behrman, Pollak, and Taubman (BPT) from 1986, and entitled "Do Parents Favor Boys?" BPT focus on parental investments in the human capital of their children. They begin by noting the existence of three distinct mechanisms that might lead to systematic gender differences in human capital investment.
First, parents may respond to expected gender wage differentials (or differences by gender in economic opportunities more broadly) in the labor market. This is the Rosenzweig-Schultz theme. Second, parents may respond to systematic differences by gender in the price of human capital investments. Such differences might exist with respect to opportunity cost if one gender's youthful work at home or work in the market has higher value than that of the other gender.
Third, parental preferences might favor girls or boys in the sense that parents value identical outcomes at identical cost more highly for one sex than for the other (cf., patriarchy and old-age support, which could also be seen as reducing the net costs to parents of human capital investments). This last "unequal concern" is the focus of the title of their paper.
In the BPT model, children's earnings (as distinct from wealth or income) are a distinct argument in the parents' utility function. If parents' choices about investments in children's schooling were made solely on the basis of efficiency grounds, parents would invest in each child until the expected rate of return on the child's human capital was equal to the market rate of interest (gifts and bequests could then be used to equalize or move to more equal incomes).
However, in the BPT approach, parental investments in the schooling of their children will reflect not only the children's earnings functions, but also parents' preferences about their children's earnings. In particular, allocation of resources to human capital investment will reflect parental concerns about equality as well as efficiency.
BPT add an additional wrinkle by focusing on human-capital-dependent (HCD) income. This essentially incorporates consideration of spouse's income (in addition to the child's own income), on the assumption that marriage market outcomes depend on human capital and parents take account of these outcomes in allocating human capital investments among their children.
The model, then, is one in which parental welfare depends on the expected HCD income of each child. Welfare is maximized subject to a budget constraint for family resources devoted to human capital investment. The price of human capital investments is assumed equal across children. Further, there are HCD production functions for each child, in which income depends on the child's education, other human capital investments, gender, genetic endowments influencing earnings/HCD income, and the same variables for the child's spouse, plus the weight attached to the spouse's expected earnings.
BPT work through the equilibrium conditions for optimal parental investment in children's education, and these conditions may be represented as follows:
That is, the utility of parents will be maximized when the marginal utility they receive from increased child income multiplied by the marginal productivity (in generating income) of expenditures on investment in the child's education is equal for all children.
This equilibrium condition simply means that utility is maximized when the "bang for a buck" to parents (the utility increment associated with increased spending on education) is equalized across all children. If this condition is not satisfied, a reallocation of investment in education away from one child (with either relatively low marginal productivity of expenditures on education or relatively low marginal utility to the parents of increases in the child's income) and toward another would increase utility.
This equilibrium condition is represented geometrically in Fig. I.E.1 (BPT's Fig. 1). The parental welfare (utility) function represents the parents' preferences regarding the HCD incomes of child i and child j. If parents' preferences were for complete equality, the welfare function would be L-shaped. If, by contrast, parents were concerned only with efficiency, the welfare function would be a straight line with a slope of -1.
The curvature of the parental utility function reflects a tradeoff between equality and efficiency, and parental preferences for the distribution of income among their children. Further, the symmetry of the welfare function around the 45 degree line reflects equal concern for children. Unequal concern entails a shift away from symmetry at the 45 degree line -- e.g., a shift to the right would entail greater concern for child i.
The HCD-income possibility frontier shows the opportunities, given the resources available for investments in human capital, of generating different incomes for child i and child j. The dashed production possibility frontier in the figure shows a situation where the genetic endowments of child j relative to child i are more favorable for producing income as compared to the solid production possibility frontier.
If there are gender differences in labor market outcomes (e.g., gender wage differentials) and i and j are of different genders, the dashed frontier may also be viewed as representing a situation where gender i is disadvantaged relative to gender j.
BPT test their model using U.S. data, and they conclude that parental behavior regarding investment in education demonstrates substantial concern about equality. They find weak evidence of unequal concern favoring girls, and also support for their expanded model taking account of expected spouse earnings. The unequal concern finding implies that parental investments do not simply reflect the labor market incentives to invest more in boys.
Evidence from Kinshasa: Gender differences in access to schooling and resource transfers across households
We've already looked a little at gender differences in access to schooling in Kinshasa, and it is useful to reconsider them briefly here. The data show little in the way of a gender difference up through the age at which many children complete primary school, but beyond that age there is distinctly higher enrollment for boys as compared to girls. Given that labor force participation of men in Kinshasa is about 90 percent while that of women is about 40 percent, this is certainly consistent with the Rosenzweig and Schultz argument that gender differences in investments in children will reflect corresponding gender differences in labor market outcomes.
Further, the Kinshasa data show some evidence of narrowing of gender differences as household economic well-being increases. This appears to be broadly consistent with the R & S emphasis on efficiency considerations in a low-income setting, in conjunction with the BPT finding of equality concerns in a high-income setting.
At the same time, this narrowing is by no means universal. Gender differences beyond primary school appear to be largest for not for the poorest of the poor but for those from poor households at the next highest rung on the economic ladder.
Consider now the phenomenon of interhousehold resource transfers. Such transfers constitute what my colleague Professor Tambashe refers to as "solidarity" behavior, and typically take place within the extended family. These transfers may be for a variety of reasons, including simply "making ends meet," but transfers in support of children's schooling or child fostering are especially significant. These latter transfers are the focus of our analysis.
As can be seen from overhead Table 1, there is widespread participation of Kinshasa households in solidarity networks. Two-thirds of households have provided such assistance, and a quarter of the households have received it. This imbalance presumably reflects exchanges with rural areas (incomes are higher in Kinshasa, and there is better access to good schools). Further, among those who have received assistance a large percentage has also provided assistance (i.e., there are reciprocal flows).
Examination of the sources and recipients of assistance (overhead Table 2) documents that these resource flows indeed take place primarily within the extended family, and may include both intergenerational and intragenerational flows. Thus, for example, parents and siblings were the main sources of support for those who had received assistance, while siblings and nieces or nephews were the main recipients of support provided by those who had given assistance.
Finally, multivariate analyses of giving and receiving behavior (overhead Table 3) identify a number of highly significant determinants of this behavior. Economic well-being of the household, employment status and educational attainment of the wife, fertility and the age and sex composition of own children, age, and ethnic group are all significantly related to solidarity behavior.
The existence of solidarity behavior has potentially important implications. To the extent that individuals are able to rely on financial support from other members of the extended family, budget constraints are relaxed. In particular, in the presence of child fostering and interhousehold assistance with expenses for children's schooling, the full costs to parents of having and raising children are reduced. Presumably this tends to have a positive impact on fertility.
More generally, solidarity behavior clearly makes analysis more complex, because of the fact that the extended family comes into the picture. We're no longer looking at resource allocation within the household; resources are also allocated across households within the extended family.
A related issue of interest here is how such solidarity behavior has been influenced by the acute economic crisis that Zaire has gone through since 1990.