M E D I C A R E B U Y - I N Betwixt And Between: Targeting Coverage Reforms To Those Approaching Medicare Should eligibility for a Medicare buy-in be based on age or ability to pay? by Dennis G. Shea, Pamela Farley Short, and M. Paige Powell ABSTRACT: Recent Medicare buy-in proposals agree on setting eligibility at age sixty-two but disagree on linking eligibility to loss of employer insurance or ability to pay. We examine arguments for targeting incremental coverage for older Americans in these ways. While access to retiree health insurance is declining, we question whether targeting loss of employer insurance can address many older Americans insurance problems. Furthermore, focusing on persons ages sixty-two to sixty-four misses a large group of persons in poor health with limited resources. Efforts to improve coverage for older Americans should consider trade-offs between defining eligibility by age versus ability to pay. DATAWATCH 1 In each of t he last three years of his presidency, Bill Clinton proposed in his State of the Union message to allow older Americans to buy into Medicare when they reach age sixty-two. Until the third year, when Clinton introduced a general percent tax credit to defray buy-in costs, the plan conspicuously avoided government subsidies. All buy-in costs were to be borne by enrollees. Even in its final iteration, the Clinton plan makes no special provision for those who cannot pay for insurance. However, Clinton does explicitly target displaced workers for assistance, allowing people who lost their jobs and insurance because of plant closings, company relocations, slack work, or the elimination of their positions to buy into Medicare at their own expense after age fifty-five. The final version of the Clinton plan also includes a general tax credit to help former employees exercise their right to extend employer insurance for eighteen months under the Consolidated Omnibus Budget Reconciliation Act (COBRA). 1 Another Medicare buy-in proposal, developed by Pamela Loprest Dennis Shea and Pamela Short are professors of health policy and administration, and Paige Powell is a doctoral candidate in the same department at the Pennsylvania State University. H E A L T H A F F A I R S ~ J a n u a r y / F e b r u a r y 1 1 ProjectHOPE ThePeople-to-People Health Foundation, Inc. Downloaded from HealthAffairs.org on February, 1.
D a t a W a t c h MEDICARE BUY-IN and Marilyn Moon, also allows everyone ages sixty-two to sixtyfour to buy into Medicare. In contrast to the Clinton plan, the Loprest-Moon proposal is specifically designed with subsidies according to ability to pay (with free enrollment for the poor and reduced premiums up to percent of the federal poverty level). There are no provisions in this plan for people who lose their jobs or employer coverage. There seems to be agreement on setting the minimum eligibility age at sixty-two for a Medicare buy-in but considerable disagreement on linking eligibility to the loss of employer coverage or ability to pay. Here we examine the arguments for and against targeting assistance for older Americans in each of these ways. Our findings rely on analysis of data from the March 1 Current Population Survey (CPS) and Waves 1 (11) and (1) of the Health and Retirement Survey (HRS). The surveys provide complementary information. The CPS offers a recent, cross-sectional look at the insurance, employment, income, and health status of the population nearing age sixty-five. The HRS, although less recent, provides information about changes in insurance, employment, assets, and income as persons approach age sixty-five. All differences discussed in the text are significant at the p <. level or better. Coverage Patterns By Age There are good reasons for targeting the age group, since access to employer insurance is reduced as people approach age sixty-five and retire. Consequently, older Americans rely most heavily on individual insurance, which is expensive and limited for people with serious health problems. Furthermore, because average health expenses increase sharply with age (by percent from the age group to the age group, according to unpublished data from the Medical Expenditure Panel Survey), the group closest to age sixty-five faces the greatest risk of being uninsured and being charged the highest premiums in the individual market. Despite these concerns, we do not find marked differences in coverage by age that would require policymakers to draw the line at age sixty-two. Retirement from full-time work peaks at age sixtytwo, when older Americans become eligible for Social Security (Exhibit 1). However, in contrast with retirement rates, the disenrollment rate from employer insurance across older age groups is flat. This may partly be the result of lower retirement rates among persons who would not be covered by employer insurance if they retired. n Job-based insurance. Overall enrollment in employer insurance declines well before age sixty-two (Exhibit ). Although enrollment in employer insurance in our 1 data fell still farther H E A L T H A F F A I R S ~ V o l u m e, N u m b e r 1 Downloaded from HealthAffairs.org on February, 1.
