Rural, Suburban and Urban Single Mothers' AFDC and FSP Participation and Labor Supply: Lessons for Welfare Reform

Since passage of the Personal Responsibility and Reconciliation Act of 1996 (PRWORA), states have the responsibility of developing and implementing their own Temporary Assistance to Needy Families (TANF) programs to operate in tandem with the Federal Food Stamp Program (FSP). The context for this welfare reform included a booming economy and broad public perception that welfare programs severely reduce the work effort of recipients. This study focuses on a period when the economy was in recession and investigates how the old cash welfare program, AFDC, and the FSP affected labor supply (weekly hours of work) decisions for single mothers, the majority of welfare recipients, across the rural-urban continuum. The central question is how the labor supply of single mothers responded to the availability of AFDC and FSP benefits, respectively, and whether their responses differed depending upon whether they reside in rural or urban areas. To answer this question, we utilize data from a special in-house Census Bureau extract from the Survey of Income and Program Participation with accurate rural and urban sub samples to estimate our three equations model, one for labor supply and one each for AFDC and FSP participation. The econometric model involves two linked components. The first is Bivariate Probit estimation of the AFDC and FSP participation decisions to account for possible correlation between the error terms of the participation equations. The participation estimates are linked to the second component, estimation of the labor supply equation, due to the endogeneity of the participation decisions and the possibility of bivariate selection. Single mothers may and do choose participation in either or both AFDC and the FSP and unobserved characteristics associated with those participation choices are likely to be negatively correlated with unobserved factors affecting labor supply. The participation estimates are used to calculate bivariate sample selection correction factors added as auxiliary variables in the labor supply equation. Because wages play important roles in all three equations, yet are observed only for women who work, we first impute wages based upon Heckman’s two-step sample selection bias correction procedure for rural and urban sub samples. The results show that bivariate rather than univariate participation estimation is necessary. The bivariate selection corrections in the labor supply equations, however, yield mixed results. Nonetheless, the estimated model reasonably explains linkages between AFDC and FSP programs and labor supply. The results show that increasing the AFDC tax on earnings by 10 percent generates almost identical average increases in labor supply by rural and urban single mothers, 0.12 and 0.11 hours per week, respectively. Their responses are also similar with respect to a 10 percent increase in the FSP earnings tax, 0.03 and 0.04 hours per week on average. A 10 percent increase in AFDC and FSP unearned income tax rates yield average rural labor supply increases of 0.11 and 0.02 hours per week, respectively, and corresponding urban responses of 0.01 and –0.02 hours per week. With one exception, rural and urban single mothers reduce hours of work as expected when AFDC and FSP guaranteed benefits increase by 10 percent. For the AFDC benefit increase, rural single mothers reduce labor supply less on average than do urban single mothers, -0.08 vs. -0.12 hours per week. The FSP benefit increase generates the largest reduction in labor supply, -0.35 hours per week for rural single mothers, compared to a labor supply increase of 0.14 hours per week by urban single mothers.