Association for Public Policy Analysis & Management

The Head Start program has long aimed to provide high quality early care and education (ECE) experiences for low-income children. Despite also being a program for families, research on Head Start has paid less attention to parents’ human capital outcomes. This is troubling considering that increased human capital – employment, earnings, education, and welfare independence – promotes family wellbeing and children’s early learning (e.g., Magnuson, 2007; Reardon, 2011). Moreover, Head Start is a promising avenue for promoting parental human capital. For example, parents whose children are in Head Start should face fewer time and cost constraints while they attend work or school, may be exposed to beneficial social networks through center-based ECE communities (e.g., Foster, 2002; Sommer et al., 2012).

An existing body of research has estimated positive impacts of subsidized child care on parents’ human capital outcomes (e.g., Blau & Tekin, 2007), and one prior study associates Head Start participation with increased parental educational attainment (Sabol & Chase-Lansdale, 2015). However, this study exclusively examined average effects of program participation, neglecting the possibility that program effects on parents’ human capital might vary across Head Start centers. Furthermore, this study focused on a subset of parents’ human capital outcomes and thus did not account for the full set that might be amenable to change as a result of Head Start.

The present study takes advantage of the random assignment and multi-site design of the Head Start Impact Study to 1) estimate average effects of program participation on the full set of human capital outcomes and 2) quantify variation in these impacts by exploiting center-level random assignment of participants. We examine average and then explore site-level variation using random assignment as the “intent to treat” (ITT) within a 2-level random coefficients framework. In addition to educational attainment and employment, we also seek to estimate impacts on earnings and receipt of public benefits.

We initially replicate prior work estimating an average ITT effect of Head Start on educational attainment for the first study cohort observed three years after random assignment. In this same cohort, we additionally estimate marginal average effects of treatment on parents’ likelihood of employment one year after assignment (b=.04; SE=.02; p=.06) and a significant positive change in families’ income-to-needs ratio two years later (b=.09; SE=.02; p=.03). We also estimate significant variation in Head Start’s effects on families’ income-to-needs ratios four years after random assignment (Q=332.78; p=.05).

These preliminary results demonstrate that some Head Start programs are more effective than others in promoting participant earnings, but that centers more consistently promote educational attainment and employment status. Future analyses will account for noncompliance likely to attenuate ITT estimates using an instrumental variables approach to estimate average and heterogeneous local average treatment effects (LATE). Using these methodologies to quantify both the average effects of and variation in program impacts on parents’ human capital may serve to inform program administrators as they design and implement more intentional two-generation programs that provide ECE services in Head Start contexts.