This study is an analytical and empirical assessment of the impact of household structures on children’s welfare using household level data from South Africa. Specifically, the thesis measures the relationship between different types of household structures and their impact on children’s health, namely their height-forage z scores. Households are classified as “Female-only” and “Mixed-gender”, with the latter further differentiated as “Nuclear” and “Extended”. The study thus moves beyond the traditional “male-female” headship commonly adopted in the Women in Development (WID) literature in defining households.  Instead, the classification suggest that gender, kinship and other relationships characterize households and that children live not only in nuclear families but also in various forms of extended families.  The study analyzes children’s welfare from within the reality of these complex household structures.  This framework is used to look at resource allocation impact of different household structures on children’s welfare – the focus of many studies in the WID literature.  It also analyzes the impact of non-tangibles on children’s welfare, such as the presence of fathers and mothers – an emphasis placed in household studies in industrialized countries. 


When it comes to considering children’s welfare, does it matter who controls the intra-household resources, especially monetary resources? A growing body of evidence indicates that more resources in the hands of women mean greater household resource allocations to children (The World Bank, 2001). In Brazil, for example, additional income in the hands of mothers is associated with substantially larger improvements in child survival and nutrition than additional income in the hands of fathers. For child survival the marginal effect of female income is nearly 20 times larger than that of male income (The World Bank, 2001). And for child nutrition, the effect is four to eight times larger (Thomas 1990, 1997). Rogers (1996) too, suggests a positive relationship between women’s control over resources and the welfare of children. Wang (1997), using a data set from Zambia, confirms the importance of gender in the allocation of household resources but also suggests a more complex relationship in terms of the final impact on the welfare of children.

This story has been significantly influenced by a rapidly growing literature on Women in Development (WID) which advocates that public policies should focus on creating income-earning opportunities for women for important reasons.

First, currently women are heading many households with young children. This is attributed to male absence, due to widowhood, divorce or desertion (Desai and Ahmad, 1998). Given labor market segmentation and discrimination against women, these households are considerably poorer compared to households that contain men

(Buvinic and Gupta, 1994; Bruce, 1989).

Second, even when men are present in the household, women contributing to a large proportion of household income are likely to have some control over how income is spent. Since women are more child-centered than men (Bruce, 1989; Rogers, 1995), income under female control has greater positive impact for children’s health, nutrition and education than that earned by men (Blumberg, 1986; Knudsen and Yater, 1981; Tripps 1981).

Thus on the one hand, the argument is that children in “Female-only” households are more vulnerable because women are poorer, and on the other hand, children in these households are better off because women have greater pro-child expenditure preferences. These two approaches have coalesced into the literature on ‘female-headed households’ (FHH) which attempts to link female headship with positive child outcomes.

This study attempts to examine gender differences in resource allocation within the household and its relationship to children’s welfare in the South African context. Specifically, height-for-age of children between ages 0 to 90 months will be used as a proxy for a welfare measure. Data from the South African Living Standards Measurement Study (LSMS) from 1994 will be used.

The framework of this thesis is embedded in the approach adopted by researchers that view household outcomes as part of a bargaining process within the household. However, in classifying households along a gender dimension, the study moves beyond the simple “male-female” headship dichotomy, introducing instead a more elaborate taxonomy that recognizes that relationships within households are not only influenced by gender but also by kinship and other connections[1]. The status of children in “female-headed” households – in this thesis classified as “Female-only” – is compared to “Nuclear”, “Extended” and “Other” households – collectively called “Mixed-gender” households – instead of the simple “male-female-headship” dichotomy.

The overall analysis suggests that “Female-only” households are on an average poorer compared to other types of households. However, the thesis does not confirm that the welfare of children in “Female-only” households is protected through expenditure-switching strategy despite the lower incomes in these households. In fact the results suggest that children’s welfare is enhanced in family structures that have the presence of both males and females. The relationship between household structures and children’s welfare is far more complex than suggested by the “male-female-headship” dichotomy.