This analysis is divided into two related components. In the first component, referred to as the demand-side analysis, exits from the Medicaid program are modeled as a function of local labor market conditions, the availability of private health insurance, and, among single mothers on welfare, the quantity and quality of available spouses. Separate models are developed for five groups of Medicaid participants: (1) single-parent families, (2) two-parent families, (3) blind or disabled children, (4) blind or disabled adults, and (5) immigrants. In the second component, referred to as the supply-side analysis, exits from the Aid to Families with Dependent Children (AFDC) into work, marriage or the Supplemental Security Income (SSI) program are modeled as a function of maternal and child health status. Additionally, several unique variables are included to measure how participants expectations of private health insurance coverage or continued Medicaid coverage affect their decisions to work. In both the demand-side and supply-side analyses, results are estimated using a discrete duration model in which the monthly exit probability is a function of the independent variables of interest, demographic characteristics, and fixed time and county effects. To conduct the analysis, several unique administrative databases on public assistance participation in California from 1987-1995 will be linked. These data were compiled as part of the California Work Pays Demonstration Project. Additionally, data on public assistance participation will be merged with extracts on the availability of health insurance from the Current Population Survey, quarterly Unemployment Insurance (UI) reports in California, known as 202 data, and California's Medicaid payments data. The overall analysis tests several assumptions embedded welfare reform and significantly extends knowledge of the interdependencies between Medicaid and the various public assistance programs that feed into the Medicaid program. The analysis will help improve models of projecting Medicaid expenditures and caseloads and projecting work participation by those on public assistance. Finally, the analysis will help policy- makers simulate how changes in the private health insurance market will affect Medicaid caseloads.
