The main findings, which are estimated using minimal controls for other potential determinants of food stamp receipt, confirm that different types of households were affected in different ways by the economy and policy changes. The economy has an especially strong effect on caseloads from multiple adult households with children and adults living separately. Shorter recertification periods also reduce food stamp caseloads from these two types of households, which include many working poor food stamp participants. TANF sanctions reduce caseloads from households with children, and Electronic Benefits Transfer (EBT) systems increase caseloads.

The main findings indicate that the economy explains at least 19 percent of the total caseload decline from 1994 to 1999, FSP reporting requirements explain another 8 percent of the decline, and several measures of specific components of TANF plans together account for another 21 percent of the decline. Restricted eligibility for noncitizens and adults without dependents could account for perhaps 10 percent of the decline. These main findings should, however, be viewed with caution because the estimated effects of TANF are sensitive to the inclusion of additional controls for other factors that may also influence caseloads, and because some of the estimated effects of TANF policies persist among households that do not include children. These findings show that it is not easy to separate the effects of policy changes and other factors on caseloads trends in the late 1990s. The results nevertheless indicate the recent policy changes may account for some of the recent food stamp caseload decline.