In an influential article, Bitler, Gelbach and Hoynes (American Economic Re- view, 2006; 96, 988-1012) illustrate the importance of estimating heterogeneous impacts of welfare reform experiments. They find that the mean treatment effect offers an uninfor- mative summary of opposing effects, while the treatment effects are significantly different across quantiles. We replicate their results and evaluate the robustness of their findings to accounting for individual-specific heterogeneity possibly associated with welfare program participation. We find results that are in general similar to Bitler’s et al. findings, although the interpretation of labor supply effects in the upper tail is revised. We find no evidence of behavioral induced participation.

Document Type

Research Paper

Publication Date


Discussion Paper Number

DP 2014-11