Part of the Demography, Population, and Ecology Commons, Education Commons, and the Public Health Education and Promotion Commons

The circumstances that characterize poor, rural communities in Malawi suggest that children's health-wealth gradient can vary from other settings. This dissertation begins with a description of the methods used to create a household wealth variable using assets data in the Malawi Longitudinal Study of Families and Health project. By using a fixed effects model to minimize omitted variable bias, I determine the influence of participating in a farm subsidy program on the levels of household wealth in 2004, 2006 and 2008. The results show that the program is positively associated with the wealth index score and this association is stronger when using lagged explanatory variables. This chapter demonstrates how asset data broadens the possibilities of wealth-poverty research that can be undertaken in poor settings. In the next chapter, I use the wealth index to identify a health-wealth gradient for children under 5 years, and I determine whether the gradient varies with age. I find that children in wealthier households have decreased risk of stunting but this is not significant until the oldest age groups (36-47 and 48-59 months). While there is no apparent health-wealth gradient across these ages, there is evidence of an emerging gradient as children get older. The final chapter explores the role of maternal social capital in children's schooling outcomes, using an index measure of women's membership in community groups and instrument variable analysis to address endogeneity concerns. I fin