How High Is Medicaid Participation?
With no adjustment for the Medicaid undercount in the SIPP, the reported participation in Medicaid is 65 percent among children who we simulated to be eligible.¹ If we exclude from the denominator those eligible nonparticipants who reported some other source of coverage, the participation rate rises to 79 percent. With an undercount adjustment, the 65 percent figure would rise to about 75 percent.

What Proportion of Uninsured Children Are Eligible for Medicaid?
Our simulation of Medicaid eligibility revealed that 2.9 million, or 33 percent, of the estimated 8.9 million uninsured children in September 1994 were eligible for Medicaid. This is comparable to what other researchers have reported with more limited Medicaid eligibility simulations (Lewis et al. 1997).² With longitudinal data, however, we could also address the question of what happens to these Medicaid-eligible uninsured children over time; and what we found suggests that policymakers need to think about the Medicaid-eligible uninsured differently than they may have in the past.

Our findings do not support the inference that all 3 million of these children would be enrolled in Medicaid if only Medicaid outreach were more effective. First, much of this group is in transition--including transition to Medicaid. Between one month and the next about 20 percent of the Medicaid-eligible uninsured children completed the transition, with roughly equal shares enrolling in Medicaid, obtaining employer-sponsored or other coverage, or experiencing a rise in family income that made them ineligible for Medicaid. Within five months, more half of the initial 2.9 million Medicaid-eligible uninsured children had left that state. The mobility among the Medicaid-eligible uninsured suggests that while all of these children could theoretically be brought into Medicaid, or brought into Medicaid more quickly, it is useful to distinguish between those who would remain eligible and otherwise uninsured for only a few months and those who would remain in that state for a much longer period. Second, 25 percent of the children who transition in and out of Medicaid eligibility and between 40 and 50 percent of those who remain Medicaid-eligible and uninsured for longer periods were actually enrolled in Medicaid previously. Thus, there is a sizable group of uninsured and Medicaid-eligible children for whom it may be more relevant to ask why have they left Medicaid than why have they not enrolled. Better Medicaid outreach is not the answer to removing them from the uninsured.

In 1994, about one million children remained uninsured for a year or more despite being eligible for Medicaid, and the number is likely to have grown since then. Policymakers would do well to focus on this group, whose persistence in this condition suggests that they are hard to reach and so will require extraordinary outreach techniques if they are to be brought into Medicaid. The demographic and socioeconomic characteristics that we examined shed little light on what differentiates this group of children from those who more quickly become insured or Medicaid-ineligible. Ultimately, understanding why children remain uninsured and Medicaid-eligible for a year or more may require different kinds of data than our surveys can provide, but we have not exhausted what longitudinal surveys like the SIPP can tell us.

What Do Medicaid Participation Rates Tell Us about Strategies for Outreach?
Medicaid participation rates among children vary widely by the basis of eligibility. In September 1994, we observed the following participation rates:

• 100 percent among AFDC participants, who are automatically enrolled
• 75 percent among reported SSI recipients and foster children
• 54 percent among children in families that appeared to be eligible for AFDC but did not report that they were receiving cash assistance
• 37 percent among children eligible under the poverty-related expansions
• 23 percent among other eligibility groups³

The overall pattern of participation and the nature of changes in the cash assistance caseload post-welfare reform carry implications for CHIP. First, it is clear that outreach efforts will have to be very strong to achieve high participation rates in both the Medicaid expansions and the state programs. In the absence of strong outreach, we might expect to see participation rates that are no better than what we observed in the populations eligible for Medicaid without cash assistance. Second, to the extent that CHIP extends eligibility to families and children who were not eligible previously, participation rates could be even lower than we have seen for non-cash families and their children historically. Many newly eligible families may not perceive that a public program is truly for them; they may not even be aware that they are eligible and, lacking experience with public programs, may not think to inquire. They may also hesitate because of the stigma that they believe accompanies participation in such programs. Third, different outreach strategies may prove to be differentially effective with different eligibility groups and different subgroups of the population. At a minimum, strategies should be designed to deal with three different types of eligibles, who will be found in different proportions in different eligibility groups and population subgroups:

• Children who have been eligible for Medicaid but resisted participation
• Children who have already participated
• Children who are becoming eligible for the first time

Another factor that plays into this issue of outreach is the role of coverage for the parent. The differences in participation by eligibility group provide at least a suggestion that children are more likely to participate if a parent is eligible as well. Some states have begun to consider strategies that would allow them to use CHIP funds to partially fund coverage for the parents of eligible children. Further research into the importance of parents' access to coverage in determining their children's likelihood of participating in public insurance programs is clearly in order and may lend additional support to efforts intended to increase the coverage that is available to parents.

Notes
1. Our simulation did not encompass all categories of eligibility; nor could it take account of the full range of income disregards that states might apply to individual cases. For this reason, we do not include in our participation rate those children who were reported as Medicaid participants but were simulated to be ineligible.

2. While the level of detail in our simulations and the use of monthly income would tend to raise estimates of eligibility relative to many of these other efforts, the SIPP obtains more complete reporting of Medicaid participation than most other surveys, which reduces the number of simulated eligibles who are identified as uninsured.

3. This category does not include children whose eligibility depended on their families "spending down" their income to meet thresholds specified in a state medically needy program. We did not simulate this aspect of Medicaid eligibility because data on health care expenditures are very limited in the SIPP.