The views expressed are those of the authors and do not necessarily reflect those of the Urban Institute, its board, its sponsors, or other authors in the series.
I. Introduction
Extending health insurance coverage to more children is currently of considerable policy interest as evidenced by the number of bills so far introduced in the 105th Congress. 1 These legislative initiatives propose a range of mechanisms for reducing the number of uninsured children including: tax credits and vouchers for families to assist them in purchasing insurance coverage for their children, grants to states to design and finance new health insurance coverage programs, and a new entitlement program. The motivation behind these initiatives is the belief that providing health insurance coverage to uninsured children will improve access to and use of health care which will in turn lead to improved health. In addition, these initiatives seek to provide some financial relief to working uninsured families.
While only one of the legislative initiatives on child health insurance coverage proposes expansions of the Medicaid program, the experience of the Medicaid expansions for children and pregnant women is relevant for today's policy debate. The Medicaid expansions provided full subsidies for the health insurance coverage of certain low-income children. Many of the proposed initiatives would provide full or partial subsidies of health insurance coverage for children who live in families with incomes that exceed current Medicaid eligibility thresholds, with the more generous including families up to 300 percent of poverty. The expansions provide important lessons regarding the ability of programs that provide health insurance coverage to reach their target population and reduce the number of uninsured children. In addition, they can shed light on the amount of "crowding-out" that can be anticipated.
Crowding-out is a phenomenon whereby new public programs or expansion of existing public programs designed to extend coverage to the uninsured prompt some privately insured persons to drop their private coverage and take advantage of the expanded public subsidy. The issue of crowding out can be important because it may lead to fewer improvements in access to care and greater program costs than expected.
Our testimony today will provide information on the extent and nature of the population of children without health insurance and describe the lessons learned from the expansions in Medicaid coverage for children and pregnant women that occurred over the last decade. We will also identify areas where more information is needed in order to make informed policy choices.
Our testimony can be summarized by the following six conclusions:
- Policy solutions aimed at reducing the number of uninsured children must take a multi-pronged approach. This type of approach is necessary because about a quarter of uninsured children are currently eligible for Medicaid but not enrolled, older children living in poverty will not all be covered by Medicaid until 2002, and the remainder, almost three quarters of all uninsured children, live in families with incomes above poverty.
- New programs that provide public subsidies for health insurance coverage will result in some crowding out of private coverage. The magnitude of this effect will depend on the income eligibility level of the program, the success of the attempts made to minimize the substitution of public coverage for private coverage, the magnitude of premium cost-sharing for employer-ponsored coverage faced by those eligible for the new program, and the generosity of the benefit package under the new program relative to under employer-sponsored coverage.
- Programs that limit coverage to lower income groups will find that a relatively small percentage of new public dollars will be replacing private employer and individual payments. Programs that offer coverage to children at higher incomes could potentially see a large share of public dollars replacing employer and individual contributions thus affecting the distribution of who pays for health insurance coverage.
- In an era of scarce resources, it is important to reduce the incentives to substitute public dollars for private dollars. While mechanisms to reduce this crowd-out effect are important, it is difficult to prevent substitution without creating inequities in access to coverage.
- While programs that phase out subsidies as income increases will discourage the substitution of public coverage for private coverage, they may also discourage families with uninsured children from purchasing insurance for their children. Without large subsidies, the ability of a new program to reduce the number of uninsured children may be compromised.
- In order to assure that most uninsured children receive health insurance coverage, we may need to accept a shifting of the distribution of who pays for such coverage from the private to the public sector as part of the cost of this coverage.
II. The Problem
According to 1994 estimates from the Current Population Survey (CPS), more than seven million children lack health insurance coverage. Uninsured children come from all income and age groups (see Table 1). Altogether 54 percent of uninsured children live in households with income less than 185 percent of the federal poverty line and almost 23 percent of uninsured children live in households below the federal poverty line.
The distribution of insurance coverage for children varies by household income and age of child (Table 2). The risk of being uninsured increases with the age of the child, particularly in poorer families. Overall, children age 13 to 18 are a third more likely than those under 6 to lack health insurance. In households with incomes below the federal poverty line, older children were three times as likely to lack health coverage relative to the younger children. For children of all ages, lower rates of uninsurance occurred at the very bottom and top of the income ranges; living in households with incomes between 133 and 185 percent of the federal poverty level puts children at the greatest risk of not having health insurance coverage.
