This study contributes to our understanding of the importance of cultural context for development policy. We show how the success of education policies depends on a widely-practiced marriage custom called bride price, which is a payment made by the husband to the wife's parents at marriage. We begin by developing a model of parental investment in children's education when there is a bride price. The model generates a number of predictions that we test in two countries -- Indonesia and Zambia -- that had large-scale school construction projects. Consistent with the model, we find that the amount of bride price received by the parents is increasing in their daughter's education. As a consequence, among bride price ethnic groups, female educational attainment is higher, and female education is more responsive to the school construction programs. In fact, we find that the programs had no discernible effect on the education of girls from non-bride price groups, but had large positive effects for girls from bride price groups. Thus, the success of the programs depended critically on the marriage custom.