The Effect of PRWORA on Welfare Caseloads in Pennsylvania: Fixed Effects Versus Random Effects Models

Samuel Sturgeon, Pennsylvania State University

This paper uses administrative panel data from various government agencies from 1990-2002 to test the impact of the 1996 welfare reform on caseload numbers in Pennsylvania counties. The results of fixed and random effects models are compared. Though both models suggest that PRWORA had a significant effect on welfare caseloads, the fixed effects model assumes that this effect was constant across counties, does not allow a test of individual county slopes, and does not allow a test of specific, measured county level fixed effects. The random effects model allows the effect of PRWORA to differ by county and also allows for the direct test of fixed county level variables. In particular, I tested whether the effect of PRWORA differed in urban and rural counties, and found that the impact of PRWORA was not uniform across counties, but had a stronger impact on caseload reductions in urban counties.