Governments around the world now face the daunting challenge of a rapidly aging population. Although the baby boom generation has accepted this change in demographics, the problem is not transitory. The combination of better medical care, higher standards of living, and improved lifestyles is permanently and irreversibly increasing the fraction of the population over the age of 65 and, even more rapidly, of those over 75 and 85.

This aging of the population means higher government costs for pensions and medical care. In the United States, the Congressional Budget Office estimates that under current law, the combined cost of Social Security retirement benefits and government spending on health care for the aged would increase from 10 percent of GDP now to 18 percent in 2030 and would go on rising after that. To pay for that 18 percent of GDP by raising taxes would require the equivalent of doubling the personal income tax or raising the payroll tax rate from 15 percent to more than 35 percent.

The situation in Europe is even worse. While the United States will see the ratio of persons 65 and over to those aged 20 to 64 rise from 21 percent in 1990 to 36 percent in 2030, the change in Germany will be from 24 percent to 54 percent. The Organization for Economic

Cooperation and Development projects that government retirement benefits (excluding health costs) will exceed 16 percent of the GDP of Germany, France, and Italy by 2030, compared with about 7 percent of GDP in the United States. Although total health care costs are a smaller share of GDP in Europe than in the United States, the European governments' larger role in financing health care means that health spending requires more taxes. The high labor costs that result from high payroll taxes encourage businesses to substitute machinery for jobs and to shift production out of Western Europe. High payroll tax rates in combination with high effective minimum wages have thus been an important cause of the rise in European unemployment rates from less than half the U.S. level in the early 1970s to more than twice its rate today. The European economies may already have increased government spending and taxes to the point where economic activity and employment cannot continue to expand.

Raising future tax rates even further to pay for the benefits embodied in current law would do great damage to the American and European economies. However, cutting benefits enough to avoid a tax increase would mean a significant decline in the relative living standard of the aged and a failure to take advantage of future medical advances. The reductions in pension benefits that are politically possible and morally justifiable and the feasible savings in medical care costs that could be achieved by improved incentives would still require an enormous tax increase. Responsible government officials have therefore concluded that the current system of financing retirement income and health care cannot continue unchanged.

The problem with the current system is that retirees' benefits are financed on a "pay-as-you-go" basis, by taxing concurrent employees. The obvious solution is to shift to a privatized system of pre-funding those benefits through mandatory contributions to individual accounts. The increased savings during working years that such a system of individual accounts entails and the productivity of the resulting increased capital stock would permit future retirement and health care costs to be financed with much lower annual contributions than would be required under a pay-as-you-go system. Moreover, people of low to moderate means would for the first time accumulate large amounts of personal savings.

A system of mandatory individual accounts has already been adopted to finance old-age pensions in Australia and several Latin American countries. Sweden has permitted individuals to shift a portion of their annual payroll tax contributions into private accounts. The British government has reduced the relative value of its basic state pension and earlier this year suggested dropping its pay-as-you-go system completely in favor of mandatory retirement savings accounts. Japanese Finance Minister Hiroshi Mitsuzuka recently suggested that Japan end its current pay-as-you-go pensions and provide a combination of a low flat-rate benefit for everyone and mandatory private savings accounts. In the United States, the official Advisory Council on Social Security issued a report earlier this year with three alternative sets of recommendations, each of which called for some form of pre-funding Social Security obligations by investment in private financial markets.¹ Although there has been no explicit official discussion in Germany, France, or Italy about substituting such accounts for the current national pay-as-you-go systems, only individual funded accounts can provide the portability and labor mobility envisioned in the 1986 Single European Act's attempt to create a single European market for labor and products.

