Do welfare benefits cause the welfare caseload?

Analyses of the welfare system generally examine one of five competing models: (1) The work disincentive model; (2) the human capital model; (3) the macroeconomic model; (4) the public choice model; or (5) the cost-of-job-loss model. This paper employs the Granger causality concept and the multiple-rank F statistic to test the implications of all five of these models simultaneously. The results offer modest support to all but the macroeconomic model. The relationships among welfare benefits, caseloads, and labor market conditions appear to be too complex to be fully captured by a single model.