FROM ISSUE NUMBER 9 ~ FALL 2011 GO TO TABLE OF CONTENTS

TEXT SIZE A A A

Means Testing and Its Limits

ANDREW G. BIGGS

America faces a massive crisis of entitlement spending. The Congressional Budget Office projects that, just a decade from now, government debt will be greater than the nation's entire gross domestic product — and that the main driver of that debt will be entitlement programs. Because of America's aging population and rising health-care costs, Social Security and Medicare have simply been pushed to the brink.

Meanwhile, today's trillion-dollar deficits, depressed federal revenues, and ballooning discretionary spending restrict the range of possible solutions. There is no more cushion of either funding or time. Slow and steady increases in the retirement age and minor tweaks in benefit formulas can no longer stave off disaster: Reformers must now entertain policy solutions once considered unimaginable. And prominent among those solutions is subjecting Social Security and Medicare to some form of means-testing, by which poorer seniors would receive more generous benefits and the wealthy would receive less (or none at all).

The appeal of means-testing is easy to see. It would dramatically reduce entitlement spending, and in a way that could be justified as commonsense: Since we don't have the money to give benefits to everyone, we should give benefits only to people who need them most. Why should a billionaire like Michael Bloomberg have his health care heavily subsidized by overburdened taxpayers? Means-testing entitlements would save huge amounts of money by not paying out benefits to wealthy Americans who can get along just fine without them. At the same time, it would still allow Medicare and Social Security to serve their basic purpose: providing a safety net to the elderly.

But means-testing also has some serious drawbacks, especially when it comes to how the policy might shape Americans' financial decisions. Means-testing entitlement benefits could punish the very people who work the hardest and save the most, depressing economic activity and discouraging good behavior. The cure for our fiscal problems could thus end up being worse than the disease.

If reformers are to reduce the costs of Medicare and Social Security without inviting counterproductive side effects, they must pay careful attention to the design of their policy reforms, and remain open to solutions that go beyond simple means tests. Embracing the idea that the rich should receive less generous entitlement benefits than the poor could help avert America's entitlement crisis — but the devil, as always, will be in the details.

THE UNIVERSAL ETHOS

Originally, both Social Security and Medicare were designed as "social insurance" programs meant to ensure that, regardless of what might happen with their personal finances and retirement savings over the course of a lifetime, elderly Americans would be protected by a government-backed safety net that would prevent destitution in old age. Since the net extended to nearly every senior, these programs can be best understood as universal benefits.

Social Security provides a monthly retirement-benefit check to some 46 million Americans, as well as disability benefits to another 10 million. The program, which now costs roughly $730 billion per year, is the largest line item on the federal budget. It is funded by a nearly universal flat payroll tax of 12.4%, half of which is paid by the employee and half by the employer.

Each person's retirement benefit is determined by a formula that takes into account the age at which he chose to retire (those who retire early, starting at 62, receive lower benefits than those who wait until the normal retirement age, which is now 66) and a measure of his earnings during his working years — and therefore of the amount he paid in Social Security taxes. That measure is based on an average of the highest 35 years of his covered earnings (that is, of pay that was subject to the payroll tax), adjusted for wage growth and then divided by 12 to yield a monthly amount. That amount is then inserted into a progressive benefit formula designed to ensure that Social Security benefits replace a larger proportion of pre-retirement earnings for people with low average earnings than for those with higher earnings.

For instance, a worker who becomes eligible to retire this year will receive 90% of that monthly amount up to $749, plus 32% of the remaining amount between $749 and $4,517, plus 15% the remaining amount above that level. This means that workers with lower lifetime earnings receive a proportionately higher (though generally still nominally lower) benefit. The benefit formula provides an added protection to the poorest seniors, but the program is still designed to offer a fundamentally universal benefit: Rich and poor alike receive Social Security checks in retirement, and the rich tend to receive larger ones. As the Social Security Administration says on its web site, Social Security "is not and was never intended to be a program to provide benefits based on need. Rather, it is a system of social insurance under which workers (and their employers) contribute a part of their earnings in order to provide protection for themselves and their families if certain events occur."