Before describing how a pre-funded system would work, it may help to clarify the nature of Social Security's financial problems. Social Security pensions and disability insurance are financed by a separate payroll tax of 12.4 percent on wages up to a current maximum taxable level of $65,400. The government now collects about 20 percent more from this earmarked payroll tax than it spends on Social Security benefits. The program is thus in surplus, which reduces the size of the federal budget deficit. The excess collections are formally invested in U.S. government bonds that constitute the Social Security trust fund. But by 2015, according to government actuaries, the aging of the population will cause Social Security benefits to exceed tax receipts. The program will then start running deficits that will exacerbate the overall federal budget deficit. After that time, the Social Security program will finance the gap between benefits and taxes by selling bonds now held in the trust fund to the public.. By 2029, all of those bonds will have been sold, and the program will technically be bankrupt. Maintaining the benefits specified by current law would then require raising the payroll tax rate immediately from 12 percent to 16 percent and then gradually increasing it to a long-run rate of 19 percent after 2030. (Projected Medicare and Medicaid health spending on the aged would bring the total payroll tax rate to more than 45 percent.)

Although talk of the looming Social Security bankruptcy has helped promote interest in fundamental reform, in this context the notion of bankruptcy has no real substance. Social Security is said to be heading toward bankruptcy only because it uses earmarked taxes and has a trust find. Other federal programs like education and defense have no earmarked taxes and no trust fund and would therefore never be perceived to be bankrupt. Social Security will only be bankrupt and unable to pay benefits if taxpayers grow unwilling to raise taxes to pay for those benefits, a prospect that becomes increasingly likely as political support declines for what is becoming understood to be a costly and inefficient program.

The trust fund is only an accounting mechanism with no economic meaning. When benefits begin to exceed receipts in 2015, the government must borrow to finance the shortfall. That borrowing will take the form of selling Treasury bonds to the public. It will absorb savings that would otherwise finance private investment in business plant and equipment, thus reducing investment and productivity. It will not matter whether the bonds that the government sells were previously in a trust fund or are newly created.

The problem with the Social Security program is not its prospective financial bankruptcy but that it has become a bad deal for participants. Social Security is now enormously expensive and provides a paltry rate of return on the funds that employees contribute. That low return undermines political support and makes continued reliance on Social Security very risky for future retirees.

The increasing cost of Social Security reflects the growing longevity of the population and the increasing ratio of retirement benefits to wages. Because the ratio of retirees to workers is increasing by more than 50 percent, causing the ratio of retirement benefits to taxable wages to reach 19 percent, the pay-as-you-go Social Security payroll tax rate is heading from 12 percent to 19 percent.

Social Security was once a better bargain. Between 1940 and 1980 the payroll tax rate increased from 2 percent to 10 percent, allowing benefits to rise sharply and to exceed greatly the taxes that individuals paid during their working years. Social Security was a wonderful deal for those who got in on the ground floor in the 1940s, 1950s, and 1960s. The double-digit rates of return that they earned on their tax contributions lent Social Security widespread public support. But those days are gone forever. The relation between the taxes and benefits projected under current law implies a rate of return of only 1.5 percent, and even that low rate is projected to decline.

Although saving during working years is the natural way for individuals to finance the expenses of their own old age, not everyone has the necessary foresight and discipline. In a privatized Social Security based on mandatory contribution, individuals (and their employers on their behalf) would be required to make contributions to individual savings accounts (like the popular 401k plans) that would be invested through mutual finds into diversified portfolios of stocks and bonds. At retirement, most of the accumulated balances would be paid out in the form of annuities.

For most Americans, mandatory contributions to individual savings accounts would represent additional savings that would add dollar tor dollar to national savings and capital accumulation. Although some people may reduce other savings in order to maintain consumption, the vast majority of Americans do not have other savings to reduce; half of American households at age 6o have accumulated financial assets equal to less than six months of earnings. The new savings would flow into investment in business plant and equipment. The interest and dividends earned in those mandatory accounts would also be saved until the individuals reach retirement age, adding further to the national savings available to finance new business investment.

