This Article Abstract Submit a response to this article Services E-mail this article to a friend Alert me to new issues of the journal

M E D I C A R E
M E A N S - T E S T I N G
W E B  E X C L U S I V E   

8 December 2004  
  

Means-Testing In Medicare

Means-testing may be one of the few ways to preserve
Medicare’s generous funding and access,
without across-the-board rationing.

By Mark V. Pauly

ABSTRACT:

The Medicare Prescription Drug, Improvement, and Modernization Act (MMA) of 2003 introduces means-testing of premiums and benefits in two ways. It will means-test the Part B premium, setting higher premiums for better-off seniors. More importantly, it will offer much more generous drug benefits, at low or zero premiums, to lower-income beneficiaries. This paper argues that additional means-testing could improve Medicare’s financial picture. It proposes a strategy in which future Medicare beneficiaries with higher incomes will pay for cost-increasing but quality-improving new technology, possibly with prefunding that begins before retirement.

Medicare as we know it today cannot be sustained over the next fifty years and probably will run into financial difficulties within the next fifteen. Even before the addition of the Part D drug benefit, continuation of Medicare’s current coverage package and the trend toward adoption of new technologies would, under virtually any plausible set of assumptions about demographic change, workplace productivity growth, and changes in input prices and new technology, require very high income and payroll taxes.¹ In the judgment of many, those tax rates are politically implausible and economically undesirable. The additional cost of the new drug benefit, even under hopeful forecasts, will add to the financial challenge. Although some observers think that improvements in health or other unrecognized boons may save the system, many argue that something will have to change; here I take the need for change as a given.

In this paper I discuss a dramatic change in Medicare design and philosophy embodied in incipient form in the Medicare Prescription Drug, Improvement, and Modernization Act (MMA) of 2003 that could make a major contribution toward solving the long-term financing problem. That change is the introduction, in explicit if still small-scale fashion, of formal means-testing of premiums and benefits in Medicare. I consider the political and economic feasibility and desirability of preserving Medicare by what is, in effect, a plan to reduce its net benefits for the wealthiest Americans. I discuss not only the current provisions but also the rationale and possible forms of extending means-testing in the future. Any realistic plan for Medicare’s survival must have multiple elements, including tax increases and benefit modification; however, I argue that it is desirable to preserve and expand means-testing.

Means-Testing And The New Law

MMA will for the first time allow both Medicare premiums and insurance benefits to vary by beneficiaries’ income. That is, the premium will be based on ability to pay, and benefits will be based on need for help.

On the premium side, the law will change the way the Part B premium is calculated for households at different income levels. Above a relatively high income, the Part B premium will be set at a higher fraction of average Part B costs than the current approximately 25 percent benchmark. Beneficiaries with adjusted gross incomes of $80,000 (single) and $160,000 (joint) will be subject to higher premiums. The increase will be phased in over a five-year period, with the goal of increasing the percentage of average Part B expenses covered by premiums from the current 25 percent to 50 percent for these high-income beneficiaries. The short-run difference in revenue arising from this provision (compared with the continuation of uniform Part B premiums) is estimated to be less than $2 billion per year.²

The other means-testing provision in the bill makes premiums lower and benefits greater for the Part D drug coverage for low-income beneficiaries than for other beneficiaries. For those with incomes below 135 percent of the federal poverty level and with low wealth, the premium would be completely subsidized, and cost sharing would be limited to no more than $5 per prescription. The subsidy and the reduced cost sharing adjust on a sliding scale, phasing out at an upper limit of 150 percent of poverty. Note that this means testing occurs within an entirely federally funded program, not, as in the past, through supplementation of uniform federal Medicare by state-administered Medicaid programs for the poor. The revenue consequences of this change, compared with uniform benefits and premiums, are substantial, probably amounting to about a quarter of the forecast cost of the Part D benefit.

Where To Go From Here?