Medicare — which provides comprehensive health insurance to essentially every American over the age of 65 — carries this universal ethos even further, offering the same health benefits to all retirees regardless of their circumstances. The program now costs roughly $550 billion per year, a figure that is growing much more quickly than the cost of Social Security, thanks to galloping health-care inflation. Medicare, too, is funded in large part by a flat payroll tax (at a rate of 2.9%, shared equally between employers and employees), though some elements of the program are also funded by general revenue and by premiums paid by enrollees. Eligibility for benefits is based upon Social Security eligibility, but unlike Social Security, Medicare benefits do not vary based upon prior earnings or contributions to the program. Therefore, those benefits, too, are not based on need, nor are they based on what retirees paid into the program while they were working.

Indeed, although Social Security and Medicare are both considered earned social-insurance benefits, neither involves a direct relationship between money paid into the program and money drawn out of it. The structure of both programs means that current workers fund the benefits of current retirees; workers do not put money into a fund that will support their own benefits later. In Medicare, the average retiree actually draws far more money in benefits than he put into the program while working: According to Eugene Steuerle and Stephanie Rennane of the Urban Institute, a two-earner couple with average wages turning 65 this year will have contributed $119,000 to Medicare in taxes but will receive $357,000 in benefits during retirement. In Social Security, a similar average couple retiring today at age 67 will actually receive somewhat less in benefits ($560,000) than they put in to the program while working ($611,000).

In both cases, however, the level of benefits bears no relation to economic need. It would thus be a significant departure from the programs' histories and purposes to subject them to means-testing. Under a means test, the government assesses the resources available to an individual — his income and assets, aside from the benefit payment in question. To the degree that those other resources exceed some pre-set level, the government then reduces (or eliminates) the individual's benefit payment.

Means-tested benefits are relatively common in the United States in welfare programs aimed at reducing poverty. Familiar means-tested benefits include the Earned Income Tax Credit, Medicaid, and Supplemental Security Income, while a large number of tax credits or deductions (like the child tax credit, new homebuyers tax credit, and deductions for pension contributions) are phased out as incomes rise. America's reliance on means-tested programs stands in contrast to most European countries, where universal benefits are far more prevalent. Many European countries pay, for instance, a universal child credit to all parents, and provide universal health coverage (not only to the elderly) regardless of income.

Means-tested benefits have the advantage of being more narrowly targeted, and therefore less costly, than universal benefits. One disadvantage, however, is that a means-tested benefit imposes an implicit marginal tax on people with earnings close to the income level at which the benefit phases out. For instance, for an eligible taxpayer (or couple) with two or more children, the benefit received through the Earned Income Tax Credit declines by 21 cents for each dollar earned above the prior year's income, imposing an implicit marginal tax rate of 21% on recipients' increased wages.

When one considers the implicit penalties for earning additional income imposed by all federal welfare programs, and combines them with explicit federal and state taxes, households in poverty can face higher marginal tax rates on taking a new job or working more hours than do households earning significantly more (as much as five times the poverty level, in fact).

The negative effects of these incentives on work are mitigated, however, by the fact that most people cannot easily choose how many hours to work. For the overwhelming majority of Americans, the choice is essentially to work a full-time job (40 hours per week) or not to work at all. Moreover, because of the complicated ways in which federal welfare programs and taxes affect one another, the effective marginal tax rates can be nearly impossible to decipher; their negative incentives thus become difficult to act upon. The result is that most people in the prime of their working years — whether they qualify for government benefits or not — will work as much as they can, regardless of marginal incentives.

But the story is different with older Americans. Often, seniors have far more flexibility — fewer immediate financial obligations, and more accumulated resources to cover the obligations they do have — as well as real choices about whether to work and how much to save. Marginal incentives can therefore make a very real difference for them (as we shall see). This is one reason why means-tested benefits — though common among federal programs targeted at the poor — have, for the most part, not been applied to Social Security and Medicare.

