Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
ALTERNATIVE MODEL DESIGNS: PROGRAM PARTICIPATION FUNCTIONS AND THE ALLOCATION OF ANNUAL TO 107 MONTHLY VALUES IN TRIM2, MATH, AND HITSM number of eligible persons in the March CPS. Participation rates vary by benefit amount (four categories) and whether the person is aged or disabled and single or married. These participation rates are calibrated to replicate the number of participants by state. Eligible people reporting SSI benefits in the March CPS are selected to participate first. Other eligible people are selected based on their benefit amount, unit type, and marital status. Comment The SSI participation functions are quite similar across all three models. TRIM2 develops SSI participation rates for the most detailed set of categories. State-level calibration is automatic in HITSM and optional in TRIM2 and MATH. AFDC Participation Functions TRIM2 In generating a baseline file the AFDC module determines participation for AFDC-eligible units in three steps (see Giannarelli [1989a] for a complete description). In the first step the module evaluates a probit function for each unit simulated to be eligible for AFDC for at least one month in the year. The probit includes explanatory variables for annual simulated benefit amount; the head's marital status, age, race, disability status, education, and employment status; presence of children under 6; absence of household members with earnings who are not part of the assistance unit; whether the household is in a state that has the unemployed-parent program; and whether the unit passed the full needs standard (net income test) in a majority of the months in which it is eligible. (See Figure 1 for the variable specifications and coefficients of the probit. Note that whether or not the unit reported AFDC participation in the March CPS is not one of the variables.11) In the next step the probit index is converted to a participation probability for each unit, the AFDC module is run, and the output is examined to determine how closely the simulated counts of participants match administrative data for the following control totals: average monthly AFDC recipients by state, average monthly AFDC families in the unemployed-parent component of the program for the nation, and the national percentage of AFDC families with earnings. If 11 However, eligible AFDC reporters tend to be selected to participate at higher rates than nonreporters. For example, based on tabulations of the 1987 baseline file prepared by the Urban Institute for the Panel to Evaluate Microsimulation Models for Social Welfare Programs, 86 percent of eligible AFDC reporters were chosen to participate, compared with 68 percent of eligible reporters who did not report receiving AFDC to the CPS. Currently, the Urban Institute is developing a new AFDC participation function that will always allow eligible AFDC reporters to participate in the baseline AFDC simulation and that will also provide automatic alignment to administrative totals.
ALTERNATIVE MODEL DESIGNS: PROGRAM PARTICIPATION FUNCTIONS AND THE ALLOCATION OF ANNUAL TO 108 MONTHLY VALUES IN TRIM2, MATH, AND HITSM the alignment is not close (as is typically the case), values for two additional parameters are specified for the probit. The PARTADJ parameter includes an additive adjustment by state. The PARTADJ2 parameter includes three additional additive adjustments: the first applies to all units and is essentially an adjustment to the constant term in the probit equation, the second applies to unemployed-parent units, and the third applies to other units with earnings in more than 6 months of the year. The AFDC module is run again, and, as necessary, the values for the PARTADJ and PARTADJ2 parameters are modified until the correspondence of the output with the control totals satisfies the TRIM2 analyst. Dummy that is turned on if this is a state with the unemployed-parent program (whether or not G6+0.0714 this is an unemployed-parent unit) AFDCBN+0.0003 Annual potential simulated AFDC benefit WEDSTA+1.121 Dummy that is turned on for single-parent families HDAGE1â0.2074 Dummy that is turned on if the head is age 35 or older HEADEDâ0.3361 Dummy that is turned on if the head has completed at least one grade past high school FMEMP+0.5132 Dummy that is turned on if the head and the spouse (if present) did not work at all during the year FMDISA+0.2712 Dummy that is turned on for disabled units Dummy that is turned on if there is no one in the household who is not in the AFDC unit and SUPABS+0.1021 who has earnings CHLD6+0.2011 Dummy that is turned on if there is a child in the unit less than 6 years old HDRACE+0.2566 Dummy that is turned on if the head is nonwhite Dummy that is turned on if the AFDC unit passed the full needs standard (net income) test in a AFELIG+0.6170 majority of the months in which it was eligible (for eligibility, units are required to pass this standard only in the first month) Constantâ2.673 FIGURE 1 Probit equation for simulating the AFDC participation decision in the TRIM2 model. SOURCE: Giannarelli (1989a: Table 1). In simulating participation under alternative AFDC programs, the AFDC module uses the probit equation with the values of the two adjustment parameters that were determined in the baseline run. Hence, alternative scenarios will embody the same participation relationships as the baseline program.