Over the past four decades in the United States, the real (net of inflation) rate of return

on such investment has been slightly more than nine percent. That will undoubtedly sound high to those accustomed to lower returns on their individual savings. The difference reflects the big bite that taxes take out of that nine percent return before it reaches individuals. Corporations pay 40 percent of the return as taxes to federal, state, and local governments, reducing it to 5.4 percent. That is just about the return that a "tax free" investor (like a private pension, IRA or 401k account) would have earned over the past 40 years by investing in a portfolio of stocks and bonds. But the government could rebate to each account the extra 3.6 percent it would collect in additional tax revenue from the profits earned on the increased stock of business capital.

In contrast to this nine percent return based on the productivity of the capital stock, contributions to a pay-as-you-go system earn a much lower return equal to the rate of increase in tax revenues resulting from the growth in incomes and the labor force. As noted above, pay-as-you-go Social Security contributions now earn an implicit return of about 1.5 percent. That return will decline to 1.1 percent over the coming decades as labor force growth declines further.

A funded account that earns a nine percent real rate of return can provide for retirement income at a fraction of the cost that would be required in a pay-as-you-go system. To get a sense of how big this difference would be, think about a 45-year-old (in the middle of his working years) who earns $40,000 a year. At the current 12.4 percent tax rate, he and his employer now pay $4,960 annually in Social Security taxes. With a rate of return of 1.5 percent, that $4,960 would grow to $7,753 when the man is 75 years old (in the middle of his retirement years). In contrast, in a funded system with a nine percent rate of return, accumulating that $7,753 at age 75 would require saving only $584 at age 45. In short, the $4,960 tax in the pay-as-you-go system could be replaced by $584 of savings or, equivalently, the 12.4 percent tax could be replaced by a mandatory savings contribution equal to 2 percent of earnings.

Although this is just a rough calculation, it captures the potential gain from shifting to a pre-funded system. In a recent study at the National Bureau of Economic Research, Andrew Samwick and I used census and Social Security data to calculate the mandatory contribution that would be needed to replace the payroll tax in the pay-as-you-go system to maintain existing benefit rules.² We found that with a 9 percent rate of return, the 19 percent payroll tax that would eventually be needed could be replaced by a mandatory contribution of only 2 percent of taxable earnings.

A comparable gain could be achieved by pre-funding health care expenditures for the aged. The projected long-run cost of 12 percent of GDP would require a separate payroll tax of 30 percent in a pay-as-you-go system but could be financed by mandatory savings of only about 3.5 percent of earnings in accounts that earn the full pretax real rate of return.

Shifting Social Security pensions from a pay-as-you-go system to a funded program based on individual accounts would gradually lead to a substantial increase in the nation's stock of investment in plant and equipment, raising total national income, real wages, and the general standard of living. Although the mandatory savings rate would be a small fraction of each individual's wages, total additional savings would be large because the nine percent return on the accumulated savings would also be saved until the individual reaches retirement age.

Samwick and I estimated that mandatory savings would cause the nation's capital stock to rise by about 12 percent after 25 years and by about 34 percent in the long run, after 75 years, when the new system is fully phased in. This investment would cause real GDP and average wages to grow more rapidly, raising both an additional 7.5 percent after 75 years.

More important, the combination of the higher real wage and the lower payroll tax in the pre-funded system would cause the real spendable wage income of the typical employee to be more than 37 percent higher than under a pay-as-you-go program.³ Replacing the 19 percent payroll tax with a much lower mandatory savings contribution would also sharply reduce the marginal tax rate of the typical employee. Since high marginal tax rates distort all kinds of economic behavior in wasteful ways -- discouraging work effort and risk-taking and encouraging the substitution of untaxed fringe benefits for taxable wages -- reduction in marginal tax rates would further improve the efficiency of the economy and the real incomes of individual employees.

Although these long-run gains involve an increased cost in the short run, it would be relatively small. Critics of shifting to a funded system often dismiss the possibility by saying that it would require the first generation to "pay twice," that is, paying to support current retirees while also saving for their own retirement. Since the payroll tax is currently 12.4 percent, that appears to imply that the current generation would be required to pay more than 24 percent. Fortunately, that perception is mistaken. The cost is more on the order of an additional 2 percent, with taxes eventually falling.