Patterns of insurance coverage are influenced by Medicaid eligibility policies which offer protection to all poor children under 14 years of age. In response to evidence indicating declining health status for low-income children and growing disparities in access to health care between the insured and the uninsured, Medicaid coverage for children was expanded in the late 1980s. Congress permitted and eventually mandated states to provide phased-in Medicaid coverage for children up to age six in families with incomes up to 133 percent of the federal poverty level and to all children born after September 30, 1983, in families with incomes at the poverty line or below. Under the phase-in, all children under age 18 living in households beneath the federal poverty line will be eligible for Medicaid by the year 2002. States were also given the option to cover infants with family incomes up to 185 percent of poverty. 2 These expansions represented a dramatic change from the past, when children qualified for Medicaid only if their families' incomes were below AFDC thresholds (which had averaged only about 50 percent of the poverty level prior to the expansions).
III. What Did We Learn From the Medicaid Expansions for Pregnant Women and Children?
The expansions in Medicaid coverage for pregnant women and children that took place in the late 1980s and early 1990s provide important lessons regarding the ability of programs that subsidize health insurance coverage to enroll eligible children and reduce the number of uninsured children, as well as the amount of crowding-out that can be anticipated under such programs.
The intent of the Medicaid expansions was to reduce the number of uninsured children and pregnant women, increase access to health care, and thus, improve children's health. Between 1988 and 1993, the number of children receiving Medicaid-covered services grew by 10.7 to 16.5 million a 54 percent increase (unpublished tabulations of HCFA Form 2082 data). The number of births financed by Medicaid also increased substantially (Sing, Gold, and Frost, 1994).
Over the same period, which also witnessed an economic recession, employer-sponsored insurance coverage was declining (Holahan, Winterbottom, and Rajan, 1996; Peat Marwick, 1994), and the number of uninsured children grew (Dubay and Kenney, 1996). The simultaneous decline in employer-sponsored coverage and increase in Medicaid coverage of children, coupled with the increase in uninsured children has led some observers to suspect that the Medicaid expansions for children and pregnant women "crowded-out" employer-sponsored coverage. To try to identify whether, and if so how much, the expansions crowded out private coverage, researchers at the Urban Institute conducted the following analysis.
The Dubay-Kenney Study
In the this study, we examined changes in health insurance coverage for children and pregnant women using CPS data edited by the Urban Institute's Transfer Income Model (TRIM2) and representing 1988 and 1991/1992 3. We focused our analysis on the target population of poor and near poor pregnant women and children ages 10 and under. Our overall approach in assessing the impact of the expansions on insurance coverage was to examine aggregate changes in health insurance coverage separately for pregnant women and children 10 years old and younger by income group.
We first assessed the extent to which the expansions were covering the target population. We found participation rates for expansion eligible children and pregnant women to be less than that under the traditional Medicaid program where over 90 percent of those eligible enroll. Only 44 percent of pregnant women eligible for the expansions who did not have employer-sponsored insurance enrolled in Medicaid. Sixty-nine percent of the children eligible under the expansions who did not have employer-sponsored coverage enrolled in the Medicaid program. Whether the lower participation rates for the expansion population are due to lack of knowledge about the new eligibility rules, unwillingness to enroll in Medicaid, or persisting problems with the Medicaid eligibility determination process is unclear. The fact that such a large percentage of uninsured children are Medicaid eligible suggests that large inroads into the problem could be made by increasing Medicaid participation rates and illustrates the importance of understanding why the participation rate is so low.
We then estimated the extent to which the expansions crowded out employer-sponsored insurance. We compared the declines in employer-sponsored coverage for children and pregnant women to the declines for men ages 18-44—a group unlikely to be affected by the expansions. We did this to control for the portion of the decline in employer-sponsored coverage for children and pregnant women that would have occurred in the absence of the expansion. The difference between the decline in employer-sponsored coverage for children and pregnant women and that for men is the amount of the decline in employer-sponsored coverage for children and pregnant women attributable to crowd-out. We then divided this decline by the increase in Medicaid enrollment. This provides our estimate of how much of the Medicaid enrollment increase resulted from crowd-out.