M E D I C A R E B U Y - I N EX HIBIT 1 Changes In Employment Or Insurance, By Age Group, 11 1 Percent Left full-time work Left employer-based insurance 1 1 Age 1 SOURCE: Health and Retirement Study, Wave 1 (11) and Wave (1). among the age group, the health insurance differences before and after age sixty-two were less dramatic than the employment differences. Furthermore, despite lower rates of employer insurance at older ages, the percentage uninsured in the and 1 age groups was the same and only slightly higher than the percentage uninsured in the age group. Even without a Medicare buy-in, Medicare covered a large percentage of persons ages under existing disabled coverage and covered almost two-thirds of the difference in employer insurance between persons in their early DATAWATCH 1 EXHIBIT Employment And Insurance Coverage Of Americans Ages Fifty To Sixty-Four, 1 Health insurance a Medicare Employer insurance Individual insurance Medicaid Uninsured Employed status Employed Full time/full year Part time/full year Full time/part year Part time/part year Not employed Retired Other % 1 % 1 % SOURCE: Current Population Survey, March 1. NOTE: Subgroup percentages may not sum to 1 percent because of rounding. a Persons with more than one type of insurance are assigned to the first applicable category shown. 11 1 1 1 11 1 % 1 1 H E A L T H A F F A I R S ~ J a n u a r y / F e b r u a r y 1 Downloaded from HealthAffairs.org on February, 1.
D a t a W a t c h MEDICARE BUY-IN fifties and early sixties. In addition, the gap in employer insurance among those ages was filled by increased individual insurance. As a result, new policies that focus on persons ages may miss a large number of older persons who would benefit from coverage reforms. Little more than a quarter of the 1 uninsured population ages were age sixty-two or older (about. million out of. million), and fewer than a third of the individually insured in this age group were age sixty-two or older (about. million out of 1. million). These patterns raise additional issues about targeting the loss of employer insurance. Partly because people retire only if they have retiree benefits, the percentage of persons losing employer insurance is relatively small (Exhibit 1). In the early 1s employers continued to cover more than four of every five insured persons who quit working in their fifties and early sixties (HRS data not shown). The differences we found in employer insurance by age still imply that many older Americans successfully negotiate employment transitions with their insurance intact, but access to the benefits that allow this transition is declining. However, since recent data suggest that access to retiree coverage declined by more than 1 percent from 11 to 1 among large employers, people may be more likely to lose coverage at retirement in the future. Relatively few of the uninsured or individually insured at older ages are people who recently lost employer insurance. In the early 1s about two-thirds of uninsured persons ages had been without coverage for two years or more. More recent data also indicate that more than half of uninsured persons ages have been uninsured for three or more years. Employers covered fewer than one of every five uninsured persons in the HRS within the preceding two years. After age sixty-two, more uninsured were recent disenrollees from individual insurance (1 percent) than from employer insurance ( percent), and people who bought individual insurance after age sixty-two were more likely to have been uninsured than to have lost employer insurance. n Policy implications. These statistics imply that offering relatively short-term coverage extensions to people who lose employer insurance, such as through COBRA or a Medicare buy-in, will miss most of the uninsured and individually insured. Providing long-term access to Medicare, beginning at a much earlier age than sixty-two, could eventually fill the gaps between employer insurance and Medicare for most people. However, such a policy would initially miss people who had already lost employer coverage. Even in the long run, focusing on people who leave employer insurance after age fifty-five still misses a relatively large number of people in their early H E A L T H A F F A I R S ~ V o l u m e, N u m b e r 1 Downloaded from HealthAffairs.org on February, 1.