There are many ways of gradually shifting from the pay-as-you-go system to a fully funded plan at low cost. It is possible, for example, to maintain permanently the level of pension benefits specified in current law, involve no new government borrowing (other than eventually selling the trust fund bonds), limit the initial increase in the combined payroll tax rate and mandatory saving to 2 percent (to a total of 14.4 percent), and in less than 20 years reach a combined total of less than the current 12.4 percent. In the year 2030, when continuing the pay-as-you-go system would require raising taxes to 16 percent, the combination of the payroll tax and mandatory saving would be less than 7 percent. And in the long run, the mandatory saving required with a 9 percent rate of return would be only 2 percent of earnings.

It is useful to contrast this two percent savings rate with the "two percent solution" of those who would preserve the existing pay-as-you-go system by adding about two percent to the current 12.4 percent payroll tax rate for the next 75 years. According to Social Security actuaries, that increase in payroll taxes would temporarily build up the trust fund enough to permit benefits to be paid by a combination of taxes and trust fund bond sales for 75 years. There are two important differences. In the pay-as-you-go solution, the tax rate remains at 14.4 percent for 75 years, while in the transition to the fully funded system it gradually drops from 14.4 percent in the first year to less than 7 percent after 35 years and to only about 2 percent after 50 years. Second, with the pay-as-you-go solution the tax rate would have to jump after the 75th year to 19 percent, while it would remain at about 2 percent with the funded program.

More generally, the small extra cost during the transition years must be compared with the adjustments that would inevitably be made if the current pay-as-you-go system continues. Congress is unlikely to wait until 2030 and then sharply raise taxes. A more likely prospect is gradual tax increases and perhaps benefits reductions before then. The typical current employee might well pay less in taxes and receive more benefits in a transition to a funded system than in a continuation of the existing pay-as-you-go program.

The key advantage of pre-funding old age benefits is the high real rate of return available on increases in the capital stock. Achieving that nine percent return requires additional saving but does not require the savings to be accumulated in individual private accounts. The same gains could in principle be achieved if the government collected the increased savings and invested them in a portfolio of stocks and bonds.

Without extra savings and an increase in the nation's capital stock, nothing would be gained by shifting some of the existing trust fund from government bonds to private stocks and bonds. The trust fund's income would rise, but the income of the private pension funds and other investors that bought the government bonds sold by the trust fund would fall by an equal amount. There would be no net gain in national income. Moreover, getting the private sector to hold more government bonds would raise the interest rate that taxpayers have to pay on the entire national debt. This lack of real economic gain and the increased interest burden on taxpayers are the basic fallacy in the Ball plan, the proposal by the group of Social Security Advisory Council members led by former Social Security Commissioner Robert Ball to invest 40 percent of the existing pay-as-you-go trust fund in private securities.

With extra mandatory savings, a Social Security trust fund managed by the government could in principle finance future benefits with the same low mandatory savings as individual savings accounts. The government could keep separate account records for each individual, effectively treating the funds as a defined contribution plan with the investment decisions made by the government. Or the government could use the funds to finance a defined benefit plan that involved redistribution based on such things as years of work, family status, and income.

Why then would a system of individual accounts be much better than a government fund? There are three reasons. First, a large and growing government fund would undoubtedly tempt the government to run larger budget deficits, thus absorbing some or all of the additional savings and precluding the increase in the capital stock. That is already happening with the existing Social Security funds. Although the U.S. Social Security system is officially "off budget," the Social Security surpluses of the past decade have made it easier for the government to finance deficit spending. In the recent agreement to achieve a balanced budget in the year 2002, the projected surpluses in the Social Security program ($108 billion in 2002) are used to balance deficits elsewhere in the government. Although sophisticated budget analysts might

see that the incremental retirement savings were being used to finance other deficits, the general public's focus on the bottom-line overall government deficit would make it very easy to dissipate the added funds. Past experience with putting Social Security "off budget" shows that accounting techniques cannot prevent this from happening. In contrast, if the new funds are invested in private accounts similar to IRAS and 401k accounts, they could not be used to fund an expanded budget deficit.

