adjustments to participation in order to meet state-by-state targets. Future plans call for the development of a probit participation function for food stamps, and hence the parameter PARTADJ was designed as an additive adjustment to a probit index. During the initialization phase of the FSTAMP module, the probabilities in PCTELD and PCTNOELD are converted to probit indices. To determine a particular unit’s participation decision, the PARTADJ value for the unit’s state is added to the converted PCTELD or PCTNOELD value, and the result is converted back to a probability that is compared with the unit’s random number. The current TRIM2 food stamp participation function was implemented in 1988 and improves upon the prior methodology by varying the participation rate with the benefit level—units eligible for higher benefits are more likely to participate.
MATH
The FSTAMP module determines participation in the food stamp program separately for a large number of demographic and income groups, on the basis of the number of participants with those characteristics found in administrative data. Eligible households are selected to participate on the basis of the following characteristics: household size, whether the household has an elderly member, gross monthly income, whether the household was simulated to receive public assistance (including AFDC, GA, or SSI), the ratio of the food stamp allotment to the household’s poverty cutoff, and whether the household reported receiving food stamps in the March CPS. When necessary, each of these variables is transformed into several discrete values, so that a large number of categories is formed.
The FSTAMP module must calculate the participation rates for various groups in order to determine who participates. It does this by counting the number of eligible households in several categories (household size, public assistance recipiency, presence of an elderly person, and gross income level), and comparing these counts to administrative totals of participants in the same categories. If the number of participants in any category exceeds the simulated number of eligible households, the module reallocates some participants to other cells. This process is repeated until participation rates are less than or equal to 1 for all cells.
The module then expands the participation matrix so that each of the subgroup participation probabilities also varies by the household’s food stamp amount relative to the poverty cutoff and by whether the household reported receiving food stamps. The variation by food stamp benefit ratio is derived from an analysis of food stamp program participation in the ISDP by Czajka (1981). The probabilities are chosen to be consistent with Czajka’s findings on participation rates at various benefit-to-poverty-line ratios and, at the same time, to maintain the original participation rate for the subgroup. This procedure