It is by no means obvious that the means-testing of benefits and premiums built into MMA necessarily heralds a trend. For example, there is already much discussion of and support for enhancing Part D benefits for the nonpoor so that they more closely resemble those for lower-income households, primarily by closing the “doughnut hole” in coverage for expense levels from $2,250 to $5,100.³

The present small window of surpluses (predicted to be closed this fiscal year) in the Medicare Part A budget offers hope for those who think that benefits could yet be enhanced. There is likely to be a race between support for closing the drug coverage gap, on the one hand, and the worsening financing situation of Medicare’s earmarked revenue sources and the overall federal budget, on the other. It is likely, however, that means-tested premiums for Part D will remain, and it is plausible that the budgetary stringency will win the race.

Mitigating the future problems of Medicare by having the wealthiest Americans pay more and get less from that program appears to have some appeal. The idea that social insurance benefits should be limited for the wealthy was discussed favorably by Sen. John Kerry (D-MA) during his 2004 presidential campaign, although in the context of the less financially troubled Social Security program.⁴ Of course, if it were possible to continue Medicare in its current uniform-coverage format, most people would probably want to do so. This change is in prospect not because people generally want to move in this direction but because of the near-necessity of doing something.

Here are some illustrative numbers to indicate the nature of the future funding problem; they are based on estimates in the Medicare trustees’ report but differ slightly from the projections presented there in dealing less formally with the range of uncertain values. (For the moment, I ignore the taxes needed to fund the new drug benefit.)

Medicare is currently in approximate short-term fiscal balance—that is, revenues being received are approximately equal to outlays. Two important demographic parameters govern Medicare’s future cost to taxpayers under age sixty-five. One parameter has to do with the rate of growth of the beneficiary population; the trustees’ report projects that rate to be in the range of 2–3 percent per year after 2005; the simple average value between now and 2025 is 2.6 percent.⁵ The other key parameter is the rate of growth of the tax-paying, working population. Current estimates put that value at about 0.8 percent per year; it could vary depending on immigration policy, changes in labor-force participation, and (with some lag) the birth rate.

Tax rates on payrolls for Part A and the implicit tax rates on income that fund Part B could stay approximately constant if the rate of growth of Medicare revenues equals the rate of growth of spending. The potential “balancing factor” here—the third key parameter—is the rate of growth in tax base (wages or income) per worker. For example, if the composite of wages and income per worker grew at 1.8 percent per year, the resulting 2.6 percent growth in revenues (roughly 0.8 plus 1.8), equal to the growth of the Medicare population, would be just enough (ignoring demographic changes within the Medicare population) to keep real spending per beneficiary at current levels with no change in tax rates. Neither benefit cuts nor tax increases would be needed.

Historically, growth rates in real per capita wages and gross domestic product (GDP) (a proxy for income) have fluctuated over a wide range, even becoming negative in recessions; because capital income has been growing relative to wage income, wages grow slightly more rapidly than GDP.⁶ During the period 1990– 2000 the growth rate in real GDP per capita was 2 percent per year, but it was 1.1 percent in the first half of the decade and 1.9 percent during 1980–2002.⁷ The Medicare trustees now forecast slightly slower future growth than these long-term trends suggest.

A key message, then, is that real Medicare benefits per beneficiary could be held constant, at approximately current tax rates, if overall economic growth is near the historical level; if it falls slightly below that level, tax rates would need to rise a little. However, it would be much more discouraging if real Medicare benefits continue to grow at their long-term real rates of 4.7 percent per year, or even at the more moderate rate of GDP growth plus 1 percent used in the trustees’ projections. Then revenues at constant tax rates would fall well behind spending, the Part A trust fund would be drawn down, and taxes would need to be raised or benefits would need to be limited. All of these projections exclude the MMA drug benefit; its estimates are very imprecise, but it will probably add at least 20 percent to Medicare taxes upon initiation and grow thereafter.

Projections about the mid-century increased Medicare tax burden with real benefit growth are in the range of doubling to more than tripling the share of GDP going to Medicare and the associated tax rates on taxable incomes. Adding this Medicare growth to taxes and to fund Social Security indicates that the percentage of wages to pay for programs for the elderly (Medicare plus Social Security) could more than double from today’s level of about 15 percent. I assume in this paper that it is a policy goal to hold these tax rates at approximately current levels (after adding the new general-revenue contribution for the drug benefit).

