Raising The Medicare Eligibility Age: Effects On The Young Elderly

Transcription

1 DataWatch Raising The Medicare Eligibility Age: Effects On The Young Elderly To be successful, a raised eligibility age should be accompanied by a Medicare buy-in subsidy for sixty-five- and sixty-six-year-olds. by Amy J. Davidoff and Richard W. Johnson ABSTRACT: If Medicare eligibility were delayed to age sixty-seven, as proposed periodically by policymakers, most sixty-five- and sixty-six-year-olds (the young elderly ) would find alternative sources of coverage. However, the loss of Medicare eligibility would leave about 9 percent of this age group uninsured, while another 11 percent would be underinsured because they could only afford limited nongroup policies. The impact would be much greater for certain subgroups, including blacks, Hispanics, and the poor. A buy-in plan that allows young elderly people to purchase Medicare coverage could reduce uninsurance rates, but only if it subsidizes premiums for those with limited incomes. Policymakers periodically advocate raising the age of eligibility for Medicare beyond sixty-five to contain program costs, which will grow rapidly once the large baby-boom cohort begins to receive benefits. 1 Proponents argue that improvements over time in the health of the aged population now permit many older adults to work past age sixty-five, which reduces the need for Medicare coverage before beneficiaries reach their late sixties. In fact, raising the age of eligibility to sixty-seven would more closely align Medicare with Social Security, which for those born after 1959 provides full benefits only to retirees who wait until age sixty-seven to begin collecting. A delay in Medicare eligibility could reinforce incentives in the Social Security system for workers to delay retirement and remain productively engaged in the labor force. Although the vast majority of young elderly adults (sixty-five- and sixty-sixyear-olds) receive insurance coverage through Medicare, current patterns of supplemental coverage suggest that viable alternatives to Medicare exist. Many older people receive supplemental coverage through Medicaid, employer-sponsored insurance, and privately purchased Medigap policies. For the nondisabled population, these alternative sources of coverage could become the sole sources if the Medicare eligibility age were raised. A raise in the Medicare eligibility age would also change eligibility for other The authors are senior research associates at the Urban Institute in Washington, D.C. 198 July/August Project HOPE The People-to-People Health Foundation, Downloaded Inc. from HealthAffairs.org on November 11, 2018.

2 Medicare Eligibility public programs and increase incentives to participate in them. For example, about 10 percent of young elderly adults qualified for Medicare before age sixtyfive because of disability. 2 Raising the Medicare eligibility age would create strong incentives for the young elderly to apply for disability-related benefits. In addition, some young elderly might become newly eligible for Medicaid coverage through medically needy programs if the loss of Medicare benefits increases out-of-pocket health spending. They would have stronger incentives to participate, because they could not rely on Medicare to cover any of their medical expenses. However, by no longer qualifying for Medicare, they would lose access to the Medicare savings programs, such as the Qualified Medicare Beneficiaries (QMB) and Specified Low-Income Medicare Beneficiaries (SLMB) programs. Raising the eligibility age could also increase employer coverage among the young elderly. Many older Americans lose this coverage when they retire, and those without retiree health benefits sometimes delay retirement until they become eligible for Medicare. 3 Delaying Medicare eligibility could lead more workers to remain in the labor force beyond age sixty-five, to avoid the loss of subsidized employer coverage and the need to purchase expensive continuation coverage or private nongroup coverage before Medicare coverage begins. However, delaying eligibility until age sixty-seven could create a large group of uninsured people ages People with limited incomes and health problems might be unable to find coverage in the absence of Medicare. This problem could worsen if employers cut back on retiree coverage in response to the higher costs of insuring early retirees when the Medicare eligibility age exceeds sixty-five. To address this problem, an increase in the eligibility age could be combined with a Medicare buy-in so that people could purchase Medicare coverage prior to the age of automatic eligibility. Such a plan might especially benefit those with health problems, who often face prohibitively high premiums in the private nongroup market or are denied coverage altogether. However, low- income people might not be able to afford Medicare coverage if premiums were set to cover the expected cost of the services they are likely to use. A subsidized buy-in plan that relates premiums to income might better reduce uninsurance rates. The Clinton administration proposed Medicare buy-ins for all adults ages without access to employer coverage and for displaced workers ages Under the Clinton plan, participants would have paid premiums of about $300 per month while enrolled in the buy-in, plus a supplement to their Medicare Part B premiums after age sixty-five. A similar buy-in plan could be introduced for the young elderly if Medicare eligibility were raised to age sixty-seven. However, analyses of the Clinton plan indicate that it would not have greatly reduced uninsurance rates unless it subsidized premiums for those with low incomes. 4 Study Methods This paper simulates the effects of raising the age of Medicare eligibility to HEALTH AFFAIRS ~ Volume 22, Number 4 199

3 DataWatch sixty-seven in It first measures the impact on the type of insurance coverage and uninsurance rates for the young elderly without a Medicare buy-in and then considers how the availability of a buy-in plan at ages would alter the findings, under a range of premium prices. The analysis incorporates the effects of changes in the availability and cost of insurance coverage on retirement decisions. It also tests the sensitivity of the results to alternative assumptions about how employers might respond to an increase in the Medicare eligibility age. Results are from a detailed microsimulation model, based primarily on the 1998 wave of the Health and Retirement Study (HRS), augmented with new data on the cost of coverage in the private nongroup insurance market. 5 The model incorporates the effects of delayed Medicare eligibility on both retirement and insurance coverage. It simulates retirement behavior as a function of the incremental cost of health insurance premiums associated with retirement under different Medicare scenarios. 6 The model predicts that some workers who would otherwise retire by age sixty-five would instead remain in the labor force if the age of Medicare eligibility were raised. In these cases, we adjust their incomes to reflect the earnings they receive when working and assign them employer-sponsored insurance if they had coverage with their last full-time job. The model also predicts coverage for the young elderly under each possible alternative to Medicare. 7 Disability-related Medicare coverage. We estimate a regression model of disability-related Medicare coverage for HRS respondents ages and use the results to predict disability coverage at ages Medicaid. Based on program rules and respondents income and assets, we simulate eligibility for Medicaid through Supplemental Security Income (SSI) receipt or spending down to medically needy income thresholds. The model assumes that 80 percent of eligible people participate, the upper end of the range of estimated Medicaid participation rates among the elderly, with those with higher medical spending more likely to participate. 8 We assume that the young elderly would not qualify for QMB and SLMB, which are linked to Medicare eligibility. Employer coverage, active or retiree. The model assigns employer coverage from one s own or a spouse s employer if the respondent reported coverage in 1998 under the current system, unless the model already assigned the respondent Medicare disability coverage or Medicaid coverage. Retirees from firms that provide retiree health benefits only until Medicare coverage begins would not report retiree coverage at ages under the current system, but they would probably be covered if the Medicare eligibility age were raised. To account for these cases, the model assigns employer coverage to those who had retiree coverage before age sixty-five or who retired at age sixty-five from jobs offering retiree benefits. However, in sensitivity analyses we assume that some people would lose retiree coverage if the Medicare eligibility age were raised. The model also assigns employer coverage to those who would be eligible for continuation coverage under provisions of the Consolidated Omnibus Budget Reconciliation Act (COBRA). 200 July/August 2003

4 Medicare Eligibility Private nongroup coverage. The model simulates purchase of nongroup coverage among those not already assigned disability-related Medicare, Medicaid, or employer coverage. We use results from a multivariate regression that estimates the effects of premium price, health status, income, and other characteristics on purchase decisions. Premium prices were obtained from an insurance Web site for a standard preferred provider organization (PPO) plan with a $500 deductible and a 20 percent coinsurance rate for in-plan expenses, and they varied by age, sex, smoking behavior, and the presence of chronic medical conditions. 9 Monthly nongroup premiums for the young elderly in our model vary from a low of $273 for women with no serious health problems who do not smoke to a high of $915 for male smokers with at least two serious medical conditions. The model indicates that health status is an important predictor of insurance purchase. Because many adults with poor health status have limited incomes, the model predicts that some low-income people would spend unrealistically large shares of their income on health insurance. We assume that those predicted by the model to spend more than 20 percent of their income on insurance would buy less costly, limited nongroup coverage. 10 Purchase of Medicare coverage. Introduction of a Medicare buy-in could provide a less costly option for standard insurance coverage for those not eligible for public or employer coverage. The model selects the lower of the Medicare buy-in premium and the imputed nongroup premium and uses it to simulate the purchase decision and assign coverage (Medicare, standard nongroup, or limited nongroup). We examine four different pricing schemes for the buy-in program: flat monthly premiums of $400, $300, and $200, and an income-related scheme whereby those with incomes below 150 percent of the federal poverty level would pay $43.80 per month (the Medicare Part B premium in 1998), while those with higher incomes would pay $300 per month. None of the plans we model charge supplemental premiums after age sixty-six. The flat $300 monthly premium is similar to the level set by the Congressional Budget Office (CBO) when it priced the Clinton cost-neutral buy-in plan for adults ages We also estimate the cost of each buy-in plan, net of premium revenue, relative to the cost of the $400 plan. 11 Coverage After Raising The Medicare Eligibility Age Under the current system, 70 percent of young elderly Medicare beneficiaries had some form of supplemental coverage in According to our HRS estimates, 7 percent had Medicaid, 40 percent had employer-sponsored coverage, and 25 percent had privately purchased Medigap coverage. Although not every Medigap policyholder would be able to purchase a primary nongroup plan, these rates suggest that many young elderly could easily maintain some type of coverage if they no longer qualified for Medicare. With a delay in Medicare eligibility to age sixty-seven, 91 percent of the young elderly would remain insured, even if the government did not create a buy-in plan. More than half would receive employer-sponsored benefits, based mostly on their HEALTH AFFAIRS ~ Volume 22, Number 4 201

5 DataWatch own employment (Exhibit 1). Raising the age of Medicare eligibility would increase employer coverage among the young elderly by encouraging many who lose their employer coverage when they retire to remain at work. The model predicts that employment rates would increase by 21 percent among those with employer coverage and without retiree coverage offers. However, employment among all of the young elderly would increase by less than one percentage point (or about 31,000 people ages in 1998), because raising the eligibility age would not greatly affect retirement costs for those without employer coverage or with retiree coverage offers from their employers. Public insurance would continue to be an important source of coverage for the young elderly, even if most did not qualify for Medicare (Exhibit 1). Another 22 percent would purchase private nongroup coverage if they were not eligible for Medicare. Only 9 percent would have lacked coverage in 1998 if the Medicare eligibility age had been raised to sixty-seven. However, another 11 percent would have been underinsured, because they would not have been able to afford standard private nongroup insurance. Impact of Medicare buy-in plans. Many young elderly people would enroll in a Medicare buy-in program if it were offered in combination with an increase in the age of automatic Medicare eligibility. Our model predicts that 15 percent of sixtyfive- and sixty-six-year-olds (or 49 percent of that age group who lack access to other types of public insurance or employer benefits) would purchase Medicare coverage if it were priced at $300 per month (Exhibit 2). Enrollment would rise to 26 percent if priced at $200 per month. Although relatively low monthly premiums would raise plan enrollment, they EXHIBIT 1 Simulated Coverage Rates At Ages In 1998 If The Medicare Eligibility Age Were Raised To 67, With No Medicare Buy-In Plan Type of coverage Percent of population Thousands of people All No insurance Public coverage Medicaid and Medicare Medicaid only Medicare only Employer-sponsored coverage Own employer Spouse s employer Military-related insurance Private nongroup coverage Standard Limited (underinsured) , , SOURCE: Authors simulations, based on the 1998 Health and Retirement Study. 202 July/August 2003

6 Medicare Eligibility EXHIBIT 2 Simulated Coverage Rates At Ages In 1998 If The Medicare Eligibility Age Were Raised To 67, By Medicare Buy-In Plan, And Estimated Cost Of Each Plan Monthly premium for buy-in plan No buy-in plan $400 $300 $200 Income related Type of coverage All No insurance 100.0% % % % % 4.7 Medicaid and/or disability-related Medicare Employer-sponsored coverage Private nongroup coverage Standard Limited (underinsured) Buy-in plan Net cost of plan, in excess of cost of $400 plan ($ million) $563 $957 $913 SOURCE: Authors simulations, based on the 1998 Health and Retirement Study. also would increase costs for the federal government. For example, reducing the monthly premium from $400 to $300 would cost the federal treasury about $563 million in lost revenue (in 1998 dollars). Charging low-income participants $43.80 instead of $300 per month would cost another $350 million. Enrollment in the buy-in plan would exceed the number of young elderly people who gain coverage, because most participants would substitute Medicare for more costly (or less comprehensive), private nongroup coverage. For example, the introduction of a $300 per month buy-in plan would reduce the uninsurance rate at ages by less than two percentage points (from 9 percent to just over 7 percent), while the share of young elderly with private nongroup coverage would fall from 22 percent to 9 percent. Buy-in plans would, however, improve the quality of coverage for the young elderly. With a $300 buy-in plan, only 11 percent of those ages would lack standard insurance coverage, compared with 20 percent if the eligibility age were raised without the option to buy into Medicare. Characteristics of buy-in participants. Regardless of how the Medicare buyin plan was priced, many participants would have relatively high incomes and serious health conditions. If the buy-in plan were priced at $300 per month, 62 percent of participants would have family incomes exceeding 400 percent of the federal poverty level, while only 13 percent would have incomes below 200 percent of poverty (Exhibit 3). By comparison, adults with incomes below 200 percent of poverty make up 29 percent of those without access to other types of public insurance or employer-sponsored health benefits, the pool of young elderly people likely to enroll in the buy-in plan. Thus, a buy-in plan without income subsidies would attract partici- HEALTH AFFAIRS ~ Volume 22, Number 4 203

7 DataWatch EXHIBIT 3 Characteristics Of Buy-In Participants Ages In 1998 Under Alternative Medicare Buy-In Plans No other public or employer-sponsored insurance Enrolled in $300 plan Enrolled in incomerelated plan Characteristic Percent Thousands Percent Thousands Percent Thousands Income relative to poverty Less than 100% 100% 199% 200% 399% 400% or more Total Number of serious health problems None One Two or more Total Overall health status Excellent Very good Good Fair Poor Total 8.6% , % % , , SOURCE: Authors simulations, based on the 1998 Health and Retirement Study. NOTE: Components do not always add to total because of rounding. pants with disproportionately high incomes. Even if premiums were related to income, about half of participants would have incomes exceeding 400 percent of poverty, compared with only 42 percent among all potential participants. Because people in poor health exhibit strong demand for health insurance, the prevalence of health problems would be somewhat higher among those who participate in Medicare buy-in plans than among all those ages For example, 61 percent of participants in the $300 buy-in plan would have one or more serious health problems, compared with only 40 percent of those without employer or public insurance. This adverse-selection problem, in which insurance costs are driven up because those most likely to use services are also most likely to purchase insurance, highlights the difficulty of designing a cost-neutral buy-in program. Uninsurance among vulnerable populations. A raise in the Medicare eligibility age would greatly increase rates of uninsurance among blacks, Hispanics, high school dropouts, and those with family incomes below 200 percent of poverty (Exhibit 4). For example, 34 percent of Hispanics and 26 percent of blacks ages would become uninsured, compared with only 6 percent of non-hispanic whites. Income differentials in coverage are also striking. We estimate that 23 percent of the young elderly poor would lack coverage, compared with only 3 percent of those with family incomes above 400 percent of poverty. Rates of uninsurance would not differ 204 July/August 2003

8 Medicare Eligibility EXHIBIT 4 Simulated Uninsurance Rates At Ages In 1998 If The Medicare Eligibility Age Were Raised To 67, Under Alternative Medicare Buy-In Plans Percent of sample No buy-in $300 buy-in systematically by health status. High rates of disability coverage among those in poor health would offset low rates of private health insurance coverage. Vulnerable populations, such as racial minorities and the poor, would benefit greatly from the introduction of a Medicare buy-in program only if premiums were related to income. For example, with a buy-in premium of $300, the rate of uninsurance among Hispanics would decrease by only one percentage point. Under the income-related premium plan, the uninsurance rate would fall by eight percentage points for blacks, six percentage points for Hispanics, and four percentage points for whites, relative to the scenario with no buy-in option. Relating premiums for the Medicare buy-in plan to income would especially increase rates of coverage for people with limited incomes. If Medicare eligibility were delayed, the introduction of a $300 buy-in plan would not reduce unin- Incomerelated plan All 100.0% 9.1% 7.4% 4.7% Income relative to poverty Less than 100% 100% 199% 200% 399% 400% or more Self-rated overall health status Excellent Very good Good Fair Poor Number of serious health conditions None One Two or more Race/ethnicity Black Hispanic White and other Educational attainment Not high school graduate High school graduate Some college College graduate SOURCE: Authors simulations, based on the 1998 Health and Retirement Study. NOTES: The first data column reports the percentage of those ages with the given characteristic. Other cells indicate the percentage in that age group who would be uninsured under each buy-in plan scenario, if Medicare eligibility were delayed until age 67. HEALTH AFFAIRS ~ Volume 22, Number 4 205

9 DataWatch surance rates for those in poverty at all, whereas it would reduce uninsurance for those with incomes above 400 percent of poverty, from 3 percent to 0.4 percent. However, if premiums were instead related to income, uninsurance rates would fall to 9 percent for those in poverty and remain at 0.4 percent for those with incomes above 400 percent of poverty. These low-income subsidies would cost the federal government about $350 million per year (in 1998 dollars). The introduction of a Medicare buy-in plan at age sixty-five would greatly reduce rates of uninsurance among people with health problems, even if premiums were not related to income. If all participants faced $300 monthly premiums, the buy-in plan would reduce uninsurance rates from 7 percent to only 4 percent for people with two or more health problems and from 19 percent to 15 percent for people who described their overall health status as fair or poor. The effectiveness of the buy-in plan results from the strong demand for health insurance among people with health problems and from high premiums in the private nongroup market. Retiree coverage. Because a delay in Medicare eligibility would increase the costs of providing retiree coverage, some employers might respond by dropping it altogether. In the unlikely event that all employers dropped their plans, the loss of Medicare benefits would greatly change coverage rates for the young elderly population. Under this worst-case scenario, and in the absence of a buy-in plan, 14 percent would become uninsured, 43 percent would purchase insurance through the nongroup market, and 26 percent would maintain some form of employer-sponsored coverage, through their own employment, spousal employment, COBRA coverage, or military benefits (Exhibit 5). The elimination of retiree benefits would also increase employment by 4 percent among the young elderly overall (and by 80 percent among those with employer coverage), relative to simulated employment rates assuming an increase in eligibility age but no erosion in retiree coverage. The effects would be much smaller under a less extreme scenario in which 40 percent of those in small firms and 20 percent of those in large firms lose their retiree coverage. Under this assumption, the share with employer coverage falls to 45 percent, while the share purchasing nongroup coverage increases to 29 percent and the share without any coverage rises to 10 percent. Our estimates of uninsurance are less sensitive to alternative assumptions about the erosion of retiree benefits in the presence of a buy-in plan, because many who lose retiree coverage would participate in the buy-in. Relative to rates under the assumption that no employers drop retiree coverage, estimated uninsurance rates for the young elderly would increase by only three percentage points if all employers drop retiree coverage, if premiums for the buy-in were related to income (Exhibit 5). Discussion Drawbacks of raising the eligibility age. If Medicare eligibility were delayed to age sixty-seven, most sixty-five- and sixty-six-year-olds would find alternative 206 July/August 2003

10 Medicare Eligibility EXHIBIT 5 Simulated Coverage Rates At Ages In 1998 If The Medicare Eligibility Age Were Raised To 67, Under Alternative Assumptions About Erosion Of Retiree Health Insurance (RHI) And Alternative Medicare Buy-In Plans Percent covered Type of coverage No employers drop RHI a Some employers drop RHI b All employers drop RHI No insurance Medicaid and/or disability-related Medicare 9.1% 9.9% 14.0% Employer-sponsored coverage Own employer Spouse s employer Military-related insurance Private nongroup coverage Standard Limited (underinsured) Premium Percent uninsured Monthly buy-in premium No buy-in $300 Income-related 9.1% % % SOURCE: Authors simulations, based on the 1998 Health and Retirement Study. a Base scenario. b The model assumes that 40 percent of people who work in small firms and 20 percent who work in large firms lose RHI. sources of coverage, even if they were not permitted to buy into the Medicare system. Nonetheless, about 9 percent of the young elderly would become uninsured if they no longer qualified for Medicare, and 11 percent would be underinsured. The impact would be much greater for certain segments of the population such as Hispanics, blacks, and the poor. Another drawback might be that delaying Medicare eligibility could raise employers cost of providing coverage to retirees who do not yet qualify for Medicare. If some employers respond by dropping coverage, uninsurance rates for the young elderly could be even higher. Our estimates suggest that the possible erosion in retiree coverage would have minor effects on uninsurance rates, especially if people could buy into the Medicare program, because most of those who lose retiree benefits would purchase alternative types of coverage. However, the loss of retiree coverage would increase insurance costs for young elderly retirees. Even if employers do not drop coverage, they could increase cost-sharing requirements, which might lead some to forgo their retiree benefits. Need for a Medicare buy-in. These findings underscore the need for welldesigned policy initiatives to help the young elderly obtain insurance coverage if they no longer qualify for Medicare benefits. Our results suggest that such a plan would attract many participants but would greatly reduce rates of uninsurance only HEALTH AFFAIRS ~ Volume 22, Number 4 207

11 DataWatch if premiums were subsidized for the low-income population. Without subsidies, however, most buy-in participants would merely replace expensive nongroup coverage with lower-price Medicare coverage. Thus, the success of a Medicare buy-in program would depend upon the level of subsidies provided. But higher subsidies would raise the costs of the program and could also be expensive to implement, because of the difficulty of verifying income eligibility. Dealing with adverse selection. Our results also suggest that adverse selection into the buy-in program would complicate efforts to keep costs down while ensuring access to those with limited incomes. Participation rates would be higher among people in poor health than among those in good health. Across-the-board subsidies, which encourage enrollment by those in better health with low expected costs, mitigate the adverse selection problem. The cost to taxpayers increases, however, when the program charges lower prices. Accruing savings to Medicare. Because the young elderly are the least costly group of Medicare beneficiaries, the cost savings generated by an increase in the eligibility age would not keep pace with the reduction in program participation. 12 Many disabled beneficiaries ages would keep Medicare coverage if the age of eligibility were raised. As a result, program savings would accrue only from dropping coverage for relatively young, healthy beneficiaries who cost Medicare relatively little. Over time, however, these cost savings can become substantial, as both the Medicare-eligible population and expected per capita health care costs grow. According to one estimate, raising the age of eligibility to sixty-seven would save Medicare almost $28 billion per year (in 2000 dollars) by Setting an age for medicare eligibility is necessarily arbitrary. Although a clear national consensus exists for providing universal health benefits to older Americans, it is not clear when a person becomes old. 14 As health status improves at late midlife and many workers begin to delay retirement, the need for subsidized health benefits at age sixty-five may be less urgent now than it was in 1966 when Medicare was introduced. 15 As the normal retirement age for Social Security benefits slowly rises to sixty-seven, it may make sense to increase the Medicare eligibility age at the same time, since the eligibility age always coincided with the normal retirement age (before the retirement age began increasing in January 2000). However, our results suggest that many people would benefit from an option to buy into the Medicare program at younger ages. Similar to the early retirement option for Social Security, the buy-in program would permit people to receive limited subsidies for benefits before the full entitlement age. The authors thank Landon Jones and Rumki Saha for excellent programming assistance, and Marilyn Moon for helpful comments on the paper. This project was supported by the Henry J. Kaiser Family Foundation. The opinions expressed are the authors and do not reflect the positions of the Urban Institute or its funders. 208 July/August 2003

12 Medicare Eligibility NOTES 1. In 1995 the Bipartisan Commission on Entitlement and Tax Reform proposed to raise the Medicare eligibility age to seventy. Two years later the U.S. Senate approved an increase to sixty-seven as part of the 1997 Balanced Budget Act, but this provision was dropped. The 1999 National Bipartisan Commission on the Future of Medicare also recommended raising the eligibility age. 2. See M. Moon et al., Potential Effects of Medicare Program Changes on Medicaid Expenditures and Insurance Coverage, Final Report to the Health Care Financing Administration (Washington: Urban Institute, 2000). 3. See, for example, D.M. Blau and D.B. Gilleskie, Retiree Health Insurance and the Labor Force Behavior of Older Men in the 1990s, Review of Economics and Statistics (February 2001): 64 80; J. Gruber and B.C. Madrian, Health Insurance Availability and the Retirement Decision, American Economic Review (September 1995): ; and R.W. Johnson, A.J. Davidoff, and K. Perese, Health Insurance Costs and Early Retirement Decisions, Industrial and Labor Relations Review (July 2003): R.W. Johnson, M. Moon, and A.J. Davidoff, A Medicare Buy-In for the Near Elderly: Design Issues and Potential Effects on Coverage (Washington: Henry J. Kaiser Family Foundation, 2002); and J. Sheils and Y-J. Chen, Medicare Buy-In Options: Estimating Coverage and Costs (New York: Commonwealth Fund, 2001). 5. The HRS is a large longitudinal survey of middle-age Americans conducted by the University of Michigan for the National Institute on Aging. The 1998 wave includes 1,303 respondents ages See Institute for Social Research, Health and Retirement Study, hrsonline.isr.umich.edu (21 January 2003). 6. Following Johnson et al., Health Insurance Costs and Early Retirement Decisions, we quantify the effects of changes in the Medicare eligibility age on labor-force withdrawals by computing the premium cost associated with retirement. We define this cost as the monthly increase in premium expenses that workers would pay if they retired, compared with what they would pay if they remained at work, and we compute the net present value of the stream of costs from the age at which workers consider retirement until they reach the Medicare eligibility age, adjusting for survival probabilities. We include the premium cost in probit models of retirement, and we use the estimated parameters from the model to simulate retirement behavior under an increase in the Medicare eligibility age, with and without the introduction of Medicare buy-in plans. For each reform, we recompute the premium cost to retire and use the new premium costs to predict retirement behavior. 7. Additional details are available from Amy Davidoff, adavidof@ui.urban.org. 8. M. Moon, N. Brennan, and M. Segal, Options for Aiding Low-Income Medicare Beneficiaries, Inquiry (Fall 1998): See Insweb, Instant Individual and Family Health Insurance Quotes, qsindmedindex (20 May 2003). (Quotes no longer vary by chronic medical condition.) 10. Health insurance costs exceed 20 percent of total household expenditures in only 2 percent of households, according to our estimates from the 1998 Consumer Expenditure Survey. 11. Combining self-reported utilization data from the HRS and per unit cost estimates from the Medicare Current Beneficiary Survey (MCBS), we measure 1998 expenditures on hospital stays and physician services for those who we predict would participate in each plan. We then compute the share of Medicare costs, among sixty-five- and sixty-six-year-olds, spent by buy-in enrollees, assuming that Medicare spending is proportional to hospital and physician expenditures. Combining this information with the estimated share of total 1998 Medicare expenditures consumed by people ages (from the Medical Expenditure Panel Survey) and total Medicare expenditures, we calculate the total cost for each buy-in plan. We then subtract premium revenues to estimate net federal subsidies. 12. T.A. Waidmann, Potential Effects of Raising Medicare s Eligibility Age, Health Affairs (Mar/Apr 1998): D.C. Wittenburg, D.C. Stapleton, and S.B. Scrivner, How Raising the Age of Eligibility for Social Security and Medicare Might Affect the Disability Insurance and Medicare Programs, Social Security Bulletin 63, no. 4 (2000): Henry J. Kaiser Family Foundation, National Survey on Medicare: The Next Big Health Policy Debate? (Menlo Park, Calif.: Kaiser Family Foundation, 1998). 15. E.M. Crimmins, S.L. Reynolds, and Y. Saito, Trends in Health and Ability to Work among the Older Working-Age Population, Journal of Gerontology: Social Sciences (January 1999): S31 S40; and J.F. Quinn, Retirement Patterns and Bridge Jobs in the 1990s, Issue Brief no. 206 (Washington: Employee Benefit Research Institute, 1999). HEALTH AFFAIRS ~ Volume 22, Number 4 209