About 14 percent of the increase in Medicaid enrollment of pregnant women and 17 percent of the increase in enrollment of young children was attributable to crowd-out, according to our estimates. These estimates represent the degree to which public funds substituted for private funds over the period. We find no evidence of crowding out for poor (that is, below the poverty line) pregnant women and very little crowding out for poor children. For pregnant women and children with household incomes above the poverty line (that is, 100-185 percent of poverty for pregnant women and 100-133 percent of poverty for children) we find the crowd-out effect to be 45 and 21 percent respectively. The higher crowd-out estimate for pregnant women suggests that more crowding out occurs as income eligibility thresholds increase.
We also found that more than 75 percent of the increase in Medicaid enrollment over the expansion period was for children and pregnant women who would otherwise have been uninsured or would have lost their insurance as a result of secular declines in employer-sponsored coverage. This means that, without the Medicaid program, an additional 3 million children would have been uninsured in 1992.
Study Limitations. Using men ages 18-44 is not a perfect control for how the health insurance coverage of children and pregnant women would have changed in the absense of the expansions. For example, if the secular declines in coverage were greater for children and pregnant women than for men, we may over-estimate the extent of crowding out. In fact, there is evidence to suggest that this is the case: between 1989 and 1993, employee health insurance contributions rose by twice as much for family coverage as for individual coverage, providing a much more substantial disincentive to continue employer-sponsored coverage of dependents. 4
It has also been claimed (Culter and Gruber 1997) that we underestimate the Medicaid crowd-out effect by failing to count the spillover effect within families. This effect comes about when a family drops employer-sponsored coverage because some of the family members (pregnant mother and younger children) are eligible for, and choose to take advantage of, the Medicaid expansion--leaving the ineligible family members (older children) without insurance. Since these family members are ineligible for Medicaid, we do not consider this a crowd-out effect. However, even with a broader interpretation of crowd-out, the spillover effect is likely to be small. In work that we are currently doing, we find that less than 3 percent of the children living in families with Medicaid-covered children are uninsured. This 3 percent estimate is an upper bound on the spillover effect since some of these uninsured children would have been uninsured even in the absence of the expansions.
This research, and much of the other literature on crowd-out, uses the Current Population Survey which is a cross-sectional database. A definitive analysis of the crowd-out issue requires the use of a longitudinal database to shed light on the dynamic nature of health insurance coverage. In other words, in order to understand changes in insurance coverage over time, we want to be able to observe the transitions between one type of insurance coverage and another. When cross-sectional data are used, the movement of one group out of employer-sponsored coverage and into the uninsured category combined with another group moving from the ranks of the uninsured into Medicaid might appear to be a movement from employer-sponsored coverage into Medicaid.
In order to eliminate this problem, a new Urban Institute study by Blumberg, Dubay and Norton, is using the 1990 panel of the Survey of Income and Program Participation —a data base that follows the same households over a 2 and a half year period—to examine health insurance coverage transitions for poor and near poor children over the expansion period. The results of this analysis will be released in the near future.
IV. What Lessons Are Relevant for Today's Policy World?
As mentioned previously, some of the legislative initiatives currently in Congress propose to provide full or close to full subsidies to purchase health insurance coverage for children in families with incomes up to some specified level. In this way, the initiatives are similar to the Medicaid expansions and some lessons can be easily applied. At the same time, there are aspects of the initiatives that are unlike the expansions. For example, income eligibilty for many of the proposed programs would be substantially higher than under the Medicaid expansions and those with higher incomes would receive only partial subsidies.
Four lessons stand out as important for today's policy discussion regarding federal
programs to subsidize the costs of health insurance coverage for children.
- Programs that subsidize health insurance coverage for children will reduce the number of uninsured children.
- Even when the entire cost of health insurance coverage for children is subsidized, some eligible children will remain uninsured.
- Subsidizing insurance coverage for children in poor households will result in very little substitution of employer-sponsored coverage, in large part because this population has very little insurance coverage to begin with.
