Welfare migration and the duration of welfare spells

This dissertation addresses the issues of welfare migration and the duration of welfare spells in three related essays. First, it examines the role of welfare benefit levels and labor market opportunities on the migration propensities of potential welfare recipients. In particular, the question the first chapter addresses is whether poor single mothers who are likely to be eligible for welfare benefits are more attracted to high benefit regions than are similar non-eligibles. This is the first paper to address this question using a nationally representative dataset that allows us to capture intra-state moves as well as inter-state moves. County cost of living measures are used to capture variation in the real value of state welfare benefits within states. Statistical tests strongly reject the hypothesis that poor single mothers are more attracted to higher benefits. This chapter also addresses the selection bias that plagues many earlier studies of welfare migration.;Second, the dissertation examines the duration of welfare spells using a methodology previously untried in the study of welfare spells. This methodology, the split population model, allows us to model the heterogeneity of the welfare population that has been noted for a quarter century. The incorporation of the split population model generates results that suggest different policy prescriptions than those derived from a simple parametric duration model. The dissertation further demonstrates that failure to control for the correlation across equations in the model---previously ignored in the split population literature---generates its own biases as well.;Third, the dissertation tests for state dependence in welfare participation while returning to the original idea of welfare migration. In particular, the chapter addresses the dual questions of whether welfare recipients exhibit state dependence in their use of welfare over time and whether welfare migrants exhibit greater state dependence than other welfare recipients. If welfare prompts some part of the population to move, we would expect that this would be the group with the greatest present discounted value of the future welfare benefit stream, and they are likely to be on welfare longer than average. Statistical tests strongly reject the hypothesis that welfare migrants exhibit more state dependence.