Of course, means tests for these programs have not been entirely rejected; there are some very modest instances of the practice in both Social Security and Medicare. For example, since the Social Security reforms of 1983 — implemented to improve the program's shaky finances — a rising share of Social Security benefits has become subject to income taxes. Currently, retirees with "combined incomes" — defined as adjusted gross income plus non-taxable interest income, plus one-half of the Social Security benefit — between $25,000 and $34,000 pay the income tax on up to 50% of their benefits. Seniors with combined incomes over $34,000, meanwhile, pay the income tax on up to 85% of their benefits.

This tax structure can be considered a means test, especially as the taxes levied on the first 50% of benefits flow directly back to Social Security (the taxes levied on the increment between 50% and 85% of benefits flow to the Medicare program, a policy instituted under President Clinton's 1993 budget package). Taxing Social Security benefits to fund Social Security in this way constitutes an effective 3.4% "clawback" in the total benefits of high-income seniors. And crucially, because the income thresholds for benefit taxation are not indexed to inflation or wage growth, ever-increasing shares of total benefits will become subject to this means test over time. By 2050, nearly 5% of total Social Security benefits will be taxed back.

But such taxation of Social Security benefits has at least a plausible foundation in the principle of tax neutrality — by which a tax system is designed not to influence investment, the allocation of capital, or other individual economic decisions — and thus is not simply a means test. While the 6.2% employee share of the payroll tax is paid after income taxes, the matching employer contribution is not subject to either personal or corporate income taxes. As a result, imposing income taxes on half of a person's Social Security benefits is similar to the taxation of 401(k) or Individual Retirement Account withdrawals — essentially just deferring taxation from one's working years to one's retirement. Moreover, the effective increase in marginal rates through the taxation of Social Security benefits is modest. Researchers at the Urban Institute have found that eliminating the taxation of Social Security benefits entirely would reduce the implicit tax on continued work in retirement by only three to four percentage points.

Medicare offers a clearer case. The premiums for Medicare Part B (which covers physician services, outpatient care, and medical equipment) and Part D (which covers prescription drugs) are higher for wealthier retirees. For instance, Part B premiums range from $110 per month for retirees with incomes of less than $85,000 to $369 per month for retirees with incomes over $214,000. The result is that, for seniors earning between $85,000 and $214,000 a year, an implicit effective tax rate of around 2.2% of income is imposed by Medicare. Today, roughly 5% of retirees pay higher Part B premiums based on their incomes (though this share will rise to around 14% over the next decade because of Obamacare, which ends the practice of adjusting income thresholds for inflation). Under the Part D prescription-drug program, around 3% of beneficiaries currently pay income-adjusted premiums; this number, too, will rise under Obamacare, to around 9%.

These are both exceedingly modest means tests, however, and bear little resemblance to the sorts of eligibility tests employed in programs intended for the poor. Our old-age entitlements thus retain a fundamentally universal character — for as much as there is disagreement over entitlement reform, one broadly unifying theme has long been a commitment to universality. For example, the bipartisan Advisory Council on Social Security, appointed by President Clinton in 1994, was at odds over several major issues — but one conclusion the group supported unanimously was that "[c]onventional means-testing of Social Security is unwise." The group further declared:

The fact that benefits are paid without regard to a beneficiary's current income and assets is the crucial principle that allows — in fact encourages — people to add savings to their Social Security benefits and makes it feasible for employers and employees to establish supplementary pension plans.

Moreover, means-testing would send the wrong signal to young people and wage earners generally. The message would be: "If you are a saver and build up income to supplement Social Security, you will be penalized by having your Social Security benefits reduced." This message is both unfair to those who work and save and creates the wrong incentives.

These statements represent the conventional wisdom on both sides of the entitlement debate, although different ends of the political spectrum have opposed means-testing for different reasons. Thus, if any form of means-testing is to offer hope of resolving the problems with Social Security and Medicare, it is essential for reformers to understand the nature of this opposition — and then formulate an adequate response.

LEFT AND RIGHT

People on the left tend to oppose means-testing based on the view — perhaps most notably espoused by Wilbur Cohen, an official of the Social Security Administration right after the program's founding and later Lyndon Johnson's secretary of Health, Education and Welfare — that "a program just for the poor is a poor program." They believe, in other words, that Americans will not adequately support (and fund) policies in which redistribution toward the poor is too overt. For instance, in the late 1990s, traditional liberals like Senator Edward Kennedy opposed increases in Medicare premiums for high earners, while in the 2000s influential liberals argued that adding personal accounts to Social Security would weaken support for the program among middle and high earners, who would see most of their benefits coming from their individual accounts rather than from the traditional program.

But this view is empirically dubious. Programs such as the Earned Income Tax Credit have retained bipartisan support and even been expanded; the Medicaid program, which provides health care to the poor, is among the largest and fastest growing in the federal budget — costing federal and state taxpayers more than $400 billion each year. Moreover, the argument that, because Americans won't support redistribution if they know it is taking place, policymakers should therefore disguise redistribution by mixing it with benefits paid to middle and high earners is insulting to voters. It implicitly holds that most Americans will act only out of selfish motives, and so must be convinced that redistributive programs in fact serve their own interests. It also suggests that government aid to the poor must be rooted in a lie.

Strangely, the concern that voters will not pay for government programs from which they themselves do not benefit hasn't chilled the left's fervor for progressive funding of entitlements. After all, high earners already pay more into Social Security and Medicare than they are likely to receive back in benefits. And most proposed liberal reforms to entitlement programs would simply extend this trend, requiring the wealthy to pay more while they are working. For instance, the most popular Social Security reform proposal on the left these days is to raise (or eliminate altogether) the annual cap on earnings subject to the Social Security payroll tax, currently $106,800. Eliminating the cap would constitute an enormous tax hike on individuals earning more than the current maximum, effectively raising the top marginal tax rate by 12 percentage points.

It is not clear why large benefit cuts for high earners would reduce their support for entitlement programs, as the left believes, but even larger tax increases would not — unless we assume that the best-educated and hardest-working Americans are extremely bad at math. One potential answer is that tax increases on high earners are more acceptable simply because they can raise more money for the program than can benefit cuts applied to the same people. Clearly, there is a limit to how much government can cut a person's benefits: The top Social Security benefit for a person retiring in 2011 is around $29,000 a year; no more than that amount can flow back into the program's coffers from cuts to his benefits. Eliminating the payroll-tax ceiling, however, could raise vastly more for the Social Security program: Under current rates, lifting the cap would mean garnering an additional $29,000 a year from every person earning $350,000 — and much more from people earning still higher incomes. Moreover, the savings from cutting or reducing a senior's benefits would accrue only in the years between his eligibility and death. By contrast, the added revenues from lifting the tax cap would stream onto the program's balance sheets over Americans' entire working lifetimes.

The right, meanwhile, has tended to oppose means tests for economic reasons. Chief among them is that reducing a person's government benefits as his outside income increases creates a disincentive to work and save. In other words, means tests can produce implicit taxes every bit as harmful as explicit taxes. And since we want individuals to work and save more to provide for themselves in retirement (let alone to sustain the American economy), it would be both counterproductive and unfair to penalize them for doing exactly that.

There is some evidence to support these arguments. For instance, seniors who retire early and earn money between today's Social Security early-retirement and full-retirement ages (that is, between the ages of 62 and 66) have their benefits reduced during those years by one dollar for every $2 they are paid over the annual earnings limit of roughly $14,000. In effect, retirees perceive this earnings test as a 50% marginal tax on income over the $14,000 threshold, and research indicates that many retirees earn up to the threshold and then cease working in order to avoid the earnings test. As it happens, an unadvertised aspect of the Social Security earnings test is that, as soon as an "early" retiree turns 66 and reaches the full retirement age, his benefits are increased to account for any past reductions he experienced because of the test. The "earnings test" is therefore not a true tax: Lifetime benefits remain roughly the same with or without the test. Nevertheless, it is widely perceived as a tax, and the effects on seniors' earnings behavior are significant and troubling.

Many conservatives are also concerned about the administration and enforcement of means tests. To run a means test, the government quite obviously needs to know the scope of a person's means, entailing a more intrusive IRS than many would prefer. For instance, individuals do not currently file information regarding the value of their assets or bank-account balances along with their tax returns, but this information would be necessary to administer an effective means test. Moreover, as is painfully evident in Medicaid's provision of long-term care for the elderly (which is available only to seniors below a certain income and asset threshold), potential beneficiaries will seek to avoid a means test if possible by, for example, shifting assets to family members. This makes the administration of means-tested programs a race between individuals seeking to game the system and administrators devising new rules to prevent them from doing so — a pattern that often results in gross inefficiency and waste, as well as fraud.

Others argue that wealthier Americans have put more into the Social Security and Medicare systems during their working years, so that reducing their benefits to pay for those who have put less into the system is tantamount to a tax increase, and would mean turning a social-insurance program into a more redistributive welfare program.

Still, despite these concerns, support for large-scale means tests has been growing among some conservatives in recent years. They view means-testing as a way to generate quick and significant savings without undermining the safety net for the truly needy. Conservative commentator Charles Krauthammer, for instance, has called for means-testing Social Security "so that Warren Buffett's check gets redirected to a senior in need." Mitch Daniels, Indiana's Republican governor and a prominent fiscal conservative, has also called for a Social Security means test for better-off retirees. And many on the right have suggested that Medicare benefits should be significantly greater for seniors in need, so that people who can afford to pay more can help avert the program's looming fiscal collapse.

But this newfound enthusiasm for means-testing has too often failed to account for the real-world consequences of such tests.

IMPLICIT TAXATION

To illustrate both the positive potential and the risks involved in means-testing our old-age entitlements, consider the provisions of a budget-reform plan proposed by the conservative Heritage Foundation in May. The plan, which Heritage calls "Saving the American Dream," would include (among other reforms) means tests for both Social Security and Medicare benefits.

It would begin reducing Social Security benefits for retirees with annual incomes of $55,000 (excluding Social Security benefits), and would then gradually reduce benefits for seniors with incomes above that level until completely eliminating benefits for retirees with incomes above $110,000. According to the Census Bureau's Current Population Survey, in 2009, the typical retiree with an income of $55,000 to $59,000 had a Social Security benefit of around $14,000 per year; thus, under Heritage's proposal, as a senior's income rose from $55,000 to $110,000, his Social Security benefits would fall from $14,000 to zero. In effect, the individual would lose around 25 cents in Social Security benefits for each dollar of income above $55,000. This would generate an implicit tax rate of 25%, in addition to other taxes this person might pay.

The plan includes a similar approach to Medicare. It proposes transforming the Medicare program into a premium-support system similar to the one envisioned by House Budget Committee chairman Paul Ryan, in which the government would contribute toward individual purchases of private insurance coverage. The Heritage proposal, however, would adopt a much more significant means test than the Ryan approach. The premium-support benefit would be modified based upon the age, health costs, and income of each senior, gradually phasing out for seniors with incomes between $55,000 and $110,000. Assuming an annual government contribution of $11,000 — roughly the level of per-beneficiary Medicare spending today — the Medicare and Social Security means tests combined would produce an implicit marginal tax rate of around 46%.

While Heritage cites the current taxation of Social Security benefits and progressive Medicare parts B and D premiums in defense of its reforms, the effects of its proposed means tests would be more than an order of magnitude larger. Even under Heritage's accompanying proposals to reform the tax code — replacing the current system with a flat tax of about 25% — total marginal tax rates in retirement could reach 72%.

These results could, of course, vary based upon other policy parameters. For instance, if, at the same time it imposed a means test, the plan adopted a significant, across-the-board tightening of Social Security and Medicare benefits for all seniors, the implicit tax rates imposed by the means test would be smaller — simply because fewer benefits would be available to be "lost" through the earning of extra income. Alternatively, if the income thresholds at which the means test was applied were not fully indexed for the growth of incomes over time, the means test could apply to larger and larger numbers of retirees each year, and its effects would be greater.

The precise long-term effects of such a plan are therefore difficult to pin down. But the broad conclusion should be clear: Means tests can impose very high implicit taxes on work and saving upon individuals whom we would otherwise wish to encourage to work and to save more. The growing embrace of means-testing on the right is thus somewhat puzzling, given the opposition of conservatives to far smaller increases in top marginal income-tax rates proposed by the Obama administration.

That puzzle can be at least partially solved by considering what else means tests offer: a vastly improved budget outlook, intensely craved by fiscal conservatives. Heritage's budget plan was produced at the invitation of the Peter G. Peterson Foundation, which asked six prominent think tanks spanning the ideological spectrum to provide their blueprints for addressing the federal government's fiscal problems. Of the six (one of which was produced by scholars at the American Enterprise Institute, including me), Heritage's was the only plan that would actually balance the budget within the coming decade; that balance would be reached largely by reducing Social Security and Medicare outlays by more than a full percentage point of GDP over a four-year period. Other proposals settled merely for so-called "primary balance," in which spending (excluding interest payments) would not exceed revenue. But while this approach would stabilize the ratio of government debt to GDP, it would not bring revenues fully into line with outlays.

The means by which the Heritage budget would achieve its impressive feat highlights a crucial point: It would be difficult, and perhaps impossible, to bring the federal budget into balance quickly without touching the Social Security and Medicare benefits of those already on the programs' rolls. The budget deficit is too large, and the non-entitlement elements of the budget too small, to make balancing the budget possible without affecting current retirees. Traditionally, reformers have pledged to protect seniors receiving benefits and people near retirement against any changes, in a (generally unsuccessful) effort to neutralize seniors' opposition to entitlement reforms. This strategy made sense as long as fiscal calamity remained in the distant future and the true budget challenge could be addressed over the long term. But in today's environment of mounting annual trillion-dollar deficits, slow and steady may no longer be a viable approach.

A significant overhaul of entitlements in the near term may therefore be simply unavoidable. And of the various options available, providing less generous benefits to wealthier retirees certainly seems like a prudent and effective way to go about such changes. Yet a stubborn truth remains: Straightforward means-testing will have some enormously counterproductive consequences. So what can reformers do?

A BETTER WAY

There is an alternative approach — one that achieves many of the ends of traditional means-testing, but without inviting many of its drawbacks. The plan's essential and distinguishing feature would involve limiting benefits based not on individuals' incomes in retirement, but rather on their lifetime earnings. As noted above, Social Security already effectively does this, by paying proportionally lower benefits to people with higher average lifetime earnings. Today's reformers could do the same, if to a greater degree.

For instance, the Social Security Administration already tabulates individuals' average lifetime earnings as an intermediate step in calculating their retirement benefits. Reductions in Social Security or Medicare benefits, or increases in premiums, could be based upon this measure of average lifetime earnings rather than on income in retirement.

Unlike a strict means test, this approach would avoid creating powerful disincentives to save. Indeed, individuals facing lower Social Security benefits in retirement would have an incentive to save more during their working years to make up for the loss. And these increased savings would benefit not only the savers, but the economy as a whole: To the degree that retirement savings generally take the form of investment portfolios, increased saving for retirement would generate more investment capital — in turn boosting economic output and wages over the long term.

Incentives to work would still be undermined somewhat under such an approach. Reducing benefits based upon lifetime earnings would mean that, for each additional dollar an individual earns, the return through Social Security would be smaller. Thus, the "net tax rate" under Social Security — that is, the net value of taxes paid and benefits earned — would increase, and obviously higher tax rates discourage work.

But it is likely that these negative effects would be significantly smaller under a lifetime-earnings approach than under a traditional means test. After all, these net tax increases would be relatively small at any given time, since they would be spread out over a full working lifetime rather than concentrated in retirement. This makes an enormous difference, as public-finance economists hold that the negative effects of taxes — known as the "deadweight loss" — rise with the square of the tax rate. The implication is that it is far better to spread a smaller tax increase over a larger number of years than to concentrate a large increase over a shorter period of time. In effect, this approach would take the favorite maxim of tax reformers — broaden the base while lowering the rate — and apply it to reductions in entitlement benefits, yielding, in essence, a lower effective tax over a larger number of years.

Moreover, as noted, there is reason to believe that people are more responsive to marginal tax rates during or near retirement — when the option not to work is clearly available — than during their prime working years, when the need to support a family and save for retirement means that most people will work as much as they can. In a 2008 study of workers over the age of 70, Lucie Schmidt of Williams College and Purvi Sevak of Hunter College found that "a reduction in the marginal tax rate that would increase the payoff to working by 10% would increase labor force participation by 7.5% among men and 11.4% among women." These are far greater responses to tax incentives than have been found among younger workers. For the population as a whole, the Congressional Budget Office assumes a response about one-third this size. The effect on marginal tax rates of the means test proposed in the Heritage budget would be almost four times larger than the increase studied by Schmidt and Sevak — and in the wrong direction — and so would be likely to yield a powerful undesirable response. Similarly, a study by Eric French of the Federal Reserve and John Jones of the State University of New York found that labor-force "participation decisions are most sensitive to financial incentives when workers are old." Means tests would thus impose powerful negative incentives precisely when people are most responsive to them.

Of course, the major problem with cutting benefits for well-off retirees based upon their lifetime earnings is that such a reform would require a lifetime to take full effect. And the federal budget doesn't have a lifetime to wait. This dilemma highlights the cost of delay on entitlement reform: We have known since the late 1980s that Social Security would require reforms beyond those passed in 1983; had these more significant fixes been passed at the time, they would now be fully effective. They would have put the program on a sustainable path while giving individuals enough time to adjust their saving and retirement plans.

The same is true of Medicare. The scope of the coming problem, and the shape of a plausible solution, have been clear since at least the mid-1990s, when a bipartisan Medicare reform commission appointed by President Clinton and congressional leaders endorsed a premium‑support structure that closely resembles the one House Republicans are pushing today. Time, then, really is the resource that budget reformers need the most — yet it has become the most elusive.

If we wish to balance the budget in the near term, the key is to devise approaches that generate targeted, immediate savings without imposing significant disincentives to work and save. In economic terms, we wish to impose a negative "income effect" — that is, a loss of income that individuals will seek to make up through additional work — without imposing a negative "substitution effect" that reduces the rewards to additional work.

Striking this balance will not be easy. If immediate reductions in Social Security and Medicare benefits are required, then the best option might be a combined approach: changing the benefit formula based on lifetime earnings so that the reform takes effect over the long term, but also implementing shorter-term changes that, while rooted in the same principle, are more incremental. For instance, we might pay wealthier individuals with higher Social Security benefits lower annual cost-of-living adjustments than those receiving lower benefits. A progressive COLA could reduce high-end benefits by reasonable amounts in the near term while generating incentives — not disincentives — to work or save. A policy in which the highest third of beneficiaries received no COLA, the lowest third received a full COLA, and the middle third received half the current COLA would reduce Social Security outlays by around 12% over the first ten years. In fact, the savings from this measure alone would be enough to balance the program's finances over the long term.

There could be many variations on such an approach, of course, and they need not necessarily involve the COLA. But all must seek to reduce outlays while hurting neither the most vulnerable retirees nor Americans' incentives to work and save.

In Medicare, the fiscal challenges are even greater, but the basic logic of reform would be much the same. For instance, Medicare beneficiaries might be subject to an annual deductible based on their average lifetime earnings.

FAIR AND SUSTAINABLE

It is inevitable that Social Security, Medicare, and other government programs will become less generous toward the rich than they are today. The only alternative is ever-increasing taxes and their toll on personal welfare, individual freedom, and economic growth.

It is impossible to support a European-style universal welfare state without European levels of taxation, so the provision of relatively generous entitlement benefits to rich and poor alike must shift over time to a more targeted approach. But that targeting must bear in mind the crucial insights of market economists in recent decades — that individuals and the economy as a whole are complex and dynamic, and respond to incentives in ways that can stymie the seemingly simple answers suggested by static economic analysis.

Means tests are a good example of this very principle. On paper, they can provide significant budgetary savings while sparing the poor from benefit reductions. In practice, however, they can generate disincentives to work and to save, and do more harm than good. The challenge for today's policymakers is to craft policies that make the most of the benefits of means-testing while avoiding its worst consequences.

Andrew G. Biggs is a resident scholar at the American Enterprise Institute.