Experimental Evidence on Distributional Impacts of Head Start [Revise and resubmit, Journal of Political Economy]
- Marianne P. Bitler, University of California, Irvine and NBER
- Thurston Domina, University of California, Irvine
- Hilary Hoynes, Goldman School of Public Policy, University of California, Berkeley
- Goldman School of Public Policy Working Paper (August 2014)
This study provides the first comprehensive analysis of the distributional effects of Head Start, using the first national randomized experiment of the Head Start program (the Head Start Impact Study). We examine program effects on cognitive and non-cognitive outcomes and explore the heterogeneous effects of the program through 1st grade by estimating quantile treatment effects under endogeneity (IV-QTE) as well as various types of subgroup mean treatment effects and two-stage least squares treatment effects. We find that (the experimentally manipulated) Head Start attendance leads to large and statistically significant gains in cognitive achievement during the pre-school period and that the gains are largest at the bottom of the distribution. Once the children enter elementary school, the cognitive gains fade out for the full population, but importantly, cognitive gains persist through 1st grade for some Spanish speakers. These results provide strong evidence in favor of a compensatory model of the educational process. Additionally, our findings of large effects at the bottom are consistent with an interpretation that the relatively large gains in the well-studied Perry Preschool Program are in part due to the low baseline skills in the Perry study population. We find no evidence that the counterfactual care setting plays a large role in explaining the differences between the HSIS and Perry findings.
Current Draft (2MB)