Medicaid and private insurance: evidence and implications.

Abstract

Health insurance coverage of women and children in the United States over the past decade has been marked by two striking trends. The first is a dramatic increase in coverage through Medicaid—the public insurance program that covers low-income women and children, among others. The second is an equally dramatic decline in private coverage. Medicaid coverage of children rose from 15 percent in 1987 to 21 percent in 1992, but private insurance coverage declined by almost the same amount (from 77 percent to 69 percent). On net, the uninsurance rate was unchanged. For women of childbearing age (ages fifteen to forty-four), Medicaid coverage rose from 8 percent to 11 percent, but the rate of private coverage fell from 76 percent to 71 percent. On net, the rate of uninsurance rose by approximately two percentage points. The similarity between the increase in Medicaid coverage and the reduction in private coverage for these two groups leads one to ask whether these two trends are related. One natural linkage between them is the “substitution hypothesis;” that is, persons who are made eligible for Medicaid may drop their private insurance in favor of the public program. Since the average privately insured family pays more than $1,000 out of pocket for health care each year (including premium contributions and cost sharing) andMedicaid is essentially free, this is not an implausible hypothesis. The extent of insurance substitution cannot be inferred from time-series data alone. There was a sizable recession in 1990–1991, and there is always less private coverage and more public coverage during such a time. Also, there has been a long-term secular decline in private insurance coverage in the United States, related to changes in the nature of employment and employers’ views about the benefits they need to offer to attract workers. Determiningwhether there is some substitution of public for private insurance among these other trends is an important issue. Here we review our research on this question. n Methods. The major hurdle for any attempt at measuring the degree of substitution is figuring out how to differentiate between substitution and the other hypotheses noted above for changes in insurance coverage. There is no “correct” way to do this. If this were a controlled trial, we would randomly make some persons eligible for public insurance and others ineligible.We then would examine how many of those eligible for public insurance moved from private to public insurance and compare that with the change in private insurance in the control group. Of course, no experiment of this form has