Three Common Objections To Means-Testing

Before I outline a more detailed rationale and plans for means-testing Medicare, I address three objections often raised to a means-testing strategy. (Other objections related to higher administrative costs are now less compelling, since the administrative costs of determining income and wealth will already have been incurred for the Part D program.) These are (1) it won’t help; (2) it won’t make a difference; and (3) it isn’t politically appropriate.

Means-testing won’t help. As already noted, the future fiscal challenge to Medicare is daunting. To make the program sustainable into the indefinite future, more than $60 trillion (in present-value terms) would be needed right now.⁸ (Things will get even worse if we wait.) Compared with these higher future liabilities, could we expect to raise enough from the nonpoor elderly population to cut these liabilities down?

A review of data on the elderly suggests that in the short run, there is a basis for hoping for some improvement from means-testing, but a need for tempering that hope. The percentage of households with incomes below poverty is smaller for people older than age sixty-five than for the rest of the population. However, the percentage of households with incomes above 300 or 400 percent of poverty, possible benchmarks for ability to pay more in premiums or get less in benefits, is also smaller than in the rest of the population. Exhibit 1 shows the distribution of household incomes relative to poverty for the population over age 65, with the distribution for people ages 55–64 included as a comparator.

Are there enough elderly households that could pay higher premiums or deal with out-of-pocket payments to help Medicare’s future? In a static sense, the answer appears to be “some” but perhaps not enough. Here is an illustrative calculation: Beneficiaries now pay about 10–15 percent of the per beneficiary cost of Medicare (as Part B premiums). If the approximately 25 percent of nonpoor households paid instead for half of their total Medicare, the overall financial burden would be reduced by about 10 percent (since the share would increase by 40 percent for 25 percent of beneficiaries). This would be much-needed fiscal help, but not salvation.

Means-testing won’t make a difference. Would means-testing Medicare make a difference in the distribution of well-being among Americans? Having a higher-income household pay a larger share of its insurance cost after age sixty-five may make sense, but why not just collect this money as higher general taxes before age sixty-five, instead of waiting to collect it later? To be sure, as we have noted, the implied tax rates needed to do this job are very high, but moving the tax to later in life does not necessarily make it smaller. So what’s the point?

It is possible that distortive effects of income-related taxes or charges postponed to retirement age could be smaller. Clearly, there will not be discouragement of work effort for someone who is not working. But won’t there be a negative effect on work effort when younger? Not necessarily; if the worker does not take into account the effect of wealth on future insurance premiums, the effect may then be mitigated.⁹

Even if workers do have perfect foresight and anticipation, means-testing need not affect the labor supply, since there is no necessary connection between current wages and either future income or future Medicare benefits. There might be a negative effect on saving, but even this is ambiguous. Some people might save less in order to have lower incomes and qualify for better benefits, but those with higher incomes who do not expect to be subsidized will need to work harder and save more to replace lost Medicare benefits.¹⁰

Means-testing isn’t politically appropriate. The reason why Medicare and Social Security pay generous benefits to higher-income retirees is intrinsically tied to the political/philosophical theory surrounding social insurance. The theory is that if all citizens believe that they share benefits from these programs, there will be more political support for them. The actual transfer aspects of those programs would not, it is claimed, be supported by the middle class and above if considered as “welfare” but would be supported if thought of as social insurance.

If the net benefit to the nonpoor from social insurance could be positive, as it was for much of the history of Social Security and Medicare, this argument has merit. But with the current demographics undoing the “everyone-is-a-winner” character of social insurance, the argument is less persuasive, either as a positive political theory or as a normative welfare theory. There will in the future be transfers that higher-income taxpayers can never recover, and the fiscal illusion in which nonpoor people focus primarily on Medicare benefits may not be so prevalent once they note that their taxes paid fall well short of the value of their benefits. Normatively, it is unclear how to justify inducing the upper middle class into being more generous than it really wishes to be.

There is even reason to doubt that Medicare actually does redistribute income to any great extent. As Mark McClellan and Jonathan Skinner have shown, higher lifetime income is associated with longer life and higher-price medical services; these effects mean that although people with higher incomes (or wages) pay more Medicare taxes, they also get more lifetime benefits than people with lower wages.¹¹ If this is true, adding a requirement that higher-income people also pay more for Medicare after retirement might achieve, at long last, the modest amount of income redistribution that the program’s authors thought they were enacting.

