Health Insurance for Americans Approaching Age Sixty-five: An Analysis of Options for Incremental Reform

Transcription

1 Health Insurance for Americans Approaching Age Sixty-five: An Analysis of Options for Incremental Reform Pamela Farley Short, Dennis G. Shea, and M. Paige Powell The Pennsylvania State University Abstract This article provides a systematic evaluation of the options for incremental health insurance reforms aimed at older Americans nearing age sixty-five. It presents three basic arguments for giving special consideration to this age group: (1) early retirement and its effect on access to employer insurance; (2) changes in health and health care expenses associated with increasing age; (3) the vulnerability to unexpected economic or health shocks that will affect people throughout their retirement. The analysis of policy options begins by specifying criteria for evaluating alternative approaches to reform. The proposed criteria emphasize that reforms for this age group should be designed to fit with other financial plans and decisions made during such a transitional stage of life. Policy options should be judged according to fundamental goals such as equity and efficiency, not simply ranked according to the number of uninsured who will gain coverage. After offering a comprehensive catalog and evaluation of available options, the analysis identifies and discusses a preferred approach which preserves choices while offering universal and subsidized access to Medicare before age sixty-five. Since the Clinton administration s failed attempt to bring about universal coverage and comprehensive health care financing reform, the United States has proceeded in a more incremental fashion with legislation to improve the accessibility and affordability of health insurance. Congress Supported by a grant from the Commonwealth Fund, a New York City based private independent foundation. The views presented here are those of the authors and not necessarily those of The Pennsylvania State University or the Commonwealth Fund, its directors, officers, or staff. The authors are grateful for assistance provided by Sherry Glied and Shana McCormack at Columbia University, Cathy Schoen and Karen Davis at the Commonwealth Fund, the Actuarial Research Corporation, and Steve Maczuga at the Population Research Institute at Penn State. Journal of Health Politics, Policy and Law, Vol. 28, No. 1, February Copyright 2003 by Duke University Press.

2 42 Journal of Health Politics, Policy and Law enacted federal insurance market reforms in the Health Insurance Portability and Accountability Act (HIPAA) of 1996, the State Children s Health Insurance Program (SCHIP) in the Balanced Budget Act of 1997, and a variety of changes to Medicaid and Medicare. Expanding and subsidizing the health insurance of older Americans who are nearly old enough to qualify for Medicare eligibility is one of several other incremental reforms that have been offered for consideration. In this article, we provide a systematic evaluation of incremental coverage reforms for older Americans under age sixty-five. We begin by discussing the reasons for considering special reforms to help this age group. Then we identify a comprehensive set of policy options, evaluate the options against explicit policy goals and evaluation criteria, and make a set of policy recommendations. Why Consider Special Reforms for Older Americans? There are three basic arguments for giving special consideration to the health insurance needs of the population that is nearly age sixty-five. First, because many people retire early, before they qualify for full Social Security benefits and Medicare at age sixty-five, access to employer insurance is diminished in the years leading up to Medicare. This leaves older Americans to depend more heavily than younger age groups on the individual insurance market, where coverage for preexisting conditions is limited and premiums are much higher than for similar benefits purchased through groups. Second, because health declines and health expenditures increase with age, the age group closest to age sixty-five faces the greatest risk from being uninsured and the highest premiums in the individual market. Finally, older Americans are especially vulnerable to unexpected and unwelcome changes in their health, health insurance, jobs, or employment status that could leave them poorly positioned for old age. Although age sixty-five was the typical retirement age when Medicare was enacted (Hurd 1990), labor force participation now declines markedly before age sixty-five. Indeed, more of today s Social Security recipients began receiving reduced benefits when they first qualified at age sixty-two than at age sixty-five, when they were eligible for full Social Security benefits and Medicare (U.S. House of Representatives 1998). Older Americans rely on many different arrangements to continue employer insurance during the transitional period that precedes Medicare eligibility. Some continue to work and receive employee health benefits

3 Short et al. Older Americans and Health Insurance Reform 43 Age Group Figure 1 Percent with Changes in Employment and Insurance according to Age in 1994 ( ) Source: Authors tabulations of Wave 1 (1992) and Wave 2 (1994) of the Health and Retirement Survey. until they are sixty-five, especially if they would not have access to employer insurance if they retired (Currie and Madrian 1999). Others who quit working have retiree health insurance from former employers or rely on current or former employers of their spouses for employer insurance. Early retirees are especially likely to exercise their right to continued coverage from former employers under COBRA,the Consolidated Budget Reconciliation Act of 1985 (Flynn 1994). This law requires employers with twenty or more employees to offer continuing coverage for eighteen months after both voluntary and involuntary terminations, with former employees paying up to 102 percent of the total premium. Thanks to the varied arrangements for continuing employer insurance after early retirement, and to higher retirement rates among people with access to retiree coverage, far fewer people leave employer insurance before age sixty-five than leave full-time work (Figure 1). In other words, although employer insurance does indeed decline as people approach age sixty-five, many older Americans successfully negotiate the transition out of full-time work with their employer insurance intact (Sloan and Conover 1998; Short, Shea, and Powell 2001a). Where there are losses of employer insurance, public policies that are already in place make up for part of the shortfall (Figure 2). A disproportionate share of older Americans qualify for Medicare and Medicaid disability benefits that are available to people who are unable to work. Disabled workers are also able to extend COBRA for an additional eleven months beyond the initial eighteen-month extension in order to cover the

4 44 Journal of Health Politics, Policy and Law Employer insurance Medicare/Medicaid Individual insurance Uninsured Age Group Figure 2 Percent Distribution of Insurance Types, by Age Source: Authors tabulations of 1999 Current Population Survey. Medicare waiting period, with employers charging as much as 150 percent of the usual premium. However, with reduced access to employer insurance, older Americans are more likely than other age groups to buy health insurance on an individual basis (Figure 2). Individual insurance costs considerably more for the same benefits. Individual insurance is typically priced at 150 percent over paid claims because of marketing and administrative costs in excess of claims, compared to as little as 105 percent in large employer groups (Pauly and Percy 2000). Individual insurance is also more expensive because of adverse selection; insurers charge higher premiums to account for the likelihood that buyers who are willing to pay for expensive individual policies are poor risks who will file lots of claims. With higher rates of public and individual insurance offsetting lower rates of employer insurance, the percentage of Americans who are uninsured after age fifty-five is only slightly higher than the percentage at younger ages (15 percent compared to 13 percent for fifty- to fifty-fouryear-olds, Figure 2). However, the percentage of fifty-five- to sixty-fouryear-olds who are uninsured is increasing faster than the uninsured percentage for younger adults (Employee Benefits Research Institute 1996; Fronstin 2000). With fewer employers offering retiree health insurance now and in the future (Hewitt Associates 1999), the current situation is almost certain to get worse (Glied and Stabile 1999). Either the number of uninsured older Americans will continue to increase or more people will struggle with the cost of individual insurance. While declining access to inexpensive group insurance offers some jus-

5 Short et al. Older Americans and Health Insurance Reform 45 Hypertension Figure 3 Prevalence of Chronic Conditions, by Age Source: Published estimates from the 1996 National Health Interview Survey (National Center for Health Statistics 1999). tification for age-related coverage reforms, we think that the most compelling arguments derive from the high average health expenses of people nearing age sixty-five. Both the prevalence of chronic conditions (Figure 3) and per capita health expenditures (Figure 4) increase markedly with age. Consequently, uninsured people who are approaching age sixty-five face much greater financial risks from uninsured medical expenses than do younger people who are uninsured. As a result of the relationship between age and health expenditures, older Americans also face much higher premiums for individual insurance. According to one recent study of market practices and regulations in ten states (Chollet and Kirk 1998), sixty-year-olds are likely to pay anywhere from two to four times the premium that twenty-five-year-olds pay for the same coverage in the individual market. By the same token, older Americans are more likely to be affected by underwriting practices that most states regulations permit in the individual market, where insurers limit coverage, raise premiums, or even refuse to sell to buyers with preexisting conditions. Finally, older Americans are also disadvantaged by the uncertainties surrounding their plans for retirement and old age. As people get closer to retirement, or after they have retired at a particular age, they have little flexibility to adjust to unexpected changes in their health, health insur-

6 46 Journal of Health Politics, Policy and Law Figure 4 Age Group Mean Total Health Services Expenditures in 1996, by Age Source: Unpublished tabulations of 1996 Medical Expenditure Panel Survey by the Agency for Healthcare Research and Quality. ance, jobs, or employment status. Both private and social insurance programs can ameliorate these risks. For example, Medicare and Social Security disability benefits for people under age sixty-five are designed to protect older Americans against the triple threat of disabling health problems that could (1) cause them to lose their jobs and earned income at an unusually young age, (2) incur large and continuing health care expenditures, and (3) lose their job-related health insurance. One of the risks for older people who rely on individual health insurance is the possibility of becoming a poorer risk, someone who will either have to pay much higher premiums or will not be able to purchase certain kinds of protection at all. There are also arguments against singling out the older age group for special reforms. Empirical studies (such as those reviewed in Currie and Madrian 1999) confirm that subsidizing health insurance for people in their early sixties will artificially inflate retirement rates. The increasing cost of health insurance is a predictable result of aging that people ought to factor into their financial plans. Older Americans have had time to accumulate assets that they can use to purchase health insurance or offset reductions in their incomes. There is also sympathy for singling out other population groups for incremental coverage reforms, such as the low-income parents of children who are already targeted by SCHIP and Medicaid or all of the poor, without regard to age or family status. One could argue, for example, that covering the uninsured at younger ages might prevent longer-term health

7 Short et al. Older Americans and Health Insurance Reform 47 <100% % 149% 199% 299% Family income as a percent of poverty Figure 5 Percentage of Fifty- to Sixty-Four-Year-Olds Who Are Uninsured, by Family Income Source: Authors tabulations of the 1999 Current Population Survey. problems, thereby improving family stability and economic prospects over the long run. However, considering the high incidence of chronic conditions at older ages, the potential payoffs from improving the health care of older Americans who lack health insurance (including public health programs like Medicare and Medicaid) are also great. If it is not politically feasible to expand coverage to all of the uninsured poor at once, we are inclined to argue for starting with the oldest of the poor. They confront greater health risks, have more health care needs, and face much higher insurance costs. As it is, there are great disparities in health insurance among older Americans that are related to income (Figure 5). Thirtyfive percent of fifty- to sixty-four-year-olds with family incomes below the poverty line are uninsured, compared to 8 percent above 300 percent of the poverty line. Criteria for Evaluating Policy Options Because people who are approaching age sixty-five are going through a transitional stage of life, we think that health insurance reforms for this age group should be designed with a lifetime view of health insurance arrangements, ability to pay, and age-related health risks. Special care is

8 48 Journal of Health Politics, Policy and Law needed to develop policies that will appropriately fit a context where people are making financial arrangements for their older years, deciding when to retire, and beginning to live on assets instead of current income. Furthermore, as different people move through this transitional stage of life in different ways, their individual situations and circumstances are quite heterogeneous. Good policy making must accommodate this heterogeneity and should not favor one path over another. Accordingly, we propose that new policy initiatives to expand health insurance for older Americans near age sixty-five should be judged by six criteria. Such initiatives should: 1. provide insurance efficiently, exploiting administrative economies that can be achieved through large insurance groups; 2. pool all insurable risks, including the risk of becoming a poor health risk; 3. minimally distort decisions about work and retirement; 4. encourage saving for retirement, given the evidence that many Americans will be financially ill prepared for old age (U.S. Dept. of Labor 1998; Employee Benefits Research Institute 2001); 5. require anyone who can afford insurance to pay for it, while making insurance affordable for those cannot pay; 6. preserve and offer health insurance choices. The first four criteria are concerned with the overarching goal that economists call efficiency, that is, putting public and private resources to their best possible use in order to maximize benefits in relation to costs. The fifth criterion addresses equity concerns. The last criterion acknowledges that Americans value having choices among health plans, especially when plan choices affect their choices of doctors and other providers (Gawande et al. 1998; Davis et al. 1995; Ullman et al. 1997). Efficiency emphasizes the maximization of net social benefits over social costs, without regard to the distribution of either costs or benefits across society s members. Both politics and social justice require attention to the distribution of costs and benefits as well. While there are many ways, both conceptually and operationally, to evaluate the equity (or fairness) of a policy s effect on different types of individuals, we rely primarily on progressivity as a test of equity in our analysis. If the distribution of a program s benefits is progressive, individuals receive benefits that decline as a percentage of income at higher incomes. If the financing of a program s costs is progressive, individuals incur costs that increase as a percentage of income.

9 Short et al. Older Americans and Health Insurance Reform 49 Policy Options Recent Proposals In his last three State of the Union addresses, former president Clinton proposed allowing older Americans to buy into Medicare at their own expense or to continue employer-sponsored insurance through an expansion of COBRA. Under the Clinton plan (Sheils and Chen 1999), anyone between the ages of sixty-two and sixty-four would be able to buy into Medicare for a premium of approximately $300 a month. Because Clinton intended the plan to be self-financing and $300 per month would not cover the full actuarial cost, early enrollees would pay an additional $10 premium per month after reaching Medicare eligibility at age sixty-five for every year that they participated in the buy-in. The Clinton plan would also help some people who lost their employer insurance. After exhausting any available COBRA benefits, displaced workers would be eligible to buy into Medicare after age fifty-five. In situations where employers stopped providing health benefits to retirees between the ages of fifty-five and sixty-four, they would be required to extend COBRA coverage for as long as ten years (until such retirees reached Medicare eligibility at age sixty-five). Over time, Clinton modified his plan to include a 25 percent tax credit to offset the cost of the Medicare buy-in and a 25 percent tax credit to offset the cost of COBRA continuation coverage (Clinton-Gore Administration 2000). The Medicare buy-in proposed by Pamela Loprest and Marilyn Moon (1999), as part of the Kaiser Family Foundation s Project on Incremental Health Reform, would also be available to everyone from age sixty-two through sixty-four, regardless of work status. This plan would subsidize a Medicare buy-in in two ways. First, the premium would be community rated at the actuarial cost of covering everyone in the age group from sixty-two through sixty-four, ignoring the effect of adverse selection on the expected average costs of people who actually participate. (Because people with higher expected medical costs have a greater incentive to participate in any voluntary insurance program, the costs of participants will be higher than the community average, while the costs of nonparticipants will be lower than the average.) Second, Loprest and Moon would provide government subsidies to participants below 200 percent of the poverty line, with a full subsidy (free coverage) for participants below 100 percent of the poverty line and partial subsidies for those between 100 percent and 200 percent of the poverty line.

10 50 Journal of Health Politics, Policy and Law A Menu of Policy Choices Drawing on the ideas in these two proposals and adding to them, we have organized the key policy decisions and options into four areas (Figure 6). Like diners ordering different courses from a restaurant menu, policy makers can choose from several alternatives within each of the four areas, and they are free to make choices in one area that are largely independent of their choices in others. Given that the choices listed in each section of the menu are not mutually exclusive, and that there are many different ways to combine choices across sections, there is practically no limit to the number of different policy proposals that could be constructed from this menu. The first set of options is concerned with providing a place where older Americans can buy health insurance more economically than in the individual health insurance market. Here the objective is to make health insurance more affordable by supplying it more cheaply and efficiently. Lowering premiums in this way will provide a measure of financial relief to those who are already paying for individual insurance and will encourage some of the uninsured to buy insurance. However, a Medicare buy-in is only one of several possible ways to provide older Americans with access to a new, less expensive source of health insurance. As suggested by the elements of the Clinton proposal that deal with workers who lose their retiree health benefits or have access to COBRA, employers could be encouraged or required to provide retiree or continuation benefits to more of their former employees. Older Americans could be allowed to buy into Medicaid, SCHIP, or some other state-sponsored plan for the low-income uninsured. Alternatively, federal and state governments already operate large health insurance programs for government personnel that could be extended to other citizens. In his campaign for the 2000 Democratic presidential nomination, Bill Bradley proposed opening the Federal Employee Health Benefits Program to all Americans without employer insurance (Fuchs 2001). In the state of Washington, the Washington State Health Care Authority contracts with health plans to cover not only state employees and retirees, but also low-income families who are eligible for insurance through the Washington Basic Health Plan (Washington State Health Care Authority 2000). Finally, private purchasing pools could also be developed as an alternative to individually purchased insurance (Curtis, Neuschler, and Forland 2001). There is a crosscutting policy issue related to the choice of premiumrating methods that would affect the price of insurance from any of these

11 Short et al. Older Americans and Health Insurance Reform 51 (1) Providing access to a new, low-cost low-cost source source of of insurance Status quo Medicare Medicaid of SCHIP (state) Current/former employers Government employee plans (FEHBP, state) Private purchasing pools (3) Defining eligibility Based on age Means-tested only No access to employer insurance Nonworkers only (2) Helping older Americans to pay pay premiums Status quo Means-tested subsidies Current income test Additional asset test Lifetime earnings test (applied at specified age) General tax credits Tax preferred savings accounts (4) Financing new expenditures on health on health insurance Government expenditures Payroll tax Income tax Enrollees Pay before enrolling Pay when enrolling Pay after enrolling Figure 6 Policy Options for Incremental Reforms for Older Americans sources. If premiums were actuarially fair, as in the Medicare buy-in proposed by President Clinton, then the premiums would be set above the community-rated average to reflect the higher propensity of those with higher-than-average claims to enroll. With higher premiums, even more of the good risks would opt out of the risk pool, and poorer risks would be left to bear most of the burden of their elevated costs themselves. However, if premiums were held at the community-rated level, as Loprest and Moon propose, and relatively more poor risks enrolled, then the plan would incur losses. Because our second evaluation criterion calls for pooling all insurable risks, and community rating provides insurance against

12 52 Journal of Health Politics, Policy and Law the risk of becoming a poor risk, we will assume community rating in evaluating our list of possible insurance sources. The second set of choices in the policy menu also is concerned with making health insurance more affordable, but it addresses the issue on the consumer side. Here the two most obvious options, given precedents and current policy discussions, are means-tested government subsidies 1 (an approach illustrated by the Loprest-Moon proposal) and general tax credits (illustrated by the latest modifications to Clinton s Medicare buy-in). A third alternative is to encourage younger people to set aside savings that can be used to pay for health insurance when they are older, for example, by subsidizing such savings with tax-preferred IRAs. If means testing is to be considered, then there is a further question about how to implement means testing in a fair and consistent fashion. Loprest and Moon propose premium subsidies for individuals with family incomes below a specified percentage of the poverty line. However, in an older age group where many people have already chosen to retire on reduced incomes, current income is not a very accurate measure of the ability to pay for health insurance. Two people with similar economic resources may have quite different incomes if one has retired and the other has not. An asset test, used in Medicaid and many other public programs, is the usual mechanism for identifying people with significant resources beyond current income. Another way to adjust for the effect of retirement on current income in this age group would be to consider lifetime earnings. A lifetime earnings test would not be hard to administer, because the Social Security Administration already compiles individual earnings histories for everyone who has ever paid Social Security taxes. The Social Security Administration could make a one-time determination of eligibility at age sixtytwo, for example, by comparing the prior lifetime earnings of an individual or a couple over a fixed period of time to the average value of the poverty line during the same time period. The third set of choices has to do with eligibility. Who should have access to new source(s) of insurance or receive assistance in paying premiums? Both the Clinton and Loprest-Moon proposals provide universal access to Medicare at age sixty-two, strictly on the basis of age. While the Clinton tax credits are also available to everyone over age sixty-two, eli- 1. We use the term subsidies in its most general sense to include vouchers, a schedule of premium reductions, tax credits, or free enrollment with no premium.

13 Short et al. Older Americans and Health Insurance Reform 53 gibility for the Loprest-Moon subsidies is limited by a means test. One could also limit access to a new, publicly sponsored source of insurance with a means test. That is the current design of Medicaid and SCHIP. Proposed coverage programs often exclude people with employer insurance or access to employer insurance, so as to help the uninsured or individually insured and not replace employer insurance. Finally, in order to specifically target the health insurance needs of early retirees (or displaced workers, as in the Clinton proposal), all or part of a new coverage initiative for older Americans could focus on those who leave the labor force. If a reform initiative for older Americans succeeds in insuring some of the uninsured, then the new coverage will have to be financed through some combination of government and enrollee expenditures. The last section of our policy menu is concerned with financing these new expenditures. On the government side, how should the revenue be raised to subsidize premiums or pay the administrative costs associated with expanding a government insurance program? On the consumer side, should younger workers be encouraged or required to prepay the cost of health insurance for their older years, as they currently do through the Medicare payroll tax? Should Americans under sixty-five postpone some premium payments until after age sixty-five, as in the Clinton plan? Or should the premiums simply be paid at the time of enrollment? Evaluating Options We evaluated each of these options against the relevant evaluation criteria in our list and then summarized our conclusions in four tables (Tables 1 4) corresponding to the four sections of the policy menu. After giving serious consideration to the alternatives, we conclude that a Medicare buy-in is indeed an appealing way to supply older Americans with a new source of health insurance (Table 1). Medicare s administrative costs are low, and the program is already organized to communicate with an older cross section of the U.S. population on an individual basis. Employer-based strategies cannot provide universal access to insurance, because relatively few of the uninsured or individually insured have a recent history of employer insurance. Most of the other alternatives would require a substantial initial investment in new administrative systems to reach beyond their traditional enrollee populations or, for private purchasing pools, to start up operations. A Medicare buy-in has the further advantage of smoothing the transition to Medicare at age sixty-five. In particular, it would avoid the possibility of disrupting patient-physician

14 Table 1 Evaluation of Options for Providing Older Americans with Access to a New, Low-Cost Source of Insurance Choices Provide Insurance Efficiently Pool All Insurable Risks Help Those Least Able to Pay Status quo Administrative costs and insurer profits Risk that health problems will lead to After age 62, enrollment rate for account for about a third of individual higher individual premiums, although individual insurance is higher premiums. a,b perhaps only as much as 15% of increased below the poverty line than above About 25% of year-olds lack group risk is reflected in higher premiums. b 300% of the poverty line. b coverage through employers, Medicare, or Coverage may be denied initially, but Medicaid. c two-thirds of policies guarantee renewal. b Adverse selection raises premiums and keeps the healthy from buying insurance. Medicare Administrative costs for 1998 were 3% of Equal and universal access to insurance of Medicare expenditures. d A Medicare can be guaranteed if government guaranbuy-in would piggyback on some existing tees community-rated premium against administrative systems, but enrollment adverse selection. and marketing would cost more for a buy-in than administering nearly universal coverage for people at age 65. Medicaid or Administrative costs for 1998 were 6% Equal and universal access to insurance SCHIP of Medicaid expenditures. d Medicaid s can be guaranteed if government guaranenrollment and marketing systems are tees community-rated premium against primarily designed to reach low-income adverse selection. mothers and children, the disabled, and the frail elderly.

15 Table 1 (continued) Choices Provide Insurance Efficiently Pool All Insurable Risks Help Those Least Able to Pay Current or Administrative costs and insurer profits Risk of losing access to insurance People who have a recent history former account for 5% 30% of group premiums, because of job losses or job changes of employer insurance have higher employers depending on group size. a Only 15 16% of average incomes than those with uninsured or individually insured no employer insurance to lose. b year-olds were covered by an employer in the preceding two years. c Government Piggybacks on current systems to contract Equal and universal access can employee plans with insurers. New administrative systems be guaranteed if government for enrollment, marketing, and premium guarantees community-rated premium collection would be needed for non- against adverse selection. government personnel. e Private Curtis, Neuschler, and Forland project Equal and universal access to purchasing administrative savings of at least private pools may not be sustainable, pools 10 15% compared to individual insur- unless government subsidizes ance, after initial start-up costs. f community-rated premium. apauly and Percy bpauly, Percy, and Herring cshort, Shea, and Powell 2001a. dhcfa efuchs fcurtis, Neuschler, and Forland 2001.

16 56 Journal of Health Politics, Policy and Law relationships for enrollees in managed care plans when they turn sixtyfive. Not unexpectedly, Table 2 shows that there is a trade-off between equity and efficiency in choosing among the strategies for providing premium assistance. All of the means-tested strategies that target assistance according to ability to pay would adversely affect work effort to some extent, as there would be an incentive to work less and earn less in order to qualify for premium assistance. General tax credits that were available to everyone, regardless of their economic circumstances, would not distort work and retirement decisions but would ignore differences in ability to pay. We note that, as these programs are usually structured, IRAs and other programs to promote savings also compromise equity goals because tax deductions or deferrals for the principal and interest in such savings accounts have greater value to taxpayers in higher income brackets who face higher tax rates. Taxpayers with more income are also more likely to participate in such savings programs (U.S. House of Representatives 1998). Of the alternative strategies for means testing, the lifetime earnings test has several advantages. It is likely to have much less effect on retirement decisions than is a current income test, because early retirement would have much more of an effect on current income than on lifetime earnings (especially if earnings after age sixty-two, when eligibility for Social Security begins, were ignored). Someone would have to be very farsighted to adjust his or her behavior at age thirty-five to increase the likelihood of passing a lifetime earnings test at age sixty-four. Furthermore, unlike an income or asset test, a lifetime earnings test would not discourage retirement savings. Nor would a lifetime earnings test encourage the kind of asset planning that undermines the legitimacy and effectiveness of asset tests in Medicaid. There is a similar trade-off between equity and efficiency in setting the criteria that determine eligibility for the program (Table 3). As we have just noted, means testing advances equity goals at the cost of distorting work, savings, and retirement decisions. In contrast to a redistributive policy that provides more program benefits to older people who earn less, a universal program that provides identical treatment for everyone in a specified age group will not distort the relative costs and benefits of early retirement. That is not to say that a universal insurance program for sixty-two- to sixty-four-year-olds, for example, would not encourage early retirement. Any new program that brings down the cost of health insurance in retire-

17 Table 2 Evaluation of Options for Helping Older Americans to Pay Premiums Minimally Distort Decisions about Work Choices and Retirement Encourage Saving for Retirement Help Those Least Able to Pay Status quo For insured workers without Because they will lose their retiree health Among year olds, 35% of the poor are retiree health benefits, the benefits if they leave a job before retiring, uninsured, compared to 8% of people above prospect of losing employer active workers are discouraged from 300% of the poverty line. a Tax exemption of insurance discourages participating in compensation plans that employer-paid premiums has greater value to retirement. trade current wages for health benefits higher-income taxpayers in higher tax brackets. after retirement. Means-tested subsidies Current Rewards loss of earnings Discourages saving by penalizing income Ignores availability of assets to finance income test associated with retirement. earned on assets. health insurance and consumption. Addition of Encourages postponement Rewards consumption during working Accounts for assets that are available to finance asset test of retirement by penalizing years ; penalizes saving for retirement. health insurance and consumption, but encourretirement saving. ages the well-off to hide or divest assets in order to qualify. Lifetime No distortion of retirement No distortion of savings decisions. Recognizes long-term, rather than short-term earnings test decisions after age when economic need. (at specified determination is made. age) Discourages earnings and work effort before determination is made. Fixed dollar No distortion of retirement No distortion of savings decisions. Ignores differences in ability to pay tax credits decisions. Tax-preferred Encourages earlier retire- Encourages saving for retirement over Tax exemption or deferral for IRA contributions health insurance ment by increasing current consumption. or interest has greater value to higher-income IRA resources in retirement. taxpayers in higher tax brackets. ashort, Shea, and Powell 2001a.

18 Table 3 Evaluation of Options for Defining Eligibility Provide Minimally Distort Insurance Pool All Decisions about Work Encourage Saving Help Those Least Choices Efficiently Insurable Risks and Retirement for Retirement Able to Pay Based on age Age-related eligi- Pools all random varia- Encourages earlier retire- Reduces need for Ignores differences in bility facilitates tion in health care ment. This is a distortion retirement savings ability to pay. link to Medicare, expenses within age only if the cost of retiree to cover insurance where group eco- group. health insurance is premiums. nomics can be lowered with subsidies realized. and is not just because of efficiency gains. Means-tested Could facilitate link Excludes healthier, Discourages work effort Depends on method Helps only those deemed only to Medicare or working population and earnings; extent of means testing. unable to pay. SCHIP, where group from risk pool. depends on method of economies can be means testing. realized. No access to Discourages Excludes healthier, Encourages earlier Discourages active Implicitly favors those employer employer insurance working population retirement for those workers from par- with lower lifetime wages insurance if enrollees are from risk pool. without employer ticipating in com- or less generous pensions, ineligible for pre- health benefits com- pensation plans that who are less likely to mium assistance. pared to those with trade current wages have employer insurance. benefits. for health benefits after retirement. Nonworkers Only about half of Excludes healthier, Encourages retirement; Reduces need for Implicitly favors those only year-olds are working population penalizes work. retirement savings. with lower current not working, reduc- from risk pool. incomes (i.e., no ing potential eco- earnings). nomies of scale. a ashort, Shea, and Powell 2001a.

19 Short et al. Older Americans and Health Insurance Reform 59 ment would encourage early retirement. Empirical studies suggest that the availability of retiree health benefits increases the early retirement rate by 30 to 80 percent (Currie and Madrian 1999). To the extent that reductions in retiree health insurance costs were accomplished through real gains in efficiency, it would be appropriate for older Americans to spend those gains by working less. However, because means-tested subsidies encourage people to retire so that they can benefit from the subsidies, thereby realizing individual savings that go well beyond any real savings in the cost of insurance, means testing distorts retirement decisions. A broadly defined program that encourages high levels of participation in all segments of the population would also further the goal of risk pooling. From this perspective, because good health is associated with work and income, it is particularly important to encourage enrollment among workers and at higher income levels. Finally, Table 4 lists the equity and efficiency arguments for financing new government expenditures through a progressive income tax rather than through the proportional payroll tax that currently funds part of traditional Medicare. Judged against our six evaluation criteria, the only advantage of a payroll tax is its favorable treatment of asset income and savings. Table 4 also summarizes the arguments for encouraging people to plan ahead for the predictably higher cost of buying insurance in their sixties. We assume that people prefer to even out consumption over their lifetimes so that their standard of living is not dramatically different at some ages than at others and is not radically affected by unexpected health or economic losses. If this assumption is correct, then the increasing cost of health insurance at older ages is a predictable aspect of aging that should be factored into financial and retirement planning. Another argument for encouraging earmarked saving for health insurance at an early age, before people know whether they will be good or bad health risks as early retirees, is that it helps to combat adverse selection. Allowing people to finance health insurance by borrowing against the future (as Clinton s post-sixty-five Medicare surcharge does) rewards shortsightedness. It also encourages adverse selection, because people who decide at the last moment that they need insurance do not have to be able to pay the entire premium.

20 Table 4 Evaluation of Options for Financing New Expenditures on Health Insurance Minimally Distort Minimally Distort Pool All Decisions about Work Decisions about Saving Help Those Unable Choices Insurable Risks and Retirement for Retirement to Pay Government Income tax Larger tax base than payroll Identical treatment of earned Progressive tax rates require tax allows smaller increase and asset income. high-income taxpayers to in tax rate. Higher tax rate contribute a larger percentage discourages earnings and of income. work effort. Increase Smaller tax base than income Favors saving by exempting High and low earners contribute Medicare tax requires larger increase in dividends and interest from same proportion of earnings. payroll tax tax rate. taxation. Favors high-income taxpayers, who have more unearned income. Enrollees Pay before Combats adverse Encourages saving to finance Requires individuals to predict enroll selection by encour- predictable increase in health future income and access to aging enrollment insurance costs at older age. alternative sources of insurance. before health status is known. Pay as Encourages adverse Allows individuals to make enroll selection by allowing enrollment decision with more participants to make information about retirement enrollment decision resources and access to alternative when health status is sources of insurance. known.

21 Table 4 (continued) Minimally Distort Minimally Distort Pool All Decisions about Work Decisions about Saving Help Those Unable Choices Insurable Risks and Retirement for Retirement to Pay Pay after Further encourages Permits adjustment to unanticienroll adverse selection by pated circumstances. relaxing liquidity constraint for participants who postpone enrollment decision until health status is known.

22 62 Journal of Health Politics, Policy and Law Policy Recommendations Combining Approaches The best policy initiatives will combine several of the options that we have listed and evaluated here. In particular, we strongly recommend that any incremental reforms aimed at older Americans should offer both a new place to buy insurance and assistance in paying premiums. Tax credits, subsidies, or premium savings accounts will not go very far at the prices charged in the individual health insurance market, and it is a waste of public money to subsidize health insurance at such a high price if it could be provided more cheaply. At the same time, because health insurance for this age group will be relatively expensive from any source, and because a significant proportion of the uninsured have limited resources, health insurance reforms are unlikely to cover many of the uninsured or to give meaningful relief to the individually insured without some kind of premium assistance. Furthermore, by combining approaches policy makers can avoid some of the trade-offs among competing policy goals that we have noted. For example, as Loprest and Moon propose, means-tested subsidies can be used to address equity concerns within the context of a broad, universal program that efficiently pools risks from all segments of the population. Our Preferred Approach Our preferred approach is to combine a Medicare buy-in with meanstested subsidies for people at the lower end of the economic spectrum and a tax-preferred savings program for people at the upper end of the economic spectrum. Elsewhere, we describe a detailed proposal that offers universal access to Medicare at a community-rated premium, awards premium vouchers to anyone with a history of low lifetime earnings, and allows tax-free interest on premium savings accounts (Short, Shea, and Powell 2001b). The voucher would cover the full amount of the Medicare community-rated premium for anyone below the poverty line, measured on a lifetime basis, and would phase out at 200 percent of the poverty line. We set the eligibility age at sixty-two. However, given the evidence that there are no dramatic changes in employer insurance at age sixty-two, we regard the eligibility age as an adjustable design element that can be used to resize the program to fit political and budgetary constraints.

23 Short et al. Older Americans and Health Insurance Reform 63 Providing Choices While we have not addressed the issue of choice in our discussion so far, we believe that it is possible to provide meaningful choice in this context. We would allow people who participate in our savings and voucher programs to buy into traditional Medicare, choose a Medicare+Choice plan, purchase employer insurance as active workers or retirees or through COBRA, or buy insurance through the traditional individual market. However, with a community-rated premium for Medicare, and participants in the savings and voucher programs having a choice between Medicare and private alternatives, adverse selection against Medicare could undermine the affordability and financing of the Medicare buy-in. Consequently, we suggest reducing the base amount of the voucher if it is used to purchase insurance outside of Medicare. Reducing the value of the voucher is effectively a risk adjustment to account for the likelihood that people who opt out of Medicare will be healthier than average. Such an adjustment is necessary because the people who are likely to find insurance at a lower cost than under Medicare are those whom insurers expect will submit fewer claims than the community average built into the Medicare premium. Otherwise, these insurers would have to charge at least as much as Medicare in order to cover their claims. If these insurers are covering a population that has fewer claims than the Medicare premium was intended to cover, then they should be paid less than the Medicare premium for each of their enrollees. We also suggest retaining current Medicare and Medicaid programs for the disabled under age sixty-five and excluding the people in these programs from the calculation of the community-rated premium for the Medicare buy-in. In this way, the Medicare and Medicaid disability programs will function as a high-risk pool that will help to reduce the buyin premium. Finally, administrative mechanisms can also be used to limit adverse selection while preserving choices. Consistent with the system of annual open enrollment that has been legislated for Medicare in the future, we suggest an annual open enrollment for the Medicare buy-in as well. If adverse selection turns out to be a particularly serious problem, participation in the buy-in could be limited to a one-time enrollment decision at age sixty-two (with some flexibility for those experiencing later changes in employment or family circumstances). The possibilities for using the vouchers outside of Medicare could also be curtailed or eliminated.

24 64 Journal of Health Politics, Policy and Law Table 5 Insurance Coverage under Current Law and for a Medicare Buy-in at Age 62 with Means-Tested Vouchers and Tax Exemption for Interest on Premium Accounts, 2001 Current law Current Family Income as a Percentage of Poverty < Total Enrollment (thousands) Medicare Employer ,208 3,452 Medicaid Individual Uninsured Enrollment rates Current Family Income as a Percentage of Poverty < Total Individual (Individual + uninsured) 22% 28% 36% 44% 47% 37% Insured total population 71% 76% 80% 87% 90% 85% Reform Vouchers based on current income Newly insured (thousands) Full vouchers Partial vouchers, administrative savings Savings program Program participants Vouchers, newly insured Vouchers, previously insured Savings program Enrollment rates Individual (Individual + uninsured) 88% 38% 44% 49% 52% 56% Insured total population 96% 79% 82% 88% 91% 89%

25 Short et al. Older Americans and Health Insurance Reform 65 Table 5 Reform (continued) Vouchers based on prior income Newly insured (thousands) Full vouchers Partial vouchers, administrative savings Savings program Program participants Vouchers, newly insured Vouchers, previously insured Savings program Enrollment rates Individual (Individual + uninsured) 50% 47% 55% 53% 53% 52% Insured total population 82% 82% 86% 89% 92% 88% Projected Effects of Recommended Approach The top panel of Table 5 shows our baseline estimates from the Current Population Survey (CPS) of the relationship between current family income and current insurance status in the population from ages sixty-two through sixty-four. The middle panel shows our estimates of the changes in insurance coverage that would result from providing premium subsidies or vouchers to this age group based on current income, as Loprest and Moon have proposed. We show estimates for subsidies based on current income because the available data permit much better estimates for that approach than for subsidies based on prior lifetime earnings, as we propose. In the bottom panel of Table 5, we approximate the effects of determining the subsidies on the basis of lifetime earnings. There are no data sources that describe the relationship between lifetime earnings and current insurance status that would allow us to model eligibility based on lifetime earnings. However, we can use data from the Health and Retirement Survey (HRS) to show the effects of basing subsidies on the annual income of the population over sixty-two in a prior year (before they turned sixty-two). Because a prior year s income reflects temporary reductions