M E D I C A R E B U Y - I N fifties who are uninsured or individually insured (1 percent, according to Exhibit ). Finally, targeting the loss of employer insurance helps a subgroup that is economically better off than most of the uninsured. According to HRS data on persons ages (not shown), more than three-fifths who lost employer insurance between 11 and 1 had incomes above percent of poverty in 1. In contrast, almost percent of the long-term uninsured, who made up two-thirds of the uninsured in 1, had incomes under percent of poverty. Ability To Pay And Health Insurance Reforms Insuring people who are close to age sixty-five is expensive, even through a Medicare buy-in. Analyses of Medicare buy-in proposals suggest that annual premiums would be $, $,. 1 Older Americans with incomes at percent of poverty who had no assets would have to spend about percent of their resources, and those at the poverty line would have to spend more than one-third of their resources, on a Medicare buy-in premium. Thus, people with limited resources are unlikely to participate without assistance. Reforms that do not consider ability to pay will exacerbate the existing economic disparities among older Americans. Coverage patterns vary more widely by income than by any other factor (Exhibit ). The uninsured percentage in 1 was more than four times higher and employer coverage four times lower for people below poverty than for people above percent of poverty. Despite high premiums, the poor enroll in individual insurance at almost twice the rate as do those with incomes over percent of poverty. Furthermore, because the uninsured and individually insured are concentrated at lower income levels, strategies that do not subsidize older Americans with limited economic resources are likely to miss a large percentage of the group who would benefit. If one considers either the or age group, about half of the uninsured have incomes below percent of poverty. 11 n Assessing financial status. While they point to economic disparities, statistics relating insurance status to current income probably understate the ability of some older Americans to pay for insurance. Because many people in this age group have already retired, current income may not be a good measure of economic status or ability to pay. Even if there were agreement on targeting by ability to pay, designing a consistent and accurate means test is challenging. To explore this issue, we examine prior income and assets of persons who turned age sixty-two between the first two waves of the HRS. Thirty-seven percent of such persons with incomes below poverty in 1 had incomes above percent of poverty in 11 DATAWATCH H E A L T H A F F A I R S ~ J a n u a r y / F e b r u a r y 1 Downloaded from HealthAffairs.org on February, 1.
D a t a W a t c h EXHIBIT Percentage Distribution Of Americans Ages Fifty To Sixty-Four, By Insurance Status And Individual Characteristics, 1 Total, % % % % 1% Age 1 Employment status Full time/full year Part time/full year Full time/part year Part time/part year Retired Other Annual income Below poverty 1 1% of poverty 1% of poverty % of poverty Over % of poverty Self-reported health status Excellent Very good Good Fair Poor Work disability Limited and not working Limited and working Not limited 1, 1,,,,, 1,1,,,1,,1,,,, 1,1,,, 1,, 1 1 1 1 1 SOURCE: Current Population Survey, March 1. NOTES: Subgroup percentages may not sum to 1 percent because of rounding. Persons with more than one type of insurance are assigned to the first applicable category shown. 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 (Exhibit ). This means that many persons ages who would qualify for free coverage under income-based subsidies would not have qualified for subsidies before turning sixty-two. Because enrollment in individual insurance is a sign of economic advantage, the individually insured with low incomes are even more likely than the uninsured to be better off. Considering the entire age cohort covered by the HRS (data not shown), where the sample size is sufficient to make these estimates, percent of the individually insured in poverty in 1 had incomes in excess of percent of poverty in 11, while almost two-thirds of the uninsured in poverty in 1 had incomes under percent of poverty in 11. The individually insured between 1 and percent of poverty were twice as likely as the uninsured in the same income range to have assets exceeding $1, ( percent versus 1 percent). Given these patterns, methods used to determine ability to pay may differentially affect these two key groups. H E A L T H A F F A I R S ~ V o l u m e, N u m b e r 1 Downloaded from HealthAffairs.org on February, 1.
