How Does Insurance Coverage Vary among Subgroups of Children?
Other research has established that uninsured children differ from insured children in terms of a variety of demographic and economic characteristics (Fronstin 1997b, Lewis et al. 1997). In our analysis, insurance coverage varied with nearly every characteristic that we examined and in ways that largely replicated earlier findings. For example, Hispanic children were more than twice as likely to be uninsured as black children, who were only marginally more likely to be uninsured than white children. Nearly three-quarters of children whose parents did not work were enrolled in Medicaid compared to about half that percentage among children whose parents were employed part-time. This latter group along with children with no parent in the household were the most likely to be uninsured, but together, they accounted for only 13 percent of all uninsured children.

While employer-sponsored insurance was the most common source of coverage, subgroups of children with comparatively low rates of employer-sponsored coverage were not necessarily more likely to be uninsured. Medicaid accounts for this situation but fills the gap more completely for some groups than others. Children in families below 50 percent of poverty, for example, have only one-fifth the employer-sponsored coverage that children between 100 and 200 percent of poverty have, but they are less likely to be uninsured. Indeed, uninsured children in families between 100 and 200 percent of poverty account for nearly 40 percent of all children without insurance, while those below 100 percent of poverty account for one-third of all children without insurance. This pattern is evidence of the need for the broad coverage that states are allowed to provide through CHIP.

How Does a Child's Insurance Coverage Compare to That of His or Her Parents?
Children's coverage tends to mirror that of their parents, but the exceptions to this rule pose interesting questions for policymakers and analysts. About one-fifth of uninsured children appear to have at least one insured parent--a situation that merits further research to establish if and why it is true. Between 7 and 8 percent of uninsured children, for example, have a parent who reports being covered by Medicaid--a fraction that appears much too high and suggests that many of these children may in fact be covered by Medicaid, contributing to the aforementioned Medicaid undercount . At the same time, between 10 and 13 percent of the children who were reported to have been covered by Medicaid in the study period appeared to have an uninsured parent. This finding is very plausible in light of the opportunity for many children to be enrolled in Medicaid under child-only provisions. In fact, we would very likely find an even greater proportion of Medicaid children with uninsured parents if the Medicaid participation rates of children eligible under the various child-only provisions were as high as they are in the eligibility categories with the greatest participation. This situation raises the policy question--to which we will return--of whether the Medicaid enrollment of children could be increased by extending eligibility to parents as well.

How Does a Child's Age Influence the Chances of Being Uninsured or on Medicaid?
It is important to measure the relationship between age and health insurance coverage because age differences in coverage reflect experiences over the life cycle. Both the probability of being uninsured and the average time without insurance increase with age--a finding established by longitudinal analysis. To generate findings like this, we followed children over the entire year at a given age. Between infancy and late adolescence, the probability of being uninsured during a year rose from 16 percent to 27 percent. Among those who were uninsured for any length of time, only 10 percent of infants versus 38 percent of 18-year-olds were uninsured for the entire year.

The likelihood of being both uninsured and ineligible for Medicaid increases even more sharply with age than does being uninsured alone. Only 6 percent of infants compared with a full 24 percent of 18-year-olds were ever uninsured and ineligible for Medicaid during the year. Furthermore, while only 5 percent of those infants who were ever uninsured and Medicaid-ineligible remained so for the entire year, nearly one-third of the 18-year-olds in the same situation remained so for the entire year.

Medicaid coverage declines sharply with age. Age differences in Medicaid participation are to a large degree a reflection of eligibility criteria that progressively favor younger children.¹ Our findings highlight the very prominent role of Medicaid among infants and how that role diminishes as children grow older. Nearly 40 percent of all infants were covered by Medicaid for some part of their infancy, and nearly two-thirds of them, or 25 percent of all infants, were covered for the entire year. With increasing age, the probability that a child was ever covered by Medicaid during the year declined gradually to about 15 percent or less, while the proportion of Medicaid children who were covered for the entire year fell to about one-third. In contrast to the 25 percent of all infants who were covered by Medicaid for the entire year, only 10 percent of 10-year-olds and less than 5 percent of 18-year-olds were enrolled for the entire year. This latter differential is likely to diminish over time as Medicaid eligibility under the poverty-related criteria is phased in and Medicaid expansions are introduced under CHIP.

What Characteristics Distinguish Uninsured Children Who Are Medicaid-Eligible?
In general, uninsured children who are Medicaid-eligible resemble Medicaid participants, which supports the contention that many of these children are either in transition to Medicaid or are enrolled but have not reported their coverage. However, these two groups of people differ notably from each other in four areas:

• Racial Composition. Uninsured Medicaid-eligible children are twice as likely to be white as black, whereas the proportions among Medicaid participants are nearly equal. Hispanic children are only slightly more represented among the uninsured Medicaid-eligibles than among Medicaid participants.
• Family Structure. Children from two-parent families account for more than half of the Medicaid-eligible uninsured whereas they are one-third of Medicaid participants.
• Poverty Status. Children in families between 100 and 200 percent of poverty make up one-third of the Medicaid-eligible uninsured compared to one-fourth of Medicaid participants.
• Parents' Employment Status. Children with a parent employed full-time make up half of the Medicaid-eligible uninsured but only 30 percent of Medicaid participants. The reverse is true of children whose parents are not working.

Do the Characteristics of Uninsured Children Vary by Spell Duration?
Differences in the characteristics of children with long versus short spells without insurance are also potentially important to the design of programs as well as targeted outreach. However, children who experience long spells without insurance do not appear to be strikingly different from children who experience relatively brief spells without insurance. Researchers have found very strong differences between children with and without insurance coverage, but once children lose their coverage, there appears to be little that differentiates between those who remain uninsured for long periods of time and those who become re-insured fairly quickly. Children with long spells were somewhat more likely to be Hispanic and to live with both parents, but no other notable differences emerged. On the whole, these findings give us little information to use in designing programs or outreach.

Notes
1. For example, under provisions included in the Omnibus Budget Reconciliation Act, states were required to cover infants and children under age 6 in families with incomes below 133 percent of the federal poverty level and to cover older children born after September 30, 1983 in families below 100 percent of poverty. Many states exercised an option to cover infants up to higher income levels, and a smaller number extended the income limits for older children as well. In addition, families with younger children were more likely than families with older children to have an income below the levels that qualified them for Aid to Families with Dependent Children (AFDC) and thus were more likely to obtain Medicaid coverage through that route.