Providing Insights that Contribute to Better Health Policy
Insurance Coverage & Costs Costs The Uninsured Private Coverage Employer Sponsored Individual Public Coverage Medicare Medicaid and SCHIP Access to Care Quality & Care Delivery Health Care Markets Issue Briefs Data Bulletins Research Briefs Policy Analyses Community Reports Journal Articles Other Publications Surveys Site Visits Design and Methods Data Files
Medicaid Eligibility Policy and the Crowding-Out Effect:
Did Women and Children Drop Private Health Insurance to Enroll in Medicaid?
Issue Brief No. 03
n the late 1980s and early 1990s, the federal government expanded Medicaid eligibility for children and pregnant women. By 1992, nearly a third of all children in the United States were eligible for Medicaid, and between 40 and 50 percent of women of childbearing age were eligible for Medicaid coverage for pregnancy-related services. During this period, the number of persons with employment-based insurance coverage declined, leading researchers to investigate whether Medicaid expansions have contributed to this decline -- a so-called crowding-out effect. This Issue Brief discusses research findings and the health policy implications of the crowding-out effect.
istorically, Medicaid eligibility among children and pregnant women was linked to participation in the Aid to Families with Dependent Children (AFDC) program. This policy effectively limited eligibility to single-parent households with incomes well below the poverty line.
Beginning in the late 1980s, Congress enacted a series of Medicaid eligibility expansions. By 1992, states were required to cover all pregnant women and children up to age 6 with incomes up to 133 percent of poverty. States had the option to receive federal matching funds to cover all pregnant women and infants with incomes up to 185 percent of poverty; more than half of the states chose to exercise this option. In addition, children born after September 30, 1983, to families with incomes up to 100 percent of poverty were automatically eligible for Medicaid, regardless of family composition.
Congress expanded eligibility to increase the proportion of the population with health coverage. A potential drawback to expansions in Medicaid eligibility is the possibility of crowding out private insurance coverage. Because of the crowding-out effect, in which Medicaid substitutes for what would have been private insurance coverage in prior years for some of the newly eligible, the net increase in the insured population is smaller than the gross increase. This phenomenon limits the effectiveness of public funds used to expand insurance coverage. The question debated at a seminar sponsored by the Center for Studying Health System Change was: how much smaller and does it matter?
Putting the Issue into Perspective
easuring the crowding-out effect is complicated by the difficulty of separating it from secular declines in employer-sponsored health insurance and variation in employment and Medicaid eligibility attributed to the business cycle. Increasing health care costs and changes in the structure of employment have led to a decline in the percentage of individuals with employer-based coverage. Much of the shift from employer-sponsored to Medicaid coverage during this period may be attributable to these secular declines in employer-sponsored health insurance rather than to crowding out.
Although this discussion of crowding out focuses on Medicaid, the phenomenon is not unique to it. Crowding out occurs to some degree in all social insurance programs because of the difficulties (practical and political) of targeting benefits precisely to those who most need them. For example, it has long been argued that Social Security displaces private savings. In recent years, some of the people who get a tax subsidy by putting money into an individual retirement account (IRA) would have put money away anyway. For these people, the IRA is crowding out taxable savings accounts.
The Cutler-Gruber Study
avid Cutler of Harvard University and Jonathan Gruber of the Massachusetts Institute of Technology were the first researchers to estimate a crowding-out effect associated with Medicaid expansions. In an April 1995 working paper for the National Bureau of Economic Research, they noted that the population that became eligible for Medicaid as a result of the expansions was much less disadvantaged than the population that was eligible previously. Only 20 percent of the newly eligible population had incomes below the poverty line, and nearly two-thirds already had private health insurance. The researchers also found that the expansion of Medicaid eligibility had markedly greater impact in some states than in others. For example, eligibility rose 43 percent in Texas but only 4 percent in Utah.
