Was it the economy or reform that precipitated the steep decline in the US welfare caseload?

Objectives of the 1996 overhaul of the US welfare system included reducing dependency, raising employment and de-incentivizing out-of-wedlock fertility. Using public use state-level panel data from 1990 to 2005, I analyse how state implementation of welfare reform simultaneously affects the caseload, employment and out-of-wedlock births (henceforth, Temporary Assistance for Needy Families (TANF) objectives). Because endogeneity and simultaneity could not be rejected, I use Three-Stage Least Squares (3SLS) method. Results indicated that most of the steep decline in the caseload is attributed to welfare reform, while the economy's overall effect paled in comparison. However, lagged and contemporaneous unemployment individually ranked second and third behind the Hispanic share of state population. The conservative tilt over the period studied ranked forth, followed in declining order by full family sanctions, Earned Income Tax Credit (EITC) payments, time limits (lagged by length), access to abortion clinics, lump-sum TANF diversion payments, and TANF benefit payments. Findings also suggest that policy does not always work as intended: caseloads are found to be higher in states that have highly regarded family formation and job retention TANF programs; and EITC payments are associated with lower not higher caseloads. The most compelling finding in this study is that low-income families likely turn first to unemployment insurance and then to TANF assistance.