MOST PEOPLE IN THE PAST never lived to be old. And those who did rarely lived much longer. The wealthy could live out their days off rents, profits, and interest payments. But most people worked for most of their lives. They turned to their children, or charity, only if they could work no longer.

In the twentieth century, however, "retirement" from labor at the end of life emerged as a new socioeconomic institution. Most of us now live to be old. We generally withdraw from the labor force in our early sixties then live, for about two decades, on income provided via federal government transfer payments and private retirement plans that our employers and the government subsidize heavily. In this old-age economy, government's role is pervasive. Social Security alone provides more than half the cash income to more than half of the nation's elderly households.

But the nation's ability to maintain this retirement system is now at risk. The reason is demographic. We have many fewer children than our parents and grandparents had, and we live much longer. If nothing changes, the size of the U.S. work force should remain all but static for the foreseeable future. But the number of retirees will take a giant leap between 2010 and 2030, as the baby boom generation retires; it then will continue to grow, albeit more slowly, as longevity continues to rise. The result will be a dramatic decline in the number of workers supporting each older American -- from 3.3 currently, to 2 by 2050.

This greying of America raises two challenges. The first is whether the economy can produce enough goods and services to meet the needs of this expanding dependent population. The second is how to transfer these goods and services to the elderly. The magnitudes needed will test both the government's ability to tax, and the ability of financial investments to yield a secure stream of old-age income.

THE FORSEEABLE FUTURE

Our children should be able to produce enough output. According to the Social Security Administration (SSA) "intermediate" projections, they should be able to pay our old-age pensions and Medicare claims in full and still have more take-home pay than we do today.

But barely. Projected after-tax incomes in 2030, after 30 years of "progress," are 15 percent above current levels. And these projections are very uncertain. Given the normal dispersion of wage growth across the business cycle and among different workers, many of our children could suffer extended periods of lower real incomes. The cost of our support would be especially burdensome if the losers turn out to be our low-paid offspring. So we might want to assure a larger economic pie, even if it involves sacrifices today in the form of saving more and consuming less.

The second concern is whether this rise in old-age dependency will overstretch the federal government's fiscal capacity. In 2050, according to the intermediate projections, Social Security expenditures will equal 17 percent and Medicare 11 percent of payroll. But lifting the payroll tax to 28 percent, from its current 15.4 percent rate, raises the "political risk" of a tax revolt. Since our children would collect from the same schedule of benefits, but face this much higher tax rate, they could claim a gross intergenerational inequity and legislate a cut in our benefits.

A 28-percent tax rate would also degrade the efficiency of the economy. Levies on wages create a "wedge" between what workers pocket and what they produce -- as measured by what employers are willing to pay. As the wedge increases, more employer-employee deals become uneconomic. More workers remain unemployed, take on a different line of work, work a different number of hours, or choose less productive but more remunerative "under-the-table" employment.

If we saved more today, we could increase output tomorrow and make our children richer on an after-tax basis. But saving would not substantially lower the payroll tax rate earmarked for cash benefits. This is because while we are working, Social Security indexes our accruing pension claims to wage growth. (Once we begin collecting, our allowances are indexed to prices.) So the system's obligations rise roughly in line with wage growth, albeit with a lag. Saving should make our children richer, both before and after tax; but it does little to lower the tax rate they face.

Nor could we readily use other taxes to fund our old-age benefits. Programs for the elderly, including supplemental cash payments and a large portion of Medicaid expenditures, already absorb about half of all federal revenues. The greying of America thus will challenge the government's overall fiscal strength.

REFORMING THE SYSTEM

Our initial response to the looming demographic shift came in the Social Security Amendments of 1983, legislation that focused only on the Social Security pension program. The law reduced future burdens by scheduling a gradual rise in the normal retirement age, from sixty-five to sixty-seven, by the year 2027; it subjected the benefits of high-income recipients to income taxation and funneled the proceeds back to the Social Security program; and it accelerated scheduled hikes in the payroll tax, which has led to the buildup of an enormous Social Security trust fund. This buildup was to smooth our transition to the new demographic structure. It was to prefund a portion of future benefit outlays. By augmenting the nation's pool of savings, it was also expected to expand investment and thereby enlarge the nation's capital stock and the future economic pie.

But the reforms did not resolve the problem (see sidebar), and the task of completing the job fell to the 1994-1996 Social Security Advisory Council, chaired by economist Edward Gramlich. The Social Security pension program, the focus of the Council's work, faced an immediate funding shortfall that economist Alan Auerbach estimates at 3.25 percent of the Social Security wage base If the levy were raised that much today, trust-fund income plus payroll tax revenues should be able to pay scheduled benefits in perpetuity. Any delay in raising taxes or cutting benefits, however, would widen the shortfall.

The Advisory Council rejected such a tax increase. It agreed to increase the income taxation of benefits, force state and local workers into the system, accelerate the rise in the "normal" retirement age, then index that age to longevity. The Council then divided sharply into three different camps. And the resulting three proposals frame the current policy debate.

Maintenance of Benefits (MB)

The group led by Robert Ball, a former Social Security commissioner, proposed a plan that preserved the current schedule of benefits. This group included the representatives of women and labor, constituencies that stand to lose the most from a diminution of the program.

Social Security is designed to provide all Americans with a minimum old-age income, and most elderly widows and low-income recipients rely on little else. The expansion of benefits between 1968 and 1972 cut the poverty rate from 30 to 15 percent of the elderly and pegged allowances to wage growth (and thus to living standards). With Medicare, which covers health-care costs, government programs now secure the economic standing of the elderly. A benefit cut could undo this achievement and push many women and low-income workers in our generation into privation, if not into poverty.

MB does not lower benefits for high-income workers. That would make Social Security more of a welfare program and threaten the system's political support. Cuts for middle- and upper-income workers would also reduce their already low "money's worth," a notion that presents old-age benefits as a financial return on a worker's contributions. Many people use money's worth to gauge system efficiency and fairness, even though the measure ignores the "worth" that workers get from payments made to their parents.(As economists Eugene Steuerle and Jon Bakija have shown, upper-income parents actually have gotten considerably larger Social Security transfers than low-income parents.) But the demographic shift sharply reduced all money's worth calculations, and MB took care not to reduce them any further.

Although MB increases taxes (in 2045!) to help fund the program, a significant deficit remains. The solution Ball's group came close to proposing (in the end they only called for its "serious consideration") was to invest 40 percent of the Social Security trust fund not in U.S. Treasury bonds, as the current law requires, but in a corporate equity "index fund."

The reasoning is clear. Over any extended period ending in the present, stocks have delivered significantly higher returns than Treasuries. If such returns continue, they could greatly ease the burden of funding the MB Social Security program.

But along with higher expected returns comes greater expected risk. Who bears the risk? Presumably our children. In the event of a prolonged market slide, they could face a 28 percent Social Security-Medicare payroll tax rate, plus the labor-market inefficiencies a tax that high would generate, before they otherwise might. If the economy and the equity markets both do poorly, depressing our children's after-tax incomes, we run a significant risk of a tax revolt to force us to share in their misery. That MB does not increase the saving rate -- does not call for sacrifices on our part to ease our children's burdens -- only augments the prospect for discontent and a cut in our benefits.

Individual Accounts (IA)

The smallest faction, composed of Edward Gramlich and Marc Twinney, retired director of pensions at Ford Motor Company, would fix the payroll tax at its current level and trim benefits to fit within the resulting revenue stream. Their IA plan also requires all workers to invest, through the Social Security system, in one of several low-cost "index" funds. The accumulations would raise the nation's pool of savings and the balances, annuitized at retirement, would augment our old-age incomes.

To freeze the tax rate, IA must trim allowances either across-the-board, or by targeting specific groups.

The 1983 reforms did both. Raising the "normal" retirement age trimmed benefits across-the-board Because Social Security cuts the benefits for "early" retirement, raising the "normal" age cuts the benefits for retirement at any age. The entire Advisory Council would speed this increase, then index the normal age to longevity.

This rise in the normal age, however, basically matches our expected rise in longevity. It maintains the division between the active and retired portions of adulthood. Looked at this way, the increase in the retirement age stabilizes, more than it cuts, our old-age income benefits.

But the targeted cuts in 1983, and those IA proposes, are true reductions. The 1983 reforms means-tested the program. They subjected the benefits of better-off recipients to income taxation and returned the proceeds to the Social Security system. The whole Advisory Council endorsed an expansion of benefit taxation. IA would also trim the before-tax allowances paid to middle- and upper-income recipients.

Both IA and the 1983 reforms protect low-income recipients, in part, because Social Security pays rather meager benefits. The average benefit for recipients retiring in 1994 was $750 a month. While workers can raise their pensions if they postpone retirement, low-income workers generally are less able to extend their careers; low income, itself, reflects limited demand for their labor. So maintaining the minimum benefit protects those of us who will have few economic options as old age approaches.

IA also adds a mandatory savings component -- the "individual accounts." It requires an additional 1.6 percent contribution that the Social Security Administration deposits, in our name, in index mutual funds that we choose.

This program raises the national saving rate and should expand investment, the nation's capital stock, and the size of the economic pie. Some of this saving would likely be undone. To maintain our standard of living, some of us would save less in other areas. But young and low-income workers tend not to save very much; unable to offset these mandated contributions, they, at least, are forced to save more.

Requiring contributions to "individual accounts" may be a far more politic way to increase the saving rate than raising the payroll tax. Increments to these accounts seem a more real and secure gain in wealth than the enlargement of Social Security's invisible trust fund. And the contributions are ours, individually, and they exactly equal what we put in; what each of us gets for our payroll-tax payment, by contrast, is far from clear. So compared to a tax increase, it is argued, mandated saving generates less political resistance, avoidance, and labor-market inefficiency.

Gramlich and Twinney estimate that if we invest our individual accounts in equities as we currently do our 401(k)s, and investment returns replicate those in the past, then our old-age income would be much the same as currently promised by Social Security.

The essential difference, of course, is risk. How much income our accounts will provide is uncertain. But IA does convert our balances into inflation-proof annuities, eliminating two basic sources of risk (inflation and excess longevity). Since this income is over and above our basic allowance, we run no increased risk of impoverishment. And if our annuity seems too low, we might work a bit more to increase our Social Security pensions and save a few more paychecks.

While IA increases our financial risks, it could cut the political risk to our old-age incomes. Flattening the benefit structure does make Social Security more of a welfare program and diminishes the "money's worth" for middle- and upper-income workers. But IA limits the burden on our children by freezing the tax rate and by shielding them from the financial risks implicit in MB. By pushing up the saving rate, IA also equips our children with more capital goods. They should have higher incomes and, hopefully, more magnanimous attitudes toward their parents.

Personal Savings Accounts (PSA)

The final group, led by Sylvester Scheiber, a vice president at Watson-Wyatt Worldwide, and Carolyn Weaver, of the American Enterprise Institute, would also cut government pension benefits and mandate individual retirement saving. But their PSA plan shifts the old-age income system from public to private mechanisms to such an extent that it represents not a midcourse correction, but a radical transformation.

PSA uses only half the system's current revenues for old-age pensions -- the tax our employers pay; the amount we pay as employees goes to a personal savings account. So PSA can offer only half the government benefits that IA provides. To assure the greatest cushion against destitution, it gives us each the same amount, whether our employer contributed a lot or a little on our behalf -- $410 a month at "normal" retirement, with that amount indexed to wage growth. The rest of our "Social Security" income then comes from our savings accounts.

We have great flexibility in how we handle these accounts. We can withdraw it all at age 62. We can invest it all in common stock. MB and IA both use equities to ease the strain on our old-age income system. PSA gives us far more opportunity to travel that road, and to do so individually, not collectively.

PSA solves certain problems. For the sum of money converted from a tax to a forced contibution to a savings account, it eliminates the "money's worth" issue and reduces avoidance and its associated costs.

PSA also raises the nation's saving rate -- though not by replacing a tax with mandated saving. PSA honors existing commitments to retired and active workers. So for a time, the government needs to borrow (i.e., dissave) more money than we deposit in our savings accounts. PSA increases the saving rate only by extinguishing this debt over a 72-year period -- by imposing a tax equivalent to 1.52 percent of payroll.

Despite its advantages, PSA has drawbacks. First are its administrative costs. Expenses should run about 1 percent of assets -- more on small and less on large accounts. Because MB and IA use the Social Security machinery and index funds, costs run .01 and .10 percent, respectively. Since long-run real returns should range between 1 and 8 percent, depending on the assets chosen, PSA's costs are a serious concern.

But the basic problem is risk. Savings accounts offer no assurance of income to the end of one's life. Nor do they automatically protect us from inflation. Most problematic, PSA makes investment income, generated in these individual savings accounts, the primary vehicle for transferring output to the elderly.

PSA projections assume we invest our accounts in equities as we currently do our 401(k)s, and that our returns will replicate those of the past. But equity returns are volatile. And the demographic shift poses added risks. If the slow-growing U.S. labor force were to have little use for additional capital, returns could tumble and our accounts, plus our cut-back government pensions, would yield a meager existence. But only if we invest in equities can PSA project that we will collect, on average, more than Social Security currently promises for the sums we currently contribute.

Making PSA even riskier, the plan parallels in the public sector the private sector's shift from "defined benefit" pension to "defined contribution" retirement programs. So the risks in these two sources of old-age income are similar and our fortunes in retirement become highly sensitive to fluctuations in inflation, investment returns, and our own longevity. Should our incomes fall when we are old -- and well they might -- we would not be able to respond as we could today, by working harder or taking a second job.

PSA's guaranteed benefit, indexed to wage growth, is projected to yield the same real income as today's minimum benefit. But poverty is relative, as well as absolute. So this might not be enough. To assure an adequate income, PSA might need a larger, costlier safety net or controls on individual financial decisions.

We could protect ourselves. In our accounts, we could purchase federal inflation-proof securities and inflation-proof annuities based on those securities. But our returns would be quite low. After paying PSA's high administrative costs, this risk-averse strategy yields much less old-age income than does any other plan.

RETHINKING RETIREMENT

Low fertility and rising longevity have been good for our individual, per-capita incomes. Fewer children mean fewer young dependents to feed, educate, and equip to make a living. So more remains for us to enjoy. And longevity is among the finest enjoyments our economy provides.

But the new demography complicates the way we finance old age. The payroll tax currently extracts, for our elderly parents, a portion of the expanded aggregate income that we, their working children, earn. But the growth of aggregate labor income soon will slow to a crawl. We are raising barely enough children to replace ourselves. And we are not investing enough to significantly expand their incomes.

So to enjoy an old-age income as "adequate" as our parents do today, we must burden our children or burden ourselves. Only if we burden ourselves, saving more and consuming less, do we raise the odds that our children's after-tax wages will grow as we age, and we can extract enough output to support ourselves.

Because we do not have as many children as our parents did, the payroll tax would need to rise substantially if we fund our Social Security and Medicare claims on a pay-as-you-go basis. But since we do not invest as much in children, we should be able to afford alternate investments, in a retirement financing system, that could also support us when we grow old.

No Advisory Council faction proposed that we increase national saving and make Social Security solvent by raising the payroll tax and investing in bonds. We instead have three plans that cut Social Security's cost by making the program less social and secure.

Our path would seem to depend on our view of retirement. Federal programs now underwrite both a spell of leisure after our working careers are done and "true" old age, when we can no longer work. Social Security and Medicare provide a minimum level of support; Social Security adds an earnings-related supplement tied to earnings-related contributions; and private retirement programs deliver supplements that not all elderly share, and that carry significant risks.

The benefits we can least afford to lose -- or risk -- are those that provide the minimum level of support and those that arrive close to the end of our lives.

For benefits above this minimum and arriving earlier in life, we need to know their cost in financial, labor-market efficiency, and political terms. We must then decide what benefits to keep in the public system, and who bears the risk of a funding shortfall; whether to add a mandatory private saving program that carries risk for the elderly; and what to merely promote, by using the existing tax incentives and regulations for voluntary private plans.

And we must then begin paying our bills.

What lies ahead

The Social Security Administration is currently accumulating a trust fund to help fund future outlays and -- by expanding the nation's pool of savings -- to thereby increase investment and future output and wages.

But current SSA projections have old-age pension outlays exceeding system revenues (tax receipts plus trust-fund income) by 2019. Social Security must then sell trust-fund assets to pay benefits, which would tap the nation's pool of saving and reduce the stock of capital. The system is projected to sell its last bond in 2028, and Social Security's effect on the capital stock, wages, and output then would be totally undone.

The system presumably would proceed on a pay-as-you-go basis, with no augmentation of wage rates or national output. Social Security's portion of the payroll tax would cover just 75 percent of promised benefits; including Medicare, which has a much larger long-term shortfall, total receipts would cover just half of scheduled expenditures. So benefits must be halved or the tax rate doubled -- to 28 percent.The system presumably would proceed on a pay-as-you-go basis, with no augmentation of wage rates or national output. Social Security's portion of the payroll tax would cover just 75 percent of promised benefits; including Medicare, which has a much larger long-term shortfall, total receipts would cover just half of scheduled expenditures. So benefits must be halved or the tax rate doubled -- to 28 percent.