M E D I C A R E B U Y - I N EXHIBIT Income In 1, By 11 Income And 1 Assets, For Americans Over Age Sixty-Two Population (thousands), 11 income Below poverty 1 1% of poverty % of poverty Over % of poverty 1 assets Less than $, $, $1, More than $1, % 11 1 % 1 1 SOURCE: Health and Retirement Study, Wave 1 (11) and Wave (1). % 1 1 % 1 1 Alternative Target Groups A program that targeted individuals ages and provided no income-related subsidies would provide a source of low-cost group insurance to uninsured and individually insured persons in this age group. Three groups would primarily benefit. First, about, persons with individual insurance above percent of poverty might participate in a Medicare buy-in if its premium were lower than their current private premium. This would depend on administrative costs, adverse selection, and other factors. In addition, 1, individually insured persons with incomes below percent of poverty might also benefit from a Medicare buy-in. They have shown their willingness to buy insurance, and the buy-in might offer a cost advantage over their current coverage. Finally, the, uninsured persons with incomes greater than percent of poverty might also benefit from an unsubsidized Medicare buy-in. These people arguably have the means to pay for health insurance. Lower premiums and the elimination of healthrelated coverage limitations through a Medicare buy-in might encourage them to enroll. Given the cost of the unsubsidized Medicare buy-in premium, it is unlikely that many uninsured persons below percent of poverty would participate. There are several important differences in the characteristics of the three groups most likely to benefit from such a buy-in (Exhibit ). While more than half of the individually insured with incomes greater than percent of poverty in 1 were in excellent or very good health, almost one-third of those with incomes below percent of poverty were in fair or poor health. The health of the uninsured above percent of poverty most closely resembled that DATAWATCH H E A L T H A F F A I R S ~ J a n u a r y / F e b r u a r y 1 Downloaded from HealthAffairs.org on February, 1.
D a t a W a t c h EXHIBIT Characteristics Of Uninsured And Individually Insured Americans, Ages And 1, 1 Population (thousands) 1 1 1 Self-reported health status Excellent Very good Good Fair Poor Employment status Employed Not employed Annual income Below poverty 1 % of poverty % of poverty % of poverty Over % of poverty 1% 1 1 1 1% 1 1 % 1% % 1% 1 Population (thousands) 1,1, 1 1, 1,1 Self-reported health status Excellent Very good Good Fair Poor Employment status Employed Not employed Annual income Below poverty 1 1% of poverty 1% of poverty % of poverty Over % of poverty 1% 1 1 1 1% 1 1 1 1 1 1% 1 1 SOURCE: Current Population Survey, March 1. NOTE: Subgroup percentages may not sum to 1 percent because of rounding. 11% 1 % 1 1% of the individually insured at the same income level. Persons in both groups above percent of poverty were largely employed in 1, while just percent of the individually insured under percent of poverty worked. Adding income-related subsidies to a program for people ages, similar to those proposed by Loprest and Moon, would encourage participation by uninsured persons below percent of poverty. In most respects, the nearly, uninsured persons with incomes below percent of poverty were similar to the individually insured at the same income level (Exhibit ). Extending eligibility to a younger age group, such as those ages 1, would greatly expand the size of the program. An additional. million uninsured or individually insured persons were in this age group in 1. Without targeted subsidies, the expanded pro- H E A L T H A F F A I R S ~ V o l u m e, N u m b e r 1 Downloaded from HealthAffairs.org on February, 1.
M E D I C A R E B U Y - I N gram would again attract uninsured and individually insured persons with incomes above percent of poverty. The higher-income groups in both age groups in Exhibit have similar characteristics. A program with income-related subsidies that included the younger age group would enhance participation among those below percent of poverty. The health of those with low incomes in this age group was similar to the health of those in the older age group with the same insurance and income. While low-income persons age 1 were more likely than their older counterparts to be working, they were also more likely to be poor. A program that considered the larger age group (ages ) but restricted eligibility to persons with incomes below percent of poverty, similar to recent expansions of children s health insurance, would target almost. million persons. As noted above, lowincome persons ages 1 were in as poor health as and had lower incomes than low-income persons ages. The poor health of this low-income group would result in much higher premiums compared with programs that offered broader eligibility. Discussion Our results shed light on the merits of targeting assistance for older Americans by age, income, and access to employer insurance. Our findings raise several questions about each approach. n Targeting those who lose employer insurance. The steady erosion of employer coverage as people approach age sixty-five means that a program targeting the loss of this coverage would reach relatively few persons in the short run. Few currently uninsured or individually insured persons have a recent history of disenrolling from employer insurance. In addition, many older Americans retire before age sixty-five without loss of employer insurance, so targeting this group may crowd out current private arrangements. Finally, targeting those who lose employer insurance helps people who are economically better off than people excluded by that strategy. On the other hand, the current high rate of employer insurance among retirees occurs partly because many people only retire if they will have insurance afterwards. If retiree health benefits continue to decline, some of these persons could pay for their own insurance at little cost to the government if given access to group insurance or Medicare. Furthermore, focusing on the loss of employer coverage, as the Clinton plan does, may avoid crowding out private insurance. Whether it is better to proceed with a relatively low cost program that targets the upper middle class or a potentially more costly program for the lower middle class is an ongoing debate. n Targeting according to age. Our results suggest that age DATAWATCH H E A L T H A F F A I R S ~ J a n u a r y / F e b r u a r y 1 Downloaded from HealthAffairs.org on February, 1.
