A Markov Model of Turnover in Aid to Families with Dependent Children

Abstract

In order to gain some insight into turnover in the welfare population, a two-state (on and off welfare) Markov chain transition probability matrix is estimated with longitudinal data on welfare recipients coming on welfare in 1965. Results indicate that an enormous amount of turnover occurs in the welfare population and that the average duration of time on welfare per time on welfare is relatively modest for this group. Persons facing a wage below the minimum wage and/or high unemployment are less likely to leave welfare and more likely to return, likely to stay off for shorter periods and on for longer periods, and more likely to be on welfare than are those with low expected unemployment or wages above the minimum.