This paper builds on Maluccio et al.ís (1999) finding that social capital has a positive
effect on per capita expenditure in South Africa in 1998 but an insignificant effect in
1993. Utilizing the Kwazulu-Natal Income Dynamics Data Set involving household
surveys, this analysis determines whether the effect of social capital, for which group
membership is a proxy, on per capita expenditure varies by gender. This study also
uses a Chow test to assess whether distinguishing by group function adds any
explanatory power. Results from 1993 indicate that women experience lower returns to
social capital relative to men. In 1998, however, the returns to social capital did not
vary statistically significantly on the basis of whether one is male or female. The Chow
test indicates that the effect of the type of oneís group membership does indeed use a
different ìprocessî in predicting household welfare.