The welfare caseload evolves through a process of flows onto and off of welfare that can be described with a Markov Chain model. Using formal results for Markov models, this paper examines the dynamic properties of the welfare caseload. In particular, the authors examine steady states, the speed of convergence, and the relative importance of entry and exit for changes in the caseload. Implementing these models with administrative data for California, the authors find that the welfare caseload has considerable momentum and that adjustments are far from instantaneous. In addition, the authors find that changes in the entry rate are empirically more important than changes in the exit rate for explaining changes in the overall caseload. These findings have considerable implications for the conventional methods that are used to study the changing caseload.