Using data from the Current Population Survey (CPS) from 1987 through 1992, Cutler and Gruber estimated the extent to which expanded Medicaid eligibility was offset by declines in private health insurance among the new eligibles brought about by the expansions. Their analysis compared children and women of childbearing age who were eligible for Medicaid. Using a pooled time series analysis, Cutler and Gruber relied on cross-state and cross-age variations in the magnitude and timing of the Medicaid coverage expansions to estimate the crowding-out effect and separate crowding out from secular trends in employer coverage. Based on this analysis, Cutler and Gruber concluded that approximately 50 percent of the increase in Medicaid coverage associated with the eligibility expansions was offset by a reduction in private insurance coverage -- a crowding-out rate of 50 percent.
The Dubay-Kenney Study
ther researchers have studied the crowding-out effect and have obtained results that appear to differ substantially. In October 1995, Lisa Dubay and Genevieve Kenney of the Urban Institute reported the results of two studies they had conducted, one for the Health Care Financing Administration and one for The Robert Wood Johnson Foundation. Although they used the same data, their approach differed in several ways from that of Cutler and Gruber.
Dubay and Kenney sought to control for secular trends in employer-sponsored coverage by comparing changes between 1988 and 1993 in private coverage for Medicaid-eligible children under the age of 11 with that of men aged 18 to 44, a group not eligible for Medicaid coverage. They analyzed the extent of crowding out separately for the poor (those below 100 percent of poverty) and the near-poor (those between 100 and 133 percent of poverty). Dubay and Kenney reasoned that the potential for crowding out might be greater among individuals with family incomes above the poverty line, who are more likely to have private insurance coverage.
Using this methodology rather than the regression analysis relied on by Cutler and Gruber, Dubay and Kenney found little evidence of crowding out among children or pregnant women with incomes below the poverty line. Among the near-poor, they estimated that 21 percent of the increase in enrollment in Medicaid by children under age 11 and 45 percent of the increase in enrollment by pregnant women was offset by declines in private insurance; among poor children, the rate was only 8.5 percent. Combining the results they obtained for the poor and the near-poor populations, Dubay and Kenney estimated that overall 14 percent of the increase in Medicaid enrollment for pregnant women and 12 percent of the increase in enrollment for children under age 11 was due to the crowding out of private insurance coverage.
Common Ground between the Two Studies
t first glance the results from the two studies appear very different. While both conclude that expansion of Medicaid eligibility resulted in some crowding out of private health insurance, the magnitude of the findings varies considerably. But Cutler and Grubers analysis looked at only those made newly eligible by the Medicaid expansions. This population had higher incomes than those eligible under prior law, even within the poor and near-poor categories. Dubay and Kenney, in contrast, focused their analysis on the entire Medicaid population of children under age 11 and pregnant women. Moreover, Cutler and Grubers estimate of crowding out included the indirect effects of the eligibility of other family members -- for example, those who dropped employer-based coverage who were not themselves eligible for Medicaid but who had a child who was.
When the same population is examined, the estimates of crowding out in the two studies become more similar. The fact that Cutler and Grubers estimate of crowding out included the indirect effects of the eligibility of other family members while Kenney and Dubays did not helps explain the disparity. Because much of the population affected by Medicaid expansions had incomes that were higher than those previously eligible for Medicaid, this also helps explain why the Cutler and Gruber estimates are higher than those of Kenney and Dubay, but are not necessarily inconsistent with them.
For example, estimates of the extent of crowding out are more similar when data on the near-poor are examined separately from the poor, or data on pregnant women are compared separately from those on children. Both studies found crowding out to be much less of a problem among children than among pregnant women. Dubay and Kenney also found income level to be a decisive factor, with crowding out occurring to a much greater extent among the near-poor than the poor.
Both groups of researchers acknowledged the shortcomings of using CPS data as a basis for estimating the extent of crowding out. One significant problem is that because the CPS asks individuals about the types of health insurance they had over the past year rather than about the duration of and transitions between different types of coverage, the survey may not capture the episodic nature of health insurance for many low-income people. Frequent job changes, often punctuated by periods of unemployment, can result in people having health coverage one month but not the next. These problems call for research with a panel survey in which the same people are interviewed over a period of time. Urban Institute researchers are embarking on such a study using the federal Survey of Income and Program Participation (SIPP).
