Skip Navigation

Mr. Chairman, I appreciate the opportunity to appear before this Subcommittee to discuss the implications of revising Social Security's investment policies.

Many people are now worried about how best to prepare for paying benefits to the large cohort of baby boomers that will retire after 2010. In response to that concern, some policymakers would like to change the investment policy of the Social Security trust funds, while others favor introducing individual retirement savings accounts to replace part of the Social Security system. By law, reserves in the Social Security trust funds currently must be invested in special U.S. Treasury securities ("special issues"). But some critics argue that the future of the trust funds and the nation's economic prospects could be improved if this investment policy was changed.

Some analysts claim that changes in Social Security's investment policies could have significant effects on the national economy. Policies that seek to increase saving and investment in the economy as a whole could potentially increase resources available for both retirees and working people. Even if they were successful, however, such policies would probably do little directly to solve the narrow problem of financing Social Security. Under current rules, economic growth increases benefits nearly as much as it increases financial flows into the funds. But additional growth might ease the funding problem indirectly by providing more resources all around.

Nevertheless, the two approaches reviewed by the Congressional Budget Office (CBO)--letting the trust funds hold private securities instead of special government issues, and privatizing a portion of the funds along the lines called for in the Individual Social Security Retirement Account Act of 1993 (H.R. 306)--are unlikely to be particularly effective in increasing growth and indeed might have undesired effects on the distribution of income.¹ Unless government purchases of private securities were so large as to dominate the market, they would have little effect on borrowing costs for private borrowers. Moreover, private lenders would have to be content with lower-yielding government securities.

Letting individuals take control of their Social Security accounts might generate more interest in saving for retirement and raise private saving, but it would also introduce a host of changes in the way today's Social Security system affects the distribution of income. Mechanisms could probably be devised that would preserve the long-established principle of providing higher benefits relative to earnings for retired workers with low lifetime earnings. But maintaining the myriad other features of the system--such as benefits to disabled workers, survivors, and spouses or the indexed annuity feature--would be much more of a challenge. Moreover, individual management of retirement funds does not assure high rates of return. People could unexpectedly find their retirement goals undermined by the volatility of markets or their conservative investment strategies.

Policies that would privatize a portion of the Social Security system could have profound budgetary impacts, thereby affecting national saving and future growth. For example, the Individual Social Security Retirement Account Act of 1993 (H.R. 306) would reduce the combined employer and employee payroll tax rate by 2 percentage points, require employers to deposit an amount equal to the tax reduction in special savings accounts for their employees, but not change the benefit provisions of the Social Security Act. H.R. 306 would reduce payroll tax revenues over the 1995-1999 period by $305.5 billion and raise federal outlays by $5.7 billion. Taxes paid on withdrawals from these accounts would raise $4.6 billion during the same period. On net, then, the bill would raise the total deficit between 1995 and 1999 by $306.6 billion. The bill would continue to add to deficits in future years.

Other legislation that established individual retirement savings accounts under different conditions might avoid some of these negative budgetary effects.
 

COULD REVISING THE SOCIAL SECURITY TRUST FUNDS' INVESTMENT POLICIES PROVIDE MORE RESOURCES FOR THE RETIREMENT OF THE BABY BOOMERS?

The Social Security program promises levels of support to retirees that reflect concerns about the distribution and adequacy of benefits. No single policy prescription can solve the problem of how best to provide these benefits to future retirees.

Unless policymakers wish to augment transfers from young to old, the only way to expand resources available to future retirees is through faster economic growth. Indeed, expanding the size of the economic pie would raise living standards for all ages. Faster growth can most reliably be attained through increased national saving--saving by individuals, businesses, and governments--to spur investment and raise productivity. Faster growth in productivity--the amount each worker can produce--leads to faster economic growth by expanding the production of goods and services that can be distributed among the population. Not only would the elderly enjoy some of this bounty, but the working people who must contribute to the Social Security system to pay for the benefits of retirees would gain as well.

Although faster economic growth would help to finance the retirement of the baby boomers indirectly by making more resources available to all, it does not solve all the problems of retirement funding. For example, more rapid growth does not make funding much easier for either Social Security or private pension plans of the defined benefit type in which benefits are tied to the growth of real wages. As real wages rise, so too do benefits, meaning that the flow of revenues into the system approximately keeps pace with the flow of benefits going out of the system. If revenues and benefits do not match, higher growth does little to correct the problem.

