The analysis described above uses the 2009 ASEC to provide estimates of health insurance coverage and Medicaid/CHIP eligibility for the CS eligible population in 2008. Every other year in April, the Census Bureau fields an additional supplement to the CPS in order to obtain information on custodial parents and their child support. The Census Bureau creates a data file covering households that are in sample for both the March and April supplements. Income, health insurance, and other data from the March ASEC are combined with the child support supplement data from the April supplement, producing the CPS Child Support Supplement (CPS-CSS) data file. While the CPS-CSS can be used to produce estimates of the uninsurance rate for CS eligible children, there are several elements that make the March ASEC supplement preferable for our purposes. Due to the CPS sample design, the CPS-CSS has less than half the sample size of the ASEC and the matched sample must be re-weighted to meet US population totals. Using the ASEC alone avoids this loss of sample as well as any additional measurement error brought in by the re-weighting process. Furthermore, the Urban Institute’s Medicaid/CHIP eligibility simulation model has been designed for and tested on the ASEC. Adapting it to the CPS-CSS would require additional modifications and robustness checks. Finally, in the interest of producing more regular estimates for this population, the ASEC is fielded each year and the results are available each fall. The April supplement however is fielded every other year and the necessary merging and re-weighting further delays file availability.
The main drawback to using the ASEC alone is a lack of detail necessary to precisely identify the child support eligible population. The estimates from the 2008 CPS-CSS, which includes a series of questions designed to more accurately define this population, find a total of 20.6 million CS eligible children. Using the 2008 ASEC alone, we estimate a total of 23.7 million CS eligible children. The discrepancy in the count of CS eligible children is likely due to our inability, using the ASEC, to exclude from the eligible population those children who live with only one of their parents, but who, for a variety of reasons, do not have another parent living outside the household. For example, approximately 2.5 million children have been excluded from the eligible population on the CPS-CSS because their non-custodial parent has died, terminated their parental rights, or is no longer considered a parent by the custodial parent. Such children cannot be identified and excluded on the ASEC. These children account for the bulk of the excess eligible population identified on the ASEC.
In order to explore the possible bias resulting from our eligibility definition, we examined the income and coverage distribution on the ASEC versus that on the CPS-CSS. On the ASEC, 65 percent of CS eligible children are found to have incomes below 200 percent of the FPL and 11.9 percent are uninsured while on the CPS-CSS, 59 percent fall below the same income threshold and 12.3 percent are uninsured using a comparable coverage definition. On the basis of these similar income and coverage estimates, we conclude that our estimates of the proportion of CS eligible that are uninsured or Medicaid/CHIP eligible are not severely biased due to our sample of CS eligible children.
Due to the extra CS eligible children identified on the ASEC, estimates that report the number of CS eligible children in a given population, as well as those that estimate the CS eligible population as a proportion of all uninsured children, will likely be inflated however. The CPS-CSS estimates 20.6 million CS eligible children, but excludes children living with neither of their biological parents from the CS eligible population. Such children are generally eligible for child support however and these children are included in the estimates from the ASEC. We include children under 15 who live without either of their parents and they account for 2.2 million CS eligible children. Adding these children to the 23.7 million children living with at least one biological parent, results in a grand total of 25.9 million CS eligible children on the ASEC.
If we add the 2.2 million children living without any parents to the CPS-CSS estimate of 20.6 million CS eligible children, this results in an estimate of 22.8 million CS eligible children on the CPS-CSS compared to 25.9 million on the ASEC. The ASEC estimates of the number of CS eligible children are therefore inflated by approximately 14 percent. For example, we estimate 2.9 million uninsured child support eligible children. The details above suggest that this number should probably be closer to 2.6 million. In the analysis above, we report the estimates from the ASEC, but include alternative estimates that reflect an adjustment for the overreporting of CS eligible children in the footnotes.