Heterogeneous treatment and self-selection in a wage subsidy experiment

The Self-Sufficiency Project (SSP) is a research and demonstration project that offered a generous time-limited income supplement to randomly selected welfare applicants under two conditions. The first, the eligibility condition, required that they remain on welfare for at least twelve months. The second, the qualification condition, required that they find a full-time job within twelve months after establishing eligibility. In this paper we focus on a neglected and important feature of the program, namely that the financial reward for becoming qualified is inversely related to the expected wage rate. Under very simple assumptions we show that those who have a low expected wage rate have a clear incentive to establish eligibility. Empirical non-parametric evidence strongly suggests that individuals self-select into eligibility. We jointly estimate a participation equation and a wage equation that are correlated through individual random effects. Our results show that the omission of self-selectivity into qualification translates into slightly overestimated treatment effects.