DataWatch The Cost Of Providing Health Care Benefits To Early Retirees by Paul Fronstin, Sara C. Snider, William S. Custer, and Dallas L. Salisbury Abstract: Subsidized health care benefits would be guaranteed for early retirees ages fifty-five to sixty-four under President Bill Clinton s health care reform proposal. This is an important policy issue because persons in this age group are the least likely age cohort of the nonelderly population to be working. They are also the most likely to face high and uncertain health care costs. Previous research has shown that access to continuation of health insurance coverage encourages retirement before age sixty-five. These retirement effects can add substantially to the federal government s cost ofproviding health care benefits to early retirees. Based upon various assumptions for premium levels, the induced retirement effects of current workers, and the number of nonworkers qualifying for subsidized benefits, we present tota1 annual cost estimates to the federal government, based on 1994 figures, ranging from $9.1 billion to $19.6 billion, to provide subsidized benefits to the nonworking population between the ages of fifty-five and sixty-four. Under President Bill Clinton s health reform proposal, early retirees between the ages of fifty-five and sixty-four would be eligible for government-subsidized postretirement health benefits beginning in 1998. 1 The plan would guarantee the basic benefit package for all nonworkers ages fifty-five through sixty-four, provided they have met all of the requirements for Social Security benefits except for age. 2 The federal government would pay 80 percent of the average weighted premium, freeing firms and individuals from paying a major portion of the costs. 3 This premium would be community rated and would not vary by age. For retirees with guaranteed benefits from their employer, the employer would cover the retiree s share of the premium. 4 Health insurance costs are potentially a large percentage of retirees expenditures. This plan would encourage early retirement among workers by reducing their health insurance costs significantly. Large savings would also accrue to firms that now provide early retiree health benefits, resulting in incentives for employers to encourage early retirement, especially in firms with an older work force. 5 The authors are affiliated with the Employee Benefit Research Institute in Washington, D. C. : Paul Fronstin is a research associate, Sara Snider is a research analyst, William Custer i s director of research, and Dallas Salisbury i s president.
D ATAWATCH 247 Providing health care benefits for early retirees is an important policy issue because persons within this age group are already the least likely to be working and the most likely to face uncertain health care expenditures when compared with other nonelderly age groups. 6 The near-elderly population is a relatively high-risk population that does not qualify for Medicare benefits; therefore, privately purchased health insurance can be a costly commodity. The availability of low-cost retiree health insurance is an incentive for the near-elderly to retire early. It is almost certain that the provision of health care benefits to early retirees would reduce the labor-force participation of this group and drive up the federal government s costs of providing health care benefits. These costs must be considered in the context of the current retired population that would qualify for benefits as well as the working population that may be induced to leave the labor force. Rising total costs of providing health benefits to early retirees will become even more significant as the population ages. This DataWatch reports estimates of the costs to the federal government of providing subsidized health care benefits to nonworking persons ages fifty-five to sixty-four, based on 1992 population estimates. A range of estimates is presented based on assumptions about the premium cost and the number of persons who could be affected by this proposal. Characteristics Of The Near-Elderly Population In 1992 the near-elderly comprised 8 percent (21.2 million) of the total U.S. population. 7 Exhibit 1 contains data on the major activity of the near-elderly in 1992. According to the data, 62.2 percent were working, and 1.5 percent could not find work. Nine percent did not work because Exhibit 1 Major Activity Of Persons Ages Fifty-Five To Sixty-Four, 1992 Source: Employee Benefit Research Institute tabulations of the March 1993 supplement to the Current Population Survey.
248 HEALTH AFFAIRS Spring (II) 1994 they were either ill or disabled, 14.5 percent were retired, and 12.2 percent were taking care of their home or family. Approximately 0.01 percent never worked. If the near-elderly are guaranteed subsidized health benefits when they retire early, the number of workers will decline. In addition, nonwork ers ages fifty-five to sixty-four also will be eligible for the subsidized benefits if they have met all of the requirements for Social Security except for age. Although the near-elderly have lower labor-force participation rates than do other age groups, the majority of those ages fifty-five to sixty-four (64.1 percent) get their health insurance coverage from an employmentbased plan (Exhibit 2). This compares with 70.6 percent for persons ages Exhibit 2 Selected Sources Of Health Insurance, By Age Group And Major Activity, 1992 Source: Employee Benefit Research Institute tabulations from the March 1993 supplement to the Current Population Survey. Note: Details may not add to totals because persons may receive coverage from more than one source.
D ATAWATCH 249 forty-five to fifty-four and 66.6 percent for persons ages twenty-five to forty-four. The near-elderly are more likely than other age groups to have individually purchased private insurance plans. Almost 13 percent have such a policy, compared with 8.8 percent of persons ages forty-five to fifty-four, and 7.1 percent of persons ages twenty-five to forty-four. The near-elderly s high rate of individual coverage is a result of their weak attachment to the labor force and their increased likelihood of being disabled. They are less likely to have employment-based health coverage, yet they are more likely than others to need some form of health insurance. 8 The near-elderly are second only to the elderly in likelihood of having publicly provided health insurance coverage. In 1992 more than 17 percent of the near-elderly had some form of public coverage, compared with 10 percent of the population ages forty-five to fifty-four, and 9.8 percent of the population ages twenty-five to forty-four. The major source of public insurance for the near-elderly is Medicare; the near-elderly are more likely to have Medicare coverage because they are more likely to be disabled and to qualify for the Social Security Disability Insurance (SSDI) program. Because of their higher disability rates, the near-elderly are less likely than other nonelderly age groups to participate in the labor force. The near-elderly were less likely than other nonelderly age groups to be uninsured in 1992, Less than 13 percent of the population ages fifty-five to sixty-four were uninsured, compared with 14 percent of persons ages fortyfive to fifty-four and 18.5 percent of persons ages twenty-five to forty-four. The higher rates of insurance coverage also can be attributed to higher SSDI participation. Sources of health insurance coverage differ by the work status of the near-elderly. Workers between ages fifty-five and sixty-four were most likely to have employment-based insurance, but many nonworkers also had employment-based coverage. For example, 40.3 percent of early retirees had direct employment-based health insurance in 1992. 9 These benefits can come in the form of retiree health insurance under the Consolidated Omnibus Budget Reconciliation Act of 1985 (COBRA). 10 Interestingly, 11.5 percent of persons in this age group who were looking for work were covered by an employment-based plan. These persons were most likely using COBRA coverage as a bridge during job turnover. Cost Of Providing Health Benefits To Early Retirees Data and methods. We used the March 1993 supplement to the Current Population Survey (CPS) to provide estimates of the federal government s cost of providing subsidized health care benefits to nonworkers ages fifty-five to sixty-four. The CPS is the most detailed nationally repre-
250 HEALTH AFFAIRS Spring (II) 1994 sentative survey providing data on economic and demographic variables, including health insurance coverage, early retirement, and pension plan participation. To provide a range of estimates of the costs of insuring early retirees, it is necessary to predict the number of people who will be induced to leave the labor force. We also must know how many persons have already stopped working and would qualify for Social Security benefits at age sixty-five. We used results obtained by Jonathan Gruber and Brigitte Madrian to estimate the effects of induced early retirement. 11 These researchers found that state laws mandating the availability of COBRA-type continuation of coverage for one year result in a 20-60 percent increase in the probability that a person age fifty-five to sixty-four will retire early. Since there are strong incentives to retire early when state laws mandate continuation of coverage, we would expect even greater effects under the president s reform package because continuation of coverage would be guaranteed for a longer period of time and the cost savings to early retirees would be even greater. The Clinton administration has estimated that under the president s proposal, the cost of an indemnity plan for a single adult would be $1,932 in 1994. Other researchers have presented alternative estimates. For example, the Employee Benefit Research Institute (EBRI) estimated a premium of $2,202. Researchers from The Wyatt Company estimated a premium of $2,285; Hewitt Associates estimated the premium to be $2,440; and the Health Insurance Association of America (HIAA) estimated a premium of $2,509. 12 As mentioned above, the federal government s share of the premium would be 80 percent; therefore, we used 80 percent of the lowest estimate ($1,547) and 80 percent of the highest estimate ($2,007) to derive a range of estimates of the federal government s cost of providing health benefits to early retirees. Based upon data from the March 1993 CPS, we calculated a retirement rate of 6.9 percent, representing 820,000 early retirees ages fifty-five to sixty-four, and assumed that if the number of early retirees does not change, 6.9 percent of workers in this age group would retire in any given year. Exhibit 3 presents the results of our cost simulations under various assumptions about the premium level and the retirement effects of workers. We give estimates based upon three assumptions about the retirement effects of workers combined with two assumptions about the number of nonworkers who also would qualify for subsidized benefits. Results. Our estimates of the costs to the federal government of providing subsidized benefits to early retirees and nonworkers range from $9.1 billion to $19.6 billion. Approximately 90 percent of workers ages fifty-five to sixty-four would currently qualify for Social Security benefits; therefore, we weight our estimates by 90 percent. 13 Because we do not know how
D ATAWATCH 251 Exhibit 3 Cost Estimates Of Providing Health Care Benefits To Early Retirees, Using Various Assumptions About Premiums And Retirement Effects, By Work Status, 1994 Sources: Employee Benefit Research Institute simulations using the March 1993 Current Population Survey; J. Gruber and B.C. Madrian, Health Insurance Availability and the Retirement Decision, NBER Working Paper 4469 (Cambridge, Mass.: NBER, 1993); and premiums from the Health Security Act and the Health Insurance Association of America. Notes: Billions of dollars. Cost estimates are based on 1994 figures. many nonworkers would qualify for subsidized benefits, we use 50 percent as a lower bound and 90 percent as an upper bound to derive a range of costs. When we assume that workers do not respond to the added incentives to retire, the cost of providing health benefits to workers who would have retired even without the subsidized benefits is $1.2 billion under the low premium and $1.5 billion under the high premium. This simulation allows us to put a lower bound on the cost estimate, even though we expect some change in the number of workers who retire. For nonworkers, assuming that 50 percent would qualify for subsidized benefits, we find that the cost to the federal government for providing the benefits to the nonworkers is $7.9 billion under the low premium and $10.3 billion under the high premium. 14 If 90 percent of the nonworkers qualified for subsidized benefits, the cost to the federal government for subsidizing nonworkers would be $11.4 billion under the low premium. The cost for nonworkers jumps to $14.8 billion under the high premium. The model also is simulated using a moderate level of induced-retirement for workers, based on previous research. Gruber and Madrian found that one year of COBRA-type continuation of coverage results in an increase of 1.5 percentage points in the probability of retiring early. Under the presi-
252 HEALTH AFFAIRS Spring (II) 1994 dent s health reform plan, early retirees would be guaranteed subsidized benefits until age sixty-five; thus, we use the average number of months until a person reaches age sixty-five to predict the probability of retiring early. Based on Gruber and Madrian s research, the moderate probability of retiring early becomes 15.9 percent (up from 6.9 percent), representing 1.3 million additional early retirees, and the cost of providing health benefits to early retirees ranges from $2.9 billion to $3.8 billion, depending upon the level of the premium. A large retirement effect among current workers is also possible. Gruber and Madrian do not control for the additional effects that occur when the federal government subsidizes 80 percent of the cost of community-rated health insurance premiums. Under a continuation-of-coverage mandate, group-rated health insurance will be available for only approximately eighteen months; under the president s plan, continuation of coverage will be available until age sixty-five at only 20 percent of the average weighted community-rated premium. 15 This lower price of health insurance should have some additional effects on the cost to the federal government of providing subsidized benefits to early retirees. Assuming that the largest possible response to the guaranteed benefits package would be a 20 percent increase in the number of workers who decide to retire early (representing 1.8 million additional early retirees), we find that the range of costs to the federal government of providing benefits to workers who are induced to retire becomes $3.7 billion under the low premium and $4.8 billion under the high premium. 16 Policy Implications There are several reasons for the range of cost estimates we have derived for early retiree coverage. Costs are incurred because approximately 6.9 percent of the near-elderly population retires each year. Under the Clinton plan, additional costs will be incurred when workers retire because the federal government will pay a majority of the costs of the guaranteed benefits. Nonworkers who have met the Social Security covered-quarters requirement also will be covered under the president s reform plan. All of these costs are uncertain because the level of induced retirement, the number of qualifying nonworkers, and the actual premium costs cannot be precisely estimated. The lowest total annual cost the federal government would face is $9.1 billion, assuming that the premium is $1,932, additional workers do not retire early, and only 50 percent of nonworkers qualify for the subsidized benefits. Under the assumptions that the premium for a single adult is $2,509, 20 percent of workers retire early, and 90 percent of the nonworking population ages fifty-five to sixty-four qualifies for subsi-
D ATAWATCH 253 dized benefits, the highest total annual cost the federal government would face becomes $19.6 billion. While we know that. the costs to the federal government of providing these benefits are large and that workers will retire early if they are guaranteed subsidized health care benefits, these benefits also will change the process of retirement. If a person now retires early, he or she can always return to work. Some workers use early retirement as a way to ease themselves out of the labor force. Many of these workers retire only partially, finding part-time work until they are closer to age sixty-five. Under the Health Security Act, retired persons would lose their subsidized benefits if they returned to work. The proposed employer mandate would inhibit early retirees from returning to work because as employees they would not get subsidized benefits and their employers would have to provide them with health benefits. The human capital costs also must be accounted for when estimating the costs of providing subsidized benefits to workers ages fifty five to sixty-four. NOTES 1. The Health Security Act, if passed, would be fully implemented by 1998, but some states would have the option to participate beginning in 1996. 2. To qualify for Social Security benefits, a person must have worked at least forty quarters in Social Security-covered jobs. Under the president s proposal, single adults ages fifty-five to sixty-four with income over $90,000 and married couples with joint income of more than $115,000 would have to repay the subsidy at tax time. Only 2 percent of nonworking individuals ages fifty-five to sixty-four would have to repay the subsidy. The estimates presented here do not account. for these persons because their effect on the total costs to the federal government is very small. 3. The weighted average premium represents the average premium of all available health plans in an alliance weighted by the number of persons enrolled in each plan. 4. See J. Davis, Retiree Health Benefits: Issues of Structure, Financing, and Coverage, Issue Brief 112 (Washington: Employee Benefit Research Institute, March 1991) for ways in which employers now fund retiree health benefits. 5. For a discussion of the transition to retirement, see C.J. Ruhm, The Work and Retirement Patterns of Older Americans, Issue Brief 121 (Washington: EBRI, December 1991). 6. Between 1955 and 1985 the labor-force participation rate of males ages fifty-five to sixty-four declined by more than 19 percent, compared with a 4.5 percent decrease for males ages forty-five to fifty-four. Some researchers attribute this decline to an increase in the ease of qualifying for disability benefits. See D.O. Parsons, The Decline in Male Labor Force Participation, Journal of Political Economy (February 1980): 117-134; and J. Bound, The Health and Earnings of Rejected Disability Insurance Applicants, American Economic Review (June 1989): 482-503, for alternative theories on the decline in the male labor-force participation rate. 7. U.S. Department of Commerce, Bureau of the Census, Projections of the Population of the United States by Age, Sex, and Race: 1988 to 2080, Current Population Reports, Series P-25, no. 1018 (Washington: U.S. Government Printing Office, 1989). 8. To qualify for disability benefits, a person s disability should be expected to last at least
254 HEALTH AFFAIRS Spring (II) 1994 twelve months. In addition, there is a mandatory five-month waiting period before disability benefits begin; therefore, disabled persons are more likely to need some form of individual coverage in the interim. 9. Direct employment-based coverage is coverage in one s own name. Indirect employ ment-based coverage is coverage provided through someone else s employer. 10. The Consolidated Omnibus Budget Reconciliation Act (COBRA) of 1985 requires employers with health insurance plans to offer continued access to group health insurance to qualified beneficiaries. COBRA requires continued access for eighteen months for employees (twenty-nine months for the disabled) and thirty-six months for qualifying spouses and dependent children. Some states also have continuation-ofcoverage laws. See EBRI, Fundamentals of Employee Benefit Programs, 4th ed. (Washington: EBRI, 1990), for more information about the federal law; and J. Gruber and B.C. Madrian, Health Insurance Availability and the Retirement Decision, NBER Working Paper 4469 (Cambridge, Mass.: National Bureau of Economic Research, 1993), for a listing of state laws. 11. Gruber and Madrian, Health Insurance Availability and the Retirement Decision. The authors provide estimates predicting the probability that a male would retire early based upon individual characteristics, whether or not the person had employmentbased health insurance, whether the person was covered by a pension plan, and the number of months for which the person s state of residence mandated continuation of health insurance. We adjust their equation to reflect the number of months until a person qualifies for Medicare, which reflects COBRA-type continuation of coverage until age sixty-five. We generalize their results to females, although this may overstate the retirement effects of females. 12. The White House, Health Security: The President s Report to the American People (White House Domestic Policy Council, 1993); EBRI simulations using the March 1993 supplement to the Current Population Survey (CPS) and the 1987 National Medical Expenditure Survey, adjusted for inflation and imputed to the CPS; The Wyatt Company, The Economics of Health Reform (Report prepared for the Business Council on National Health Policy, January 1994); testimony of Dale Yamamoto and Frank McArdle, Hewitt Associates, House Energy and Commerce Subcommittee on Health and the Environment, 22 November 1993; and Health Insurance Association of America, Premiums under the Proposed Health Security Act, Actuarial Memorandum (Washington: HIAA, 31 January 1994). 13. We use data from the 1984 Survey of Income and Program Participation (SIPP) to determine the number of workers ages fifty-five to sixty-four now qualifying for Social Security. 14. Based on our data for workers, for nonworking persons reporting that they are retired, we assume that approximately 90 percent of the nonworkers would qualify for Social Security at age sixty-five. 15. This will be longer for persons in some states with continuation-of-coverage law lengths that exceed the federal government length set in COBRA. 16. The Clinton administration gives estimates for induced retirement in the range of 350,000 to 600,000, yielding a figure of $10 billion in government costs for retiree benefits, based on 1998 estimates. Analysis conducted by the Congressional Budget Office also suggests effects within this range, but closer to the upper end. CBO, An Analysis of the Administration s Health Proposal (Washington: CBO, February 1994). Both the CBO and the administration s estimates are roughly consistent with the results in B. Madrian, Labor Market Effects of Employment-Based Health Insurance (Doctoral dissertation, Massachusetts Institute of Technology, 1993). Our range is based upon estimates in Gruber and Madrian, who suggest much larger responses. Gruber and Madrian, Health Insurance Availability and the Retirement Decision.