Means-Testing And The Future Of Medicare

Because of the near-certainty of high future Medicare costs, the need to raise revenues or reduce spending by means-testing may gain grudging acceptance. The simplest ideas would be to require higher Part B premiums for the better-off, or to increase their Part A or Part B cost sharing. But any optimism about this help would be tempered by its apparently limited impact, as already noted. However, consideration of the detailed reasons for the probable fiscal train wreck may suggest an alternative rationale and an alternative, more effective strategy.

There are two reasons why Medicare is expected to become such a fiscal drag on the working population. One is the inevitable demographic shift, which will lead to fewer workers per beneficiary. But worker productivity and therefore the wage or income tax base are also expected to increase. As noted earlier, if productivity increases just a little more than its historical rate, the tax rate needed to support today’s level of real spending for even much larger numbers of future beneficiaries need not change appreciably.¹²

What plunges the fiscal future of Medicare into deepest gloom is the historical fact, mentioned earlier, that spending per Medicare beneficiary has always grown, in real terms, much more rapidly than economywide inflation or GDP, and it is expected to continue to do so. The most important root of this real spending growth is not “cost inflation” prompted by exploding inefficiency, higher wages to health care workers, or provider profits, but rather the addition of more beneficial but more costly new technology. If there were a way to avoid having the existing financing system bear the burden of this new technology for nonpoor Medicare beneficiaries, we might be able to get into the clear. I next explore this idea.¹³

Means-testing technology. The primary reason why control of future Medicare spending is needed is not related to demographics as much as it is to the expectation that the historical rate of increase in the discovery and diffusion of costly but beneficial technology—and its effect on costs per beneficiary—will continue. It is possible, and perhaps plausible, that taxpayers will wish to continue to make this new technology available to lower-income seniors. But for those elders with resources, it would seem sensible to have them bear much of the cost of this “non-poolable” item. The new Part D benefit is likely to set a precedent. The stage has been set for differentially constraining what the nonpoor will get.

Medicare could, with approximately its current tax rates, continue to guarantee to fund the real dollar amount of today’s benefit package for all seniors, present and future. If the “social contract” sometimes said to be embodied in Medicare represents a promise to cover all new technology that is adopted for use by the population under age sixty-five, that “contract” would be violated. But a more limited promise of constant real benefits would still represent much of what people have come to expect. Such a benefit could grow at the economywide rate of inflation and might, if there were a political choice of modestly higher taxes, even be set to grow at the rate of some other indicator that grows less rapidly (in inflation-adjusted terms) than Medicare has traditionally done. This kind of Medicare spending program could be financially feasible.

But, then, what of cost-increasing but health-improving new technology? This is the largest share of real spending growth historically, and it is likely to continue to be so as coverage of drugs is added. It cannot be supported in the future for all seniors at current tax rates. So nonpoor seniors could be obliged to pay for their own costly improvements in technology: out of pocket, through supplementary coverage (for traditional Medicare), and through supplementary premiums to private plans that cover it. In my view, this is the key to saving Medicare.

It might even be both proper and desirable to go a little bit further into the red to make two guarantees. First, as noted, we should provide new technology to the poor. Second, we can and probably should guarantee to existing retirees most of the technological progress that will be experienced by the population under age sixty-five; we should try to hold onto the promise that Medicare would be “about as good” as private insurance for them. The means-testing of payment for future technology should begin with future retirees. But for nonpoor people not yet on Medicare, we should not promise financing by others but instead should see to it that they fund their own new technology.

So what are the options? The illustrative calculations discussed earlier can provide evidence on the effectiveness of the need for means-tested cost containment strategies. Exhibit 2 displays benchmark estimates of 2035 Medicare spending (excluding Part D) as a proportion of GDP under alternative scenarios. (These figures are intended to illustrate the effects of alternative programs and are not intended to be precise forecasts of future growth.) The first three lines show that with no means testing, the proportional spending burden will more than double today’s 2.6 percent of GDP under the trustees’ assumption about per beneficiary spending growth, or will more than triple at the historical growth rate.¹⁴ Because Medicare’s “payroll” tax base is about half of GDP, these figures translate into a two- or threefold increase in tax rates from the current effective payroll tax rate of 4.3 percent.¹⁵ Medicare tax rates alone (adding in Part D) could exceed 12 percent.

The next two lines show the effect of limiting real spending growth for means-tested (higher-income) populations, equivalent to 40–60 percent of retirees. As can be seen, these changes greatly reduce the future spending burden: It falls to 4.6 percent and 3.9 percent of GDP.

The last line shows the result of a more aggressive strategy to limit spending, which limits real benefits growth to GDP minus 1 percent for the lowest-income 40 percent of the Medicare population and holds real benefits constant for the remainder; the tax share then would rise only modestly to 3.3 percent of GDP.

One must admit that none of the scenarios in Exhibit 2 is especially attractive. But we probably have little choice but to choose among them. It may be that despite an aversion to means-testing of Medicare, the political system will feel compelled to choose one of the means-tested approaches instead of tolerating a doubling or tripling of the tax cost or imposing strong limits on real benefits for all.

Exhibit 1 suggests another possible source of funding: the higher incomes of households just before they retire. At the most obvious level, funding new technology is probably something that today’s middle-class fifty-five-year-olds, the leading edge of the baby boom, could plan for; two-thirds of them have incomes above 300 percent of poverty.

The apparent dramatic drop in money income after retirement is also probably something of an illusion. The upper middle class, to some extent, live off their wealth after they retire, instead of cutting consumption close to the poverty level. The primary type of wealth for retirees is housing wealth; more than half of elderly households have housing wealth in excess of $100,000; they own their own homes, often with a paid-up mortgage. In contrast, nonhousing financial wealth is low among the elderly and is highly correlated with income, with only about 10 percent having nonhousing wealth in six figures.¹⁶ The primary consumption-increasing aspect of the situation for many seniors is being able to live nearly tax-free in a mortgage-free house, so housing costs are low. It is this extra consumption “cushion” that could help pay for access to new technology.

Some type of voluntary prefunding of “high-tech” Medicare plans for those with high household incomes as they near Medicare might well be desirable and feasible. Placing the new obligation at this point in the future would have two benefits: It would encourage saving as voluntary prefunding, and even some higher-income beneficiaries might decide, when faced with the explicit cost, to forgo the latest technology, at least for a while and at the margin. If so, this would presumably represent a more rational decision process for new technology than Medicare’s traditional coverage decision process of muddling through while trying to deflect lobbying and political pressure. Or people might decide that compared with what else they could do with their retirement income, new technology is the best choice; that would be fine as well, and they would be paying for what they are getting.

Design issues. If Medicare is to be means-tested, how might that work? A problem with varying only premiums with income arises if the coverage is voluntary. The problem, as Ted Marmor and Jerry Mashaw point out, is that a well-off beneficiary’s means-tested premium might actually exceed market-based premiums for the same coverage.¹⁷ This will occur if the redistributive tax exceeds the loading on private insurance (that is, the amount an insurer adds to cover its cost), and it will occur to an even greater extent if there is serious adverse selection. So it might be preferable to concentrate the means-testing on benefits instead of premiums.

How might income conditioning of benefits for new technology work? The simplest way to implement it for the general program would be through Medicare Advantage (MA) plans (formerly known as Medicare+Choice). Medicare’s contribution to these plans’ coverage could stay constant in real terms for higher-income people. Nonpoor beneficiaries then either could accept a low-cost plan slow to adopt the latest technology or could pay more for the privilege of first access. Private Medicare plans could specify something about the technology, or about the budget available for it, as part of their marketing. Specifying what rationing method will be used for new technology, even describing a specific cutoff of dollars per quality-adjusted life year (QALY) gained, could make sense.

It is much more difficult to see how means-testing could work in traditional Medicare because of the political difficulty in a publicly administered program of identifying which technologies to exclude from coverage. Some of the elements needed to adjust the benefit levels are already available in Medicare. The program has constructed input price indexes, which could be used to deflate growth in spending so as to isolate the new technology. An adjustment for improved quality is sometimes developed for Part A (although the choice is often political and budgetary). Also, there have been some tentative and uncoordinated steps to use methods of cost-effectiveness analysis for coverage policy.

