THE ESTIMATION OF FOOD STAMP SELF-SELECTION MODELS USING THE METHOD OF SIMULATION

Much research sponsored by the Department of Agriculture's Food and Nutrition Service (FNS) has evaluated the effects of the Food Stamp Program (FSP) and other food and nutrition programs on outcomes of interest (e.g., dietary intake or food expenditures). Of course, the problem of "self-selection" frequently arises in evaluations of assistance programs in general and in analyses of food and nutrition programs in particular. This report addresses the phenomenon of self-selection in the context of multiple programs. For example, to study the effect of the FSP on the food expenditures of households headed by a single woman, one must also control for the effects of Aid to Families with Dependent Children (AFDC), since so many female heads receive both food stamps and AFDC. As we will show, recently developed techniques of "simulation estimation" make it feasible to deal with selection in the context of multiple programs.