Second, a government-managed fund would almost certainly take political considerations into account in its investments. Although advocates of the government fund now say that it would be managed without political interference, legislation to prevent the fund managers from investing in certain industries or in companies that did business in certain countries or that failed to meet certain standards of "good citizenship" in their employment and environmental policies would be likely. That has been the practice of many state pension investment funds, for instance with regard to companies that did business in South Africa, and in federal contracting, as with minority and small business set-asides. It is hard to imagine that Congress would not want to influence the investment of a multi-trillion dollar fund under government control.

Third, to the extent that the government simply "indexes" its fund (that is, invests it in all stocks and bonds in proportion to their value in the market), it would undermine the role of the capital market in channeling funds to the most productive uses. Active private fund managers invest in promising companies with good performance and sell the shares of those that are not well managed. That lowers the cost of funds for the good companies and helps them expand while raising the cost of funds for the poor performers, slowing their growth and making it easier for other firms to acquire them.

Because of the very large amount of funds that would be accumulated, the choice of how they would be managed is critical. Even without pre-funding Medicare, the mandatory pension funds would grow to be about 45 percent of the nation's capital stock exclusive of owner-occupied housing and unincorporated businesses. At a time when the world is moving away from state ownership, it would be a major mistake for the United States government to effectively nationalize almost half of American business.

Opponents of shifting to individual accounts raise two important questions: How would such a system protect low-wage workers and individuals with long bouts of unemployment who benefit from the redistribution of Social Security? How would individuals be protected from the risk of stock market declines?

A pure system of mandatory saving accounts would involve no redistribution. That is why some proponents advocate combining them with a tax-financed uniform benefit to all retirees. Such a universal benefit is, however, a costly and inefficient way of dealing with the problem of old-age poverty since most of the benefits would go to those who would have had high enough income without that transfer. When 20 percent of the population is over age 65, a transfer equal to two-thirds of the poverty line (as proposed by the group within the Social Security Advisory Council that favored individual saving accounts) would require a payroll tax rate of approximately 7 percent.

One simple and more efficient alternative would be to provide a supplement to the mandatory savings accounts of individuals at age 65 whose accumulated amounts are too low to support a satisfactory retirement annuity. Although there is no clear standard for a satisfactory annuity, one possibility is half of the median retirement annuity. In an earlier study, Andrew Samwick and I used individual earnings histories to evaluate the cost of achieving such a minimum standard. We found that, with a mandatory saving plan starting at age 30, about 19 percent of the accounts at age 65 would have less than half of the median value. The cost of bringing them up to half of the median value could be financed by a one-time tax of 4.7 percent on all mandatory accounts at age 65. Each individual could save enough to finance that one-time tax by raising his lifetime mandatory savings by 4.7 percent. For example, if the basic mandatory savings rate was 2 percent of payroll, raising that to 2.1 percent of payroll would provide enough additional accumulation to pay the tax that would bring all accounts up to at least half the median.⁴ This approach would be vastly cheaper than a 7 percent payroll tax.

Many other ways of protecting the poor in a funded system are possible. The key point is that it is not necessary to have a universal benefit in order to guarantee a decent standard of living for all of the elderly and that a targeted approach can be used with relatively little additional taxes or savings.

More generally, one of the attractive features of a pre-funded system is that it can substantially increase the incomes of low- and middle-income individuals during their working years as well as in retirement. As noted above, the combination of a higher real wage and lower mandatory contribution can increase the typical employee's spendable income by more than 30 percent. In addition, if a smaller income gain is taken during working years, the level of retirement benefits can be made higher than they are projected to be under current Social Security rules.

The source of these gains is the creation of a large capital stock owned primarily by the 90 percent of the population with incomes below the maximum taxable income (currently $65,400), who currently save very little. This would represent an enormous change in the distribution of wealth as well as a major shift in the distribution of income in favor of people of low to moderate incomes.

