Analysis of Children's Health Insurance Patterns: Findings from the SIPP

This paper summarizes work using the 1992 panel of the Survey of Income and Program Participation (SIPP) to provide the most detailed look yet at the dynamics of health insurance coverage among children and the relationship between Medicaid eligibility and insurance coverage. We summarize what our technical work (contained in companion technical appendices A-D) says on five broad questions:

1. How many children lack insurance?
2. How long do children remain uninsured?
3. What are the patterns of movement between spells with and without insurance?
4. What are the characteristics of children by type of health insurance coverage?
5. How many fewer uninsured children would there be if participation in Medicaid were more complete?

We find that from October 1992 to September 1994, between 12.4 and 13.3 percent of children under 19 were uninsured at any one point in time, with one in four children (27.1 percent) uninsured for some period of time over the two years. About half of all spells without insurance last less than six months, but a quarter last 12 months or more. Further, nearly half of all children uninsured at any one time were already uninsured for 12 months or more.

Among children becoming uninsured, 52 percent were covered by employer-sponsored coverage in the previous month, 39 percent by Medicaid, and 9 percent by other sources or were newborns. More than half of the 550,000 children leaving Medicaid each month (56 percent) became uninsured. But loss of Medicaid eligibility explains fewer than half these transitions. While reporting error may explain some of this, our findings highlight the importance of shifting some attention from studying why children do not enroll in Medicaid to the question of why children lose Medicaid coverage when they may still be eligible. This is particularly relevant as our findings highlight the sensitivity of insurance coverage rates to Medicaid enrollment patterns.

The 1992 SIPP data support previous research on the relationship between children's insurance status and various demographic and socio-economic characteristics. The findings highlight how Medicaid eligibility policy affects the relationships among income, child age, and coverage. They also highlight the periodic discrepancy between a child's coverage and that of the parents. About one-fifth of uninsured children appear to have at least one insured parent, although some of this may be reporting errors. Children who experience long spells without insurance do not appear strikingly different from those who experience relatively brief spells, contrary to what some may expect.

Our data suggest that Medicaid participation rates by children range from 65 percent to 79 percent, depending on how they are calculated. Simulations indicate that about one third of the 8.9 million uninsured children in September 1994 were eligible for Medicaid; some of this group is in transition to Medicaid, however, and many more were enrolled previously. The findings also highlight the substantially lower participation rates for children who are not cash assistance participants. This reinforces the importance of strong outreach to achieving high levels of CHIP participation and highlights ways of targeting outreach to reach diverse subsets of children.

In sum, our work suggests that the number of children with some spell without insurance over time greatly exceeds the number uninsured at any point in time. The findings highlight the importance of not just enrolling children in Medicaid or CHIP but also retaining them in the programs. They also show that, contrary to common perception, Medicaid participation by eligible children with no other coverage was relatively high before welfare reform. These rates were lifted by the high proportion of children covered through cash assistance programs, reinforcing the concern over the potentially adverse effects of TANF on Medicaid coverage for children.