- The higher the income-eligibility cutoff, the greater will be the crowd-out effect. This is because as income increases, the prevalence of employer-sponsored coverage increases and the proportion of households without insurance decreases. Therefore, even if only a small percentage of families substitute private for public coverage and participation by the otherwise uninsured is relatively high, as the income eligibility cutoff for a new program increases, the percentage of entrants into that program will increasingly come from those who previously had private coverage.
These are important lessons. There are also a number of limitations in the applicability of these analyses to the types of health insurance programs currently being considered by Congress.
- Since the expansions were limited to children and pregnant women with incomes below 133 and 185 percent of poverty respectively, there is no evidence on how much public coverage would substitute for private coverage in programs with higher income eligibility levels.
- Since the expansions fully subsidized the costs of health insurance coverage, there is also no evidence from this literature regarding how premium cost-sharing would affect either participation in the program or the dropping of employer-sponsored coverage.
- The dynamics of family participation in a health insurance programs other than Medicaid may be quite different from those driving Medicaid participation, particularly if the stigma associated with the program is lower, the eligibility determination process is different, or the benefit package under the program is less comprehensive.
- There is little evidence that employers responded to the expansions by reducing their offers of and contributions to health insurance coverage (Cutler and Gruber 1996). The proposed initiatives that cover children up to 300 percent of poverty would make 60 percent of all children eligible for the program. Such large-scale initiatives could alter employer behavior, potentially further reducing employer contributions for dependent health insurance coverage.
- Finally, none of the studies tells us why those individuals who dropped their private insurance over this period did so. For example, it may be that the families who substituted Medicaid for private coverage were those with policies that had high deductibles and co-payments, covered only catastrophic illness, or did not cover preventive services. For some low-income families, the movement into the Medicaid program may have represented access to coverage that they did not previously have. Similarly, those families that dropped their private coverage may have faced premium contributions that represented large financial burdens.
V. What Can be Done to Appropriately Target New Programs and Limit the Extent of Crowding Out?
Given the current federal budget constraints, it is important that new public programs that subsidize health insurance coverage of children be appropriately targeted in order to get the most "bang for the buck." What do we know about the effectiveness of strategies to reduce the substitution effect? States have implemented a number of strategies designed to prevent crowding-out when they expanded insurance coverage (either through their Medicaid Section 1115 waivers or though their state-only health insurance programs). These strategies have included limiting eligibility to individuals a) who have been uninsured for a period of time, b) who do not have an offer of employer-sponsored coverage, or c) who face premium cost-sharing greater than 50 percent for their employer-sponsored coverage. However, these programs are relatively new and the effectiveness of these mechanisms at reducing crowd-out has not been assessed. States have also found these types of mechanisms administratively complex (Wooldridge et al 1997) making them difficult to implement.
Moreover, these types of initiatives could also prevent families who lost their employer-sponsored coverage for reasons beyond their control, such as job loss or reductions in employer contributions to premiums, from taking up the new program. Thus, they potentially create inequities in who is eligible for the subsidy.
Each of the bills introduced in Congress this year would require that some families contribute to the costs of health insurance coverage under the new program and would vary the subsidy based on family income. Families for which the offered subsidy is less than the cost sharing they currently face (including differences in benefit packages, co-payments, deductibles, and premium contributions) will be unlikely to substitute public coverage for private. Thus lower subsidies would tend to reduce the crowd-out effect. However, the subsidy schedule will also affect the extent to which families with uninsured children will participate in the program. According to research by Marquis and Long (1995), in order for low-income working families to purchase insurance, subsidies must be quite high.
VI. Conclusions
The conundrum facing policy makers today is how to cover a substantial number of uninsured children without also covering children who would otherwise be insured. On the one hand, evidence suggests that subsidies will have to be large in order to achieve significant reductions in the number of uninsured children. At the same time, concern about substitution of public coverage for private coverage is a real issue from a budget perspective and it is not clear how to prevent it.