The Social Security program today is a very risky "asset" for anyone who expects to depend on Social Security benefits as the primary source of retirement income. In the past 15 years, Congress has voted to effectively cut Social Security benefits by making benefits taxable, by subsequently increasing the fraction of benefits subject to taxes, and by postponing the age of retirement for anyone born after 1937. There is now widespread discussion of further reductions of "promised" benefits by changing the inflation adjustment, accelerating the shift in the retirement age, and taxing more of the benefits.

Participants in a pay-as-you-go system are always at the mercy of the political process. The political support that previously came from the program's popularity is declining rapidly as voters recognize that the return on their Social Security taxes has fallen close to zero. A 1995 Gallup poll found that about three-quarters of 18-to-34-year-olds and two-thirds of 35-to-54-year-olds had concluded that they would not get back as much in benefits as they and their spouses had paid in taxes.

There is no way to avoid the political risk to benefits inherent in a pay-as-you-go system. By contrast, the risks of market fluctuations of a funded system can be avoided by a relatively small increase in the savings rate during working years, effectively building up a financial cushion to absorb fluctuations in stock and bond prices.

Here is what the historical record teaches us. Over the past 40 years, the combination of the rate of return on a portfolio of stocks and bonds plus the corporate taxes paid produced an average real return of nine percent. Employees who save two percent of their earnings (up to the maximum taxable level under Social Security rules) and earn a nine percent rate of return would be able to draw an annuity equal to the benefits promised by current Social Security rules. However, although nine percent maybe the expected rate of return, the fluctuations in stock and bond prices imply that the actual rate of return might be higher or lower. Someone who saved just two percent of wages would be equally likely to have too little or too much at retirement to buy an annuity that just matches the currently promised Social Security benefits. Moreover, during retirement the fluctuation in stock and bond prices could also reduce the value of the variable annuity below the expected level.

To protect themselves from that uncertainty and to virtually guarantee having enough to fund the benefits implied by current Social Security rules, individuals would only have to increase their savings rate from two percent of earnings to three percent of earnings. That would build a sufficient cushion of extra savings to virtually rule out the possibility -- less than one chance in 1,000 -- of not being able to fund the annuity. That implies that the government would take virtually no risk if it guarantees benefits equal to those projected in Social Security law to anyone who invests 3 percent of earnings in a diversified portfolio of stocks and bonds.

The financial problems of the Social Security and Medicare systems that have attracted so much attention should be regarded as providing a political opportunity to undertake reforms that would be desirable even without the aging of the population and the financial shortfalls in these entitlement programs. The shift from the existing pay-as-you-go system to a pre-funded system based on mandatory saving can lead to a dramatic reduction in the future cost of the program and a dramatic increase in the standard of living of most lower- and middle-income individuals. It is difficult to think of any other policy that could produce such a substantial permanent rise in the standard of living of the vast majority of the population.

¹ Report of the 1994-1996 Advisory Council on Social Security, Washington: U.S. Government Printing Office, 1997.

² "The Economics of Prefunding Social Security and Medicare Benefits," NBER Working Paper, in 1997 NBER Macroeconomic Annual, ed. Ben S. Bernanke and Julio Rotemberg, Cambridge: MIT Press, forthcoming.

³ Each $100 earned under the existing pay-as-you-go system would be subject to federal and state income taxes of about 20 percent, plus a payroll tax of 19 percent, leaving $61 of spendable income. With a pre-funded system, pretax earnings would be 7.5 percent higher, or $107.50. After a 20 percent income tax and a 2 percent mandatory savings contribution, spendable income would be $83.85, more than 37 percent higher than $61 under the pay-as-you-go system.

⁴ Martin Feldstein and Andrew Samwick, "The Transition Path in Privatizing Social Security," NBER Working Paper 5761, 1996, forthcoming in Martin Feldstein, Privatizing Social Security, Chicago: University of Chicago Press, 1998.

MARTIN FELDSTEIN is Professor of Economics at Harvard University and President of the National Bureau of Economic Research.