There is no simple way to describe how traditional Medicare might shift the cost of new technology to nonpoor seniors. However, here are some thoughts on how this necessary task might be accomplished. One approach would be a “defined contribution” approach: Medicare would define in dollars what it was prepared to spend on benefits for nonpoor beneficiaries and then offer a menu of alternative sets of uncovered technologies (or technologies covered for a supplemental premium) that are consistent with that budget. Beneficiaries would choose from that menu. The other approach would be a “defined benefit” approach. Under this approach, cost-effectiveness analysis of some form would be used to limit coverage; such limits (as now) might take the form of excluding some types of care entirely from coverage or specifying coverage of some care conditional on certain clinical data. The cutoff level for cost-effectiveness would be determined, Oregon-style, by whatever could be bought with the constant real expenditure provided by the basic plan. It might also be desirable to temper the phasing out of complete coverage by increasing cost sharing for “old” technology and using the savings for partial coverage of the more cost-effective new technology. A discussion that goes further than these thoughts into the so far unsolved problem of how to make cost-effectiveness analysis in Medicare fully administratively and politically acceptable would be lengthy and inconclusive. But the main motivation, here as above, is that the alternative of spending more is not feasible.

Finally, limitations on new benefits might be linked to both current income and a person’s income at preretirement ages, with adjustments for unexpected adverse events. That is, if I am especially well off at age sixty, I should expect less generous Medicare coverage at age seventy no matter what my income at that point; I could get an exception if I could show casualty losses or other bad events. Some type of prefunding vehicle would be suitable for this purpose.

Note also that this scheme deals flexibly with whatever biomedical scientists dish out. Should the net rate of cost growth attributable to beneficial new technology dramatically slow (for those who think, for example, that drug coverage will reduce total spending, or because of spillover effects from a private-sector cutback as premiums bite into consumption), the extra premium or cost imposed on the well-off for new technology could and would be then adjusted downward and might even be unnecessary. In contrast, if there were a bumper crop of breakthrough products—cures of Alzheimer’s and therapeutic vaccines for cancer, for example—the better-off would have to pay more but should count themselves on balance as very lucky.

For the new drug benefit, means-testing new technology may be even easier. That program sets up a choice among private benefit management plans, which could charge different premiums for different formularies or other ways of limiting costly new products. Although the mechanism for explicit limitation of spending growth for the nonpoor is not clear in the law, the history of Medicare strongly suggests that Congress would put limits on spending growth when it wished to do so. Alternatively, there are provisions in the law for examining the comparative benefits of new technology, which could serve as a vehicle for planning if there is sufficient political will.

Concluding Comments

Medicare has traditionally provided more than funding for and access to a static level of basic medical care at current state-of-the-art quality. It has paid for additional beneficial and costly new technology at a rate nearly equal to that in the private sector. It is the prospect of the continuation of this trend that makes Medicare as we have known it very uncertain in the relatively near future. The economic distortions that would be required if traditional Medicare were to continue to pay for that technology for all are so large as to virtually demand that something be changed.

Means-tested Medicare payment for future (not present) technology for future (not present) retirees is perhaps a sufficiently attractive solution (relative to other more drastic and more immediate adjustments) that the small steps down that path embodied in MMA will anticipate a bigger march in the same direction. My rough calculations about the fiscal impacts of such a policy only hint at what might be politically feasible, depending on what are fundamentally policy and value judgments about how much technology is enough for the nonpoor and what income and wealth levels might define the obligation to pay.

Regardless of the precise parameters, such changes should increase the chances for continued funding for every benefit in today’s Medicare for all without greatly increasing tax rates. Access to new technology for those with lifetime low incomes should also be continued. But these actions may require that the majority of the population that is not poor accept the responsibility of explicitly paying for its access to new technology, through partial income conditioning. Far from being a threat to Medicare as we know it, means-testing may be one of the few ways to preserve the program’s generous funding and access, without across-the-board “brute rationing,” that we all want to be able to anticipate in retirement.¹⁸

NOTES

1. Boards of Trustees, Federal Hospital Insurance and Federal Supplementary Medical Insurance Trust Funds, 2004 Annual Report of the Boards of Trustees of the Federal Hospital Insurance and Federal Supplementary Medical Insurance Trust Funds, 23 March 2004, cms.hhs.gov/publications/trusteesreport/2004/tr.pdf (1 November 2004).
2. J. O’Sullivan et al., “Overview of the Medicare Prescription Drug and Reform Conference Agreement, H.R. 1,” Congressional Research Service Report for Congress, 4 December 2003, www.ahcahp.org/bppo/ Best%20Practices/crs%20summary%20of%20act.pdf (1 November 2004).
3. P. Barry, “The New Medicare—and You,” AARP Bulletin Online, January 2004, www.aarp.org/bulletin/medicare/Articles/a2003-12-24-newmedicare.html (1 November 2004).
4. Centrist Policy Network, “Kerry Talks Social Security,” 19 April 2004, www.centristpolicynetwork.org/archives/000071.html (1 November 2004).
5. Computed from Boards of Trustees, 2004 Annual Report, Table II-A-3.
6. The growth in total GDP is about one percentage point higher than the per capita rate because of population growth.
7. U.S. Department of Labor, Bureau of Labor Statistics, Office of Productivity and Technology, “Comparative Real Domestic Products per Capita and per Employed Person, Fourteen Countries, 1960–2002,” Table 6, 29 July 2003, www.bls.gov/fls/flsgdp.pdf (1 November 2004).
8. Boards of Trustees,2004 Annual Report.
9. The effect of changes in the taxation of Social Security benefits on changes in the estate tax bears on this question. The presence of uncertainty about one’s own future and future government rules, imperfect knowledge, and imperfect capital markets all suggest that a change in the way benefits are related to income is likely to have a smaller effect on work effort than would an equivalent change in working-life tax rates. See P. Diamond, Taxation, Incomplete Markets, and Social Security (Cambridge, Mass.: MIT Press, 2003).
10. I am indebted to Kent Smetters for this point.
11. M. McClellan and J. Skinner, “The Incidence of Medicare,” NBER Working Paper no. 6013 (Cambridge, Mass.: National Bureau of Economic Research, April 1997).
12. The point that the adverse prospects for Medicare’s future do not arise primarily from the aging of the population has been made forcefully in U. Reinhardt, “Does the Aging of the Population Really Drive the Demand for Medicare?” Health Affairs 22, no. 6 (2003): 27–39.
13. For a more extensive discussion, see M. Pauly, “What If Technology Never Stops Improving? Medicare’s Future under Continuous Cost Increases,” Washington and Lee Law Review 60, no. 4 (2003): 1233–1250.
14. If Part D costs are added to those estimated here, the Medicare 2035 GDP percentage under the growth assumption of GDP plus 1 percent would be 7.8 percent rather than the 6.0 percent shown here. See Boards of Trustees, 2004 Annual Report, Table II-A-2. This estimate uses a slightly higher assumed rate of real GDP growth than the 1.1 percent embodied in the most recent long-run projections.
15. This calculation adds to the 2.9 percent Part A tax an additional 40 percent to reflect funding for Part B as an equivalent payroll tax.
16. J.P. Smith, “Wealth Inequality among Older Americans,” Journals of Gerontology, Series B: Psychological Sciences and Social Sciences 52, Special Issue (1997): 74–81.
17. T.R. Marmor and J.L. Mashaw, “The Case for Universal Social Insurance,” in Policies for an Aging Society, ed. S.H. Altman et al. (Baltimore: Johns Hopkins University Press, 2002), 169–198.
18. H. Jenkins, “Too Bad He Couldn’t Find a Veep Named Gridlock,” Wall Street Journal, 7 July 2004.

Mark Pauly (pauly{at}wharton.upenn.edu) is the Bendheim Professor; a professor of health care systems, business and public policy, insurance and risk management, and economics; and chair of the Health Care Systems Department at the Wharton School, University of Pennsylvania, in Philadelphia.

DOI: 10.1377/hlthaff.w4.546
©2004 Project HOPE–The People-to-People Health Foundation, Inc.