The fact that uninsured children are not concentrated at the low end of the income distribution makes this a very challenging problem. For example, new programs that subsidize coverage for children in families with incomes up to 185 percent of poverty may produce "acceptable" levels of crowd-out, but will leave 46 percent of uninsured children still uncovered 5. In contrast, programs that include children in families with incomes up to 300 percent of poverty would make more than 80 percent of all uninsured children eligible. Yet, under such programs the share of participants that previously had private coverage will likely be large unless mechanisms to limit the substitution of public for private coverage are effective. Moreover, for some families the cost of maintaining their private insurance may be burdensome, thus limiting their access to new programs may create new inequities (Holahan 1997). Therefore, in order to assure that children in this country have health insurance coverage it may be necessary to accept a greater public role in its financing.
REFERENCES
Blumberg, Linda, Lisa Dubay and Stephen Norton. 1996. "Did the Medicaid Expansions for Children Displace Private Insurance?" Presented at the 18th Annual Conference of The Association for Public Policy Analysis and Management, Pittsburgh, PA, November 2, 1996.
Cutler, David, and Jonathan Gruber. 1996. "Does Public Insurance Crowd Out Private Insurance?" The Quarterly Journal of Economics, May.
Cutler, David, and Jonathan Gruber. 1997. "Medicaid and Private Insurance: Evidence and Implications." Health Affairs (16:1) January/February.
Dubay, Lisa and Genevieve Kenney. 1996. "Revisiting the Issues: The Effects of Medicaid Expansions on Insurance Coverage of Children". The Future of Children 6:1 (Spring 1996).
Dubay, Lisa and Genevieve Kenney. 1997. "Did the Medicaid Expansions for Pregnant Women Crowd-Out Private Insurance?" Health Affairs 16:1 (January/February).
Holahan, John. Forthcoming. "Expanding Insurance Coverage for Children." Urban Institute Monograph.
Holahan, John, Colin Winterbottom, Shruti Rajan. 1995. "A Shifting Picture of Health Insurance Coverage." Health Affairs 14:4 (Winter).
KPMG Peat Marwick. 1993. Health Benefits in 1993. Newark, NJ:KPMG Peat Marwick.
Marquis, M.S., and S.H. Long. 1995. "Worker Demand for Health Insurance in the Non-group Market." Journal of Health Economics 14:1 (May).
Wooldridge, Judith, Leighton Ku, Teresa Coughlin, Lisa Dubay, Marilyn Ellwood, Shruti Rajan, Sheila Hoag. 1996. "Implementing State health Care Reform: What Have We Learned from the First Year ? The First Annual Report of the Evaluation of Health Reform in Five States." Submitted to Office of Research and Demonstration, Health Care Financing Administration, December.
Notes
1. These bills include: the Children's Health Coverage Act (S. 13, Daschle, D-SD); the Health Assurance Act (S. 24, Spector, R-PA); the Healthy Children's Pilot Program Act (S. 435, Spector, R-PA); the Healthy Start Act (H.R. 560, Stark, D-CA): the Children's Health Insurance Act (H.R. 561, Stark, D-CA); the Child Health Insurance and Lower Deficit Act (Hatch, R-UT, Kennedy, D-MA). In addition the Family and Child Health Assurance Act (Gramm, R-TX and Coverdell, R-GA) will be introduced and the President's FY 1998 budget addresses health insurance coverage for children.
2. In addition, using other provisions of Medicaid law (Section 1902(r)(2) and Section 1115), some states have chosen to offer coverage to children in households with higher income levels than specified in the expansions.
3. Dubay and Kenney use 1991 data for their analysis of pregnant women and 1992 data for their analysis of children as their post-expansion period.
4. Authors' computations of the percentage change in average monthly employee contributions towards health insurance premiums between 1989 and 1993 in medium and large firms. From the Bureau of Labor Statistics' Employer Benefits Survey, U.S. Department of Labor, Bulletins 2363 and 2456.
5. Assuming full participation.
* The research presented in this statement was funded by the Robert Wood Johnson Foundation and the Health Care Financing Administration. This statement represents the views of the authors alone and not of the Urban Institute, its sponsors or its trustees. We are grateful for helpful comments from our colleagues John Holahan and Stephen Zuckerman. Beth Kessler provided outstanding research assistance for this testimony.
Latest Research
Browse by Topic:
Printer-friendly summary
