The Long-Run Effects of Head Start: Replicating Ludwig-Miller (2007) and a Framework for Studying Intergenerational Mobility

Abstract

This paper replicates the influential Ludwig and Miller (2007) regression discontinuity analysis of Head Start and develops a framework for extending it to study intergenerational mobility. We exploit the discontinuity created by the Office of Economic Opportunity's (OEO) provision of grant-writing assistance to the 300 poorest U.S. counties in 1965. Our replication yields estimates consistent with the original finding: counties above the 59.2% poverty threshold show approximately 1.2–1.8 fewer deaths per 100,000 children ages 5–9 from Head Start-related causes, though our preferred specification (18pp bandwidth, local linear) yields a marginally significant estimate ($\tau$ = -1.20, SE = 0.66, p = 0.07; 95% CI: [-2.49, 0.10]). Validity tests support the design's credibility: the McCrary density test shows no evidence of manipulation (log discontinuity = -0.002, SE = 0.15), and placebo tests using pre-program mortality show no spurious effects ($\tau$ = -0.64, p = 0.72). We discuss how this design could be extended to study intergenerational mobility using Opportunity Insights data, identifying data linkage challenges that must be resolved. The paper contributes a replication of a canonical RDD study and a roadmap for future research on the long-run effects of early childhood interventions. \bigskip JEL Codes: I38, J13, I24, R10 Keywords: Head Start, Intergenerational Mobility, Regression Discontinuity, Early Childhood Education