Approaches That Claim to Encourage Economic Growth

Some approaches to changing the investment policy of the trust funds would require investing in securities that in turn finance investments in sectors thought to foster economic growth. For example, in the past, advocates have urged that the trust funds invest in state or municipal government securities, an education-loan bank, or community-loan funds to provide more public infrastructure, education, or training opportunities.

However, switching the form in which assets of the Social Security trust funds are held would offer little or no improvement in the outlook for the wider economy or even for specific sectors of the economy.

Using less Social Security money to help finance the deficit in other federal accounts would mean only that private saving that now is invested in nonfederal securities would have to finance that part of the Treasury's borrowing. Hence, redirecting money from Social Security to investments other than Treasury debt could not expand nonfederal investment as a whole.

Having the trust funds invest in particular nonfederal sectors would probably not even significantly expand funds available to those sectors or significantly reduce the cost of borrowing. Private investors readily shift their funds among different assets in search of the highest returns. If a significant flow of Social Security money into a particular security depressed the return of that security just a bit relative to others in the market, investors would take their funds out of those assets in search of others with higher returns. As a result, borrowing costs and the total supply of funds would change little.

Small markets in certain securities could, however, provide exceptions to this rule. If the trust funds bought all or most of the securities offered in the education-loan market, for example, it would probably affect their rates of return. But this step would also effectively confer authority on the trust funds to make and execute policy, since it could independently determine subsidies provided to various sectors or regions. Not least, some danger exists that overinvesting in a particular sector could reduce the productivity of the economy and reduce the rate of return in that sector.

Approaches That Claim to Improve the Investment Performance of the Social Security Trust Funds

Another approach to revamping Social Security is to enhance the investment performance of the trust funds, perhaps with an eye toward expanding benefits or reducing the financing burden. Under this approach, the trust funds could hold high-yielding corporate securities rather than special Treasury securities that earn relatively low yields.

Even if higher returns were achieved, however, the cost of the Social Security system would remain unchanged--at least as measured by the benefits it pays, which have to come out of the goods and services produced by workers. And better investment performance of the trust funds would reduce income earned by other investors, in precise proportion to the amount that it increased returns to the funds.

Providing additional resources to retirees would clearly increase the main burden of retirement on the economy--namely, the resources that would have to be transferred from workers to retirees. Hence, the decision to provide such additional resources might best be made directly, not as a byproduct of a financing decision.

At the same time, changing the portfolio mix implies that the trust funds would have to take on the increased risks that, in the private market, justify high returns for corporate equities or corporate bonds. Businesses can fail, dividends can be cut, and stock prices can fall. High rates of inflation would batter the value of the special Treasury securities that the trust funds now hold, but none of these other risks apply to the special securities.

Moreover, the functioning of the economy would not necessarily be improved if the government, through the Social Security trust funds, were to underwrite substantial portions of the risk of private ventures by holding corporate equities. Many people would argue that private markets and private investors are better at evaluating those risks than the government would be.

Approaches That Attempt to Give Individuals More Control over Their Social Security Funds

Approaches that would allow individuals to choose how to invest part of their Social Security contributions would enhance private control over retirement funds and presumably reduce the financing burden on the economy. These approaches would establish mandatory defined contribution pension plans--called individual retirement savings accounts (IRSAs) for the purposes of this analysis--and would divert at least a portion of the Social Security payroll tax to such accounts.

Such approaches raise a number of thorny issues, including how the accounts could be integrated with the current system, whether such accounts would meet expectations for their rate of return, and whether they would undermine the redistributive goals of the current system. A further question is whether saving by individuals for retirement outside of Social Security would be raised or lowered if payroll tax revenues were shifted to IRSAs.

Giving individuals more control over their Social Security contributions could either increase or decrease overall private saving. If baby boomers, who have expressed some skepticism about the Social Security system, became more confident that retirement benefits from Social Security or IRSAs would be available, discretionary saving for retirement could decline. If early withdrawals were allowed, as is true for individual retