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.
1 Advisers to the Nation on Science, Engineering, and Medicine Division on Behavioral and Social Sciences and Education Committee on National Statistics May 16, 2002 Mr. Jay Hirschman Director, Special Nutrition Services Office of Analysis, Nutrition and Evaluation Food and Nutrition Service U.S. Department of Agriculture Dear Mr. Hirschman: The Panel to Review the USDA Methodology for Estimating Eligibility and Participation for the WIC Program (the Special Supplemental Nutrition Program for Women, Infants, and Children) is charged with reviewing currently used and alternative data and methods for estimating the number of people nationwide who are eligible and likely to participate in WIC. The panel gave its initial assessment of the methodology in its Phase I report, Estimating Eligibility and Participation for the WIC Program. A major conclusion of the report was that estimates of the number of infants and children who are eligible for WIC are substantially understated because current U.S. Department of Agriculture (USDA) methods do not fully account for those people whose incomes are above the WIC income eligibility threshold but who are adjunctively eligible for WIC because they participate in Medicaid, food stamps, or Temporary Assistance for Needy Families (TANF). This conclusion was supported with estimates of the number of infants and children who would be eligible for WIC because they participate in one of these programs. Specifically, the estimates showed that 45 percent more infants and 21 percent more children would be adjunctively eligible than estimated by current USDA methods. I am writing to call your attention to a problem that we found in these estimates while extending the analysis conducted for the Phase I report through additional years. The problem resulted from the data used to simulate Medicaid participation. The estimates of the additional numbers of persons eligible for WIC when adjunctive eligibility is accounted for in the estimation methodology are not as large as reported in the Phase I report. New estimates with corrected data show that accounting for adjunctive eligibility increases the number of infants eligible for WIC by 31 percent and the number of children eligible for WIC by 14 percent. Although the point estimates of the additional numbers of infants and children eligible for WIC through adjunctive eligibility were not as large as initially shown in the Phase I report, the revised estimates still show a substantial increase in the number of infants and children eligible for WIC through participation in Medicaid, TANF, or food stamps. The problem with the Phase I estimates does not affect the panelâs conclusion that failing to account for adjunctive eligibility results in an underestimation of the number of eligible people. This letter details the problem
2 found with the Phase I data, explains the effect it has on the point estimates of the size of the underestimation, and discusses the implications for the panelâs conclusions in the Phase I report. Current methods for estimating WIC eligibility make only a minor adjustment to account for persons who are adjunctively eligible for WIC. Differences in eligibility rules between WIC and Medicaid, food stamps, and TANF make it possible for an applicant whose family income is above the WIC income eligibility threshold (185 percent of federal poverty guidelines) to be eligible for WIC because he or she participates in at least one of the other three programs. The most notable differences in eligibility rules are those between the WIC and Medicaid programs. In several states, the Medicaid program has a higher income eligibility threshold than the WIC income threshold. In addition, Medicaid uses a net income measure to determine eligibility, allowing certain deductions and income disregards, while WIC only considers gross income. These allowable deductions and disregards also make it possible for someone to have a gross income above the WIC income threshold but a net income below the Medicaid threshold. In the Phase I report the panel examined how many additional infants and children would be eligible for WIC if estimates accounted for adjunctive eligibility. To estimate the number adjunctively eligible, the panel obtained data from the Urban Instituteâs Transfer Income Microsimulation Model 3 (TRIM). Data from TRIM were used instead of the data currently used by USDA to estimate eligibility, the March Income Supplement of the Current Population Survey (CPS), because the CPS suffers from underreporting of participation in Medicaid, food stamps, and TANF. For example, the 1999 March CPS (which asks questions about a familyâs income in calendar year 1998) only captured 68 percent of the Medicaid caseload, 67 percent of the food stamps caseload, and 61 percent of the TANF caseload according to 1998 administrative records from these programs (Wheaton and Giannarelli, 2000). TRIM starts with the public-use version of the CPS data, but attempts to correct for underreporting of participation in each of these programs by simulating participation. TRIM corrects for the underreporting of Medicaid by randomly selecting persons who do not report Medicaid participation, but whose incomes and other characteristics indicate they are eligible for Medicaid. The number of eligible people randomly selected as âparticipantsâ is computed so that the total number selected matches caseload totals from administrative records for each state, by broad eligibility categories. In standard practice, TRIM simulates Medicaid participants to match average monthly control totals of participants. For the Phase I report (National Research Council, 2001), the panel used 1998 income data from TRIM (from the 1999 March CPS) to estimate the numbers of infants and children who would be eligible for WIC if adjunctive eligibility were accounted for in estimating the number of infants and children eligible for WIC.1 Estimates of the numbers of infants and children who met income eligibility requirements for WIC were calculated using the USDAâs current methods. These âbaselineâ estimates were compared with estimates of infants and children who met income eligibility requirements for WIC or who were enrolled in Medicaid, food stamps, or TANF. As noted above, the panel found that accounting for adjunctive eligibility increased the number of infants eligible for WIC by 45 percent over the USDA baseline estimates and the number of children eligible for WIC by 21 percent. Based on these estimates, the panel concluded that âNot fully accounting for adjunctive eligibility results in a 1The panel focused only on infants and children because estimates of the number of eligible women are derived from the estimates of eligible infants.
3 substantial underestimation of the number of people eligible for WICâ (National Research Council, 2001, p. 40). For Phase II of the study, the panel again obtained TRIM data from 1998, as well as from the years 1994, 1996, and 1999. The additional years were obtained to determine whether the results regarding adjunctive eligibility and other findings in the Phase I report from 1998 were consistent across time. In an attempt to replicate the results used in Phase I of the study, the panel discovered that the two versions of 1998 data did not match, resulting in different estimates of the additional numbers of infants and children adjunctively eligible for WIC. Panel and staff asked the Urban Institute, which provided both sets of data, to investigate why the two sets of data covering 1998 did not match. In doing so, the Urban Institute discovered an error in the original 1998 data given to the panel for the Phase I report. The error was due to the use of the incorrect Medicaid control totals in the Medicaid participation simulation. Instead of using the average monthly control totals for 1998 as a target for how many individuals were to be simulated as Medicaid participants, the simulation used control totals that included anyone ever enrolled in Medicaid in 1998. The latter number is necessarily a larger number than the former: if someone enrolled in Medicaid for only one month, he or she would be included in target controls, while the average monthly control totals equal the number reported to be enrolled in each month averaged over the year. The average monthly control totals are more appropriate because they better reflect the actual number enrolled in a month. Monthly numbers are important because the Medicaid simulations are conducted on a month-by-month basis, not on a yearly basis. As a result of using the ever-enrolled control totals, the estimates of the numbers of infants and children adjunctively eligible for WIC reported in the Phase I report are overstated.2 The panel reestimated the number of infants and children adjunctively eligible for WIC with the correct data and compared them with estimates in the Phase I report. Table 1 shows these two estimates for infants and Table 2 shows these two estimates for children. In each table, Phase I data refer to the TRIM data used in the Phase I report (with control totals based on everenrolled in Medicaid), and Phase II data refer to the recently obtained TRIM data using the correct control totals based on average monthly enrollment in Medicaid.3 As shown in Table 1, the TRIM data used in the Phase I report estimated 671,000 infants to be adjunctively eligible for WIC through participation in Medicaid, food stamps, or TANF, which is a 45 percent increase in the baseline estimates of WIC eligibility. The corrected Phase II data show that 458,000 infants are estimated to be adjunctively eligible for WICâa 31 percent 2 Two alternative measures of the number adjunctively eligible for WIC were also considered in the Phase I report. These measures-the number eligible for Medicaid and the number eligible for Medicaid but not already covered by private insurance-are not affected by the incorrect data because they are based on Medicaid eligibility and not simulated participation. 3 The Phase I and Phase II baseline estimates differ slightly because the income definition used in the Phase I data included the simulated value of food stamps, TANF, and Supplementary Security Income (SSI) received as income. Current USDA methods for estimating eligibility do not include the value of food stamps as income. The value of food stamps should not have been included in the income definition used with the Phase I data. The correction was made with the Phase II data. TANF and SSI benefits are counted as income in determining eligibility, but the Phase II data include reported benefits, not simulated benefits.
4 increase over the baseline estimate. This is a smaller increase in eligibility estimates than was initially reported, however, it is a sizable increase in the estimated total number of eligible infants and children. Similarly, as shown in Table 2, the TRIM data used in the Phase I report estimated 1,333,000 children to be adjunctively eligible for WIC, which is a 21 percent increase in the baseline estimates of WIC eligibility. The corrected Phase II data show that 865,000 children are estimated to be adjunctively eligible for WIC, a 14 percent increase. Results obtained using the corrected TRIM data suggest that WIC eligibility estimates would increase substantially if adjunctive eligibility were taken into account. Therefore, based on the corrected TRIM data, the panelâs conclusion-the current FNS practice of not accounting for adjunctive eligibility understates eligibility-is still supported. Some infants and children who were estimated to be eligible for WIC because they enrolled in Medicaid may also have been estimated as eligible for WIC if a monthly income measure had been used instead of an annual income measure. Estimates of the numbers of additional infants and children eligible for WIC using a monthly income measure were also provided in the Phase I report and are reported in Tables 1 and 2. These monthly estimates, which are unaffected by the incorrect Medicaid simulations, also showed that the numbers eligible for WIC would increase substantially if a monthly income measure was used to measure income eligibility. The estimates also showed that the effects of monthly income and adjunctive eligibility are not strictly additive but, rather, interactive. Some of those simulated to be enrolled in Medicaid were also estimated to be eligible for WIC because they had at least one month of income below the WIC eligibility threshold (i.e. some of those mistakenly simulated as enrolled in Medicaid may have also gained eligibility because their monthly incomes were below the WIC eligibility threshold). Table 1 shows the number of infants in families with incomes below the WIC income threshold for at least 1 month during the year. Monthly income measures were simulated using TRIM. If an infant or child lived in a family with at least 1 month of income below the WIC eligibility threshold, this infant or child was considered eligible for WIC.4 Use of the monthly income measure increases the number eligible by 25 percent using the Phase I data and 24 percent using the Phase II data. The small difference is due to a small change in the baseline estimates, as explained in the note to Table 1. When both adjunctive eligibility and monthly income are considered, the total number of eligible infants increases by 43 percent using the corrected Phase II data. Although less than the 51 percent using the incorrect Phase I data, the total still represents a large increase in the estimated number of eligible infants. Table 2 shows the number of children who live in families with incomes below the WIC threshold in at least one month of the year. Use of the monthly income estimate increases the number of children eligible for WIC by 21 percent using Phase I data and by 20 percent using Phase II data. Again, the difference is due to a small difference in the baseline data. When both monthly income and adjunctive eligibility are considered, the increase in the estimated number of eligible children is 27 percent using corrected Phase II data compared with 32 percent using the incorrect Phase I data. 4Estimates reported in this letter do not account for the 6-month certification period for children. Estimates provided in the Phase I report and those that will be provided in the Phase II report consider the 6-month certification period for children.
5 In the Phase I report, the panel also concluded that, overall, current methods used to estimate WIC eligibility substantially underestimate the number of people eligible for WIC. This conclusion was based in part on the panelâs estimates of adjunctive eligibility, but also on other methodological issues that the panel identified as contributing to an understatement. Specifically, the panel also found that current assumptions about the accounting time period for measuring income and the definition of an economic unit for determining eligibility, along with the underestimation of infants in the CPS, contribute to an overall underestimate of WIC eligibility. Although the incorrect data used to estimate the numbers of Medicaid enrollees used in the Phase I report led to an overstatement of the size of this underestimation, the general conclusion that current methods result in an underestimation of eligibility for WIC is still supported. The results with the corrected 1998 data will be included in the panelâs final report. The panel will use the TRIM data from other years (1994, 1996, and 1999) to estimate the effect of accounting for adjunctive eligibility on WIC eligibility estimates. The panel is also conducting a more detailed analysis of the TRIM simulations of Medicaid participation and will examine the stability of the estimates of adjunctive eligibility over time and the characteristics of those who are simulated as Medicaid participants. These analyses will be conducted as a further check of the results reported in Phase I. The panel regrets the error reported in the Phase I report. I invite you to contact me if you have any questions about the data or the panelâs work. Sincerely, David Betson Chair, Panel to Evaluate the USDA Methodology for Estimating Eligibility and Participation for the WIC Program cc: David Smallwood, Economic Research Service, USDA
TABLE 1 Estimates of the Number of Infants Potentially Eligible for WIC: Comparison of Results Reported in the Phase I Report and Corrected Results (Numbers reported in 1,000s) Estimate Phase I Data Phase II Data Baseline estimatesa 1,475 1,492 Number of additional infants eligible for WIC through adjunctive eligibility alone 671 458 Percentage change in baseline due to including those adjunctively eligible 45% 31% Number of additional infants eligible for WIC if monthly income alone is counted 370 364 Percentage change in baseline due to including those eligible based on use of a monthly income measure 25% 24% Number of additional infants eligible for WIC through either adjunctive eligibility or monthly income 755 640 Percentage change in baseline due to including those with either adjunctive eligibility or monthly income eligibility 51% 43% aThePhase I and Phase II baseline estimates differ slightly because the income definition used in the Phase I data included the simulated value of food stamps, TANF, and Supplementary Security Income (SSI) received as income. The Phase II data do not include the value food stamps and include reported TANF and SSI income. See text footnote 3. 6
TABLE 2 Estimates of the Number of Children Potentially Eligible for WIC: Comparison of Results Reported in the Phase I Report and Corrected Results (Numbers reported in 1,000s) Phase I Data Phase II Data Baseline estimatesa 6,307 6,375 Number of additional children eligible for WIC through adjunctive eligibility alone 1,333 865 Percentage change in baseline due to including those adjunctively eligible 21% 14% Number of additional children eligible for WIC if monthly income alone is counted 1,305 1,274 Percentage change in baseline due to including those eligible based on use of a monthly income measure 21% 20% Number of additional children eligible for WIC through either adjunctive eligibility or monthly income 1,999 1,735 Percentage change in baseline due to including those with either adjunctive eligibility or monthly income eligibility 32% 27% aThe Phase I and Phase II baseline estimates differ Slightly because the income definition used in the Phase I data included the simulated value of food stamps, TANF, and Supplementary Security Income (SSI) received as income. The Phase II data do not include the value food stamps and include reported TANF and SSI income. See text footnote 3. Figures presented in this table do not account for the 6-month certification period for children but instead, assume a year-long certification period. 7