D a t a W a t c h Focusing only on persons ages sixty-two to sixty-four misses a group with significant health problems and limited resources. MEDICARE BUY-IN sixty-two is not a critical threshold affecting insurance coverage. Although insurance is more difficult to access and afford as individuals approach age sixty-five, the likelihood of losing employer insurance or being uninsured is the same among persons age 1 as among persons ages. Coverage changes steadily throughout the years approaching age sixty-five, rather than precipitously at age sixty-two. Consequently, policies directed at persons ages miss many older Americans who could benefit from access to group insurance. In particular, it misses a large number of uninsured and individually insured persons ages 1 below percent of poverty who are in as poor health as and have even lower incomes than similar persons ages. One problem with a lower eligibility age is the likely effect on earlier retirement. Many policy analysts recommend incentives for longer work lives as the baby boomers approach retirement. Targeting people at younger ages runs counter to current trends in Social Security and raises further questions about the ability to fund health and income security programs for future generations. n Focusing on the poor. Focusing reforms on adults with limited economic resources is consistent with current incremental expansions targeted at the parents of children eligible for Medicaid or the State Children s Health Insurance Program (SCHIP). 1 Although targeting by ability to pay is difficult, the connection between ability to pay and insurance is dramatic in this age group. Furthermore, persons below percent of poverty, particularly the uninsured, face more health problems than do others at the same age. n Asset testing. To be fair and reliable, means testing during early retirement might involve repeated or complicated income and asset tests. In addition to the administrative expense, asset testing penalizes persons who chose to save more during their working years. Ideally, for persons in this age group, eligibility should be based on some longer-term measure of productivity and earning capacity. Such a test could be devised by considering lifetime earnings data already compiled by the Social Security Administration. 1 In addition to the administrative challenges, a means-tested program, especially one that targeted persons ages, would require large subsidies because of low incomes and poor health in this age group. Funding from general tax revenues or the current surplus would have to be identified. The costs of such an expansion could be H E A L T H A F F A I R S ~ V o l u m e, N u m b e r 1 Downloaded from HealthAffairs.org on February, 1.
M E D I C A R E B U Y - I N prohibitive and could be even larger if the means-tested program resulted in significant crowding out, as some research of Medicaid expansions has suggested could occur. Thus, the connections between ability to pay and insurance in this age group raise important questions about the appropriate method of targeting incremental coverage reforms for Americans nearing age sixty-five. Policies that merely allow older Americans to buy into Medicare without income-related subsidies will not address the insurance problems of those with limited economic resources in this age group. At the same time, subsidies for older Americans who lack access to group coverage are likely to be expensive, because of the high health care costs and lack of economic resources in this age group. As a result, limiting the program by age is probably necessary to limit program costs. However, focusing only on persons ages sixty-two to sixty-four misses an important group of older Americans with significant health problems and limited economic resources who also need access to more affordable insurance. While a program open to everyone ages fifty-five to sixty-four might be too costly, a means-tested program that extended below age sixty-two could benefit a segment of the population that has a genuine need for assistance. Since policymakers have some flexibility in choosing an age cutoff that is either older or younger than sixty-two, efforts to improve health insurance for older Americans should give more careful consideration to the trade-offs between defining eligibility by age versus ability to pay. In the spirit of incremental reform, initial eligibility for a Medicare buy-in could be expanded over time by relaxing the means test or lowering the age threshold, so that more older Americans could benefit from Medicare. DATAWATCH This research was supported by a grant from the Commonwealth Fund. The views presented here are those of the authors and not necessarily those of the Pennsylvania State University or the Commonwealth Fund or its directors, officers, or staff. We acknowledge the assistance of Sherry Glied and Shana McCormack at Columbia University, Cathy Schoen and Karen Davis at the Commonwealth Fund, Jim Mays at Actuarial Research Corporation, Steve Maczuga at the Population Research Institute at Penn State, and the helpful comments of two reviewers. The authors assume responsibility, of course, for all errors of omission and commission. H E A L T H A F F A I R S ~ J a n u a r y / F e b r u a r y 1 Downloaded from HealthAffairs.org on February, 1.
D a t a W a t c h MEDICARE BUY-IN NOTES 1. J. Sheils and Y.J. Chen, The Potential Impact of the President s Medicare Buy-In Proposal on the Number of Uninsured Persons and Program Costs, Final Report (New York: Commonwealth Fund, August, 1); and White House, Office of the Press Secretary, The Clinton-Gore Administratio n FY 1 Budget: Maintaining Fiscal Discipline while Making Key Investments, Summary Documents, February, <www.pub.whitehouse.gov/urires/ir?urn:pdi://oma.eop.gov.us////11.text.1> ( July ).. P. Loprest and M. Moon, Medicare Buy-In Proposal, Kaiser Project on Incremental Health Reform, <www.kff.org/content/1/111k/loprestmoon.pdf> ( October ).. For more detailed information on the data and methods, see P. Short, D. Shea, and M. Powell, Health Insurance on the Way to Medicare: Is Special Government Assistance Warranted? Commonwealth Fund Task Force on the Future of Health Insurance for Working Americans (New York: Commonwealth Fund, forthcoming).. J. Currie and B.C. Madrian, Health, Health Insurance, and the Labor Market, in Handbook of Labor Economics, vol. C, ed. O. Ashenfelter and D. Card (Amsterdam: Elsevier, 1).. U.S. General Accounting Office, Retiree Health Insurance: Erosion in Employer-Based Health Benefits for EarlyRetirees, Pub. no. GAO/HEHS-- (Washington: GAO, 1); and F. McCardle et al., Retiree Health Coverage: Recent Trends and Employer Perspectives on Future Benefits, October 1, <www.kff.org/content/1// retiree.pdf> ( October ).. Short et al., Health Insurance on the Way to Medicare.. C. Schoen et al., Counting on Medicare: Perspectives and Concerns of Americans Ages to, Commonwealth Fund Task Force on the Future of Health Insurance for Working Americans (New York: Commonwealth Fund, July ).. Short et al., Health Insurance on the Way to Medicare.. Ibid. 1. White House, Clinton-Gore Administration FY 1 Budget; Loprest and Moon, Medicare Buy-In Proposal; and P.F. Short, D.G. Shea, and M.P. Powell, A Workable Solution for the Pre-Medicare Population, Task Force on the Future of Health Insurance report series, Strategies to Expand Health Insurance for Working Families (New York: Commonwealth Fund, October ). 11. Short et al., Health Insurance on the Way to Medicare.. Currie and Madrian, Health, Health Insurance, and the Labor Market. 1. L. Dubay, G. Kenney, and S. Zuckerman, Extending Medicaid to Parents: An Incremental Strategy for Reducing the Number of Uninsured, National Survey of America s Families, Report no. B- (Washington: Urban Institute, June ). 1. Short et al., A Workable Solution for the Pre-Medicare Population. Eligibility tests could compare lifetime earnings to average poverty levels during the same period and be provided to the Health Care Financing Administration. This approach assesses long-term economic resources, is unaffected by retirement decisions after the age when eligibility is determined, is easy to administer, and does not encourage asset planning. One weakness may be an inability to consider short-term, yet critical, economic reversals or the unique circumstances of older women who are divorced or widowed.. L. Dubay, Expansions in Public Health Insurance and Crowd-Out: What theevidence Says, Kaiser Project on Incremental Health Reform, October 1, <www.kff.org/ content/1/111m.dubay.pdf> ( October ). H E A L T H A F F A I R S ~ V o l u m e, N u m b e r 1 Downloaded from HealthAffairs.org on February, 1.