Measuring the Dynamic Effects of Welfare Time Limits

This paper develops a new dynamic panel data model that can formally incorporate the dynamics of welfare participation behavior under time limits. The model is estimated using data from a policy experiment in the United States and a generalized method of moments estimator. The effects of time limits are found to be larger and more dynamical than in previous regression approaches, after comparing estimation results from several approaches using the same data. Around 40 percent of the anticipatory effect of the time limit are due to individuals depleting their stock of remaining months of welfare eligibility. The anticipatory effect is also found to be much larger among disadvantaged individuals. The results call for further assessment of the importance of time limits in explaining the low welfare caseload in the post-welfare reform era.