Understanding Individual and Employer Responses
hile these studies do show evidence of crowding out, CPS data -- or any other survey data -- cannot reflect the motivating factors that might cause employer-provided coverage to fall as Medicaid eligibility increases.
Changes in employer and employee behavior provide various mechanisms through which crowding out could occur.
One hypothesis is that employers may have stopped offering health insurance or increased cost-sharing to encourage low-income workers to switch to Medicaid. However, Cutler and Grubers study found no evidence that employers used their knowledge of the expansions in Medicaid eligibility to justify dropping health insurance coverage for all employees or to demand higher employee cost-sharing. They did, however, find evidence of increased cost-sharing as Medicaid eligibility expands.
Another hypothesis is that as Medicaid eligibility is extended to more people, employees may be less likely to accept employer-sponsored coverage when it is offered or may actually drop coverage for themselves or their dependents. Given the potentially high cost of private health insurance due to premiums, deductibles, and coinsurance, pregnant women or women with Medicaid-eligible children could face an incentive to decline employer-based coverage or even to drop it. Cutler and Gruber did find some evidence that reductions in employer-sponsored coverage arose as a result of workers declining insurance when it was offered to them.
Benefits of the Expansions
stimates of crowding out are probably an overstatement of the costs of imperfect targeting, because Medicaid is often more valuable to low-income persons than is private insurance. For low-income persons, private health insurance has a variety of drawbacks. It may offer only rudimentary coverage, typically excluding preventive services that are covered under Medicaid. In addition, private insurance is expensive. According to Cutler and Gruber, under the typical private insurance policy, individuals pay roughly one-third of their total medical costs out-of-pocket through premiums, deductibles, and copayments.
Health insurance that is associated with a job presents particular problems for people whose lives are characterized by high job mobility and periods of unemployment. In these circumstances, Medicaid may provide more comprehensive and more consistent coverage than employer-sponsored health insurance. There is some evidence that Medicaid has important health benefits for the population it serves.
Another way of looking at the Medicaid crowding-out ledger is its impact on income redistribution, according to John Holahan of the Urban Institute.
Although the cost of employer-sponsored health coverage is borne by employers and workers, some would argue that much of the burden ultimately falls on workers through lower wages as well as premiums, deductibles, and copayments. Expanding Medicaid eligibility and using public funds to subsidize private insurance -- both of which result in a degree of crowding out -- are ways of shifting to the more affluent some of the burden of providing health insurance for the poor and the near-poor.
Alternatives to Expanding Eligibility
rom a health policy perspective, one question that arises from a discussion of expanding Medicaid eligibility and the crowding-out effect is whether a more effective and more efficient way to provide health coverage to the uninsured exists.
Some alternatives to expanding Medicaid eligibility include direct provision of care to the poor or offering subsidies to low-income people to purchase private health insurance. Holahan noted, however, that it is difficult to design such a program so that it does not subsidize low-income persons who currently have private coverage. In other words, subsidizing private insurance would have a substantial crowding-out effect of its own. Nor would such a program necessarily be less costly than expanding Medicaid, because experience suggests that subsidies need to be fairly generous to persuade people to purchase the subsidized product.
Crowding out of private health insurance might be reduced or ameliorated by introducing income-based fees for Medicaid services or offering a leaner package of Medicaid benefits to enrollees with higher incomes. Experience in other public programs, however, suggests that such efforts can become administratively cumbersome and costly. The way Medicaid currently operates, any eligible Medicaid recipient in a given state should be able to get covered services, once proof of eligibility is established. But if levels of eligibility and benefits schedules differ, providers of care would have to determine at the point of service which Medicaid recipient is entitled to which services. For providers, this could add to the cost and burden of a system that is already administratively difficult. And for enrollees, it could delay access to services.
Meeting Social Needs
xpanding Medicaid may be the least costly and the most fiscally progressive way of paying for health care for the near-poor, and crowding out may be a necessary tradeoff for increasing the proportion of the population with health insurance. Whether the glass is viewed as half full or half empty, researchers agree that expanded Medicaid eligibility has led to more needy people getting insurance.