In Milwaukee, Wisconsin (1994--1997) a work-based, anti-poverty intervention, "New Hope," randomly assigned an income subsidy---similar to the EITC---and a child care subsidy to a group of economically disadvantaged families. Randomly chosen applicants had access to these benefits for three years. The experimental evaluation found positive effects of the program on labor supply, income, and child care use. Notably, the program also boosted various measures of child academic achievement. Using New Hope data, this thesis offers two novel contributions to the New Hope literature and to the broader research on income and child care subsidies. First, I find that the effects of New Hope on labor supply, earnings, and child care use vary according to whether the family had a young child (six years old or less) at home or not while they were eligible to the program. Furthermore, compared to children older than six years of age, New Hope boosted child academic performance of younger children by more than double. Second, I disentangle the mechanisms that explain the impact of New Hope on the human capital of children who were young while their families were under New Hope. To this end, I develop and estimate a dynamic-discrete choice model of the household and child academic achievement. Counterfactual analyses based on the structural framework indicate that the bulk of the impact of New Hope on the academic achievement of young children is explained because of the policy-induced increase in center-based child care use. Consistently, the larger share of the impact of New Hope on child human capital is explained by the child care subsidy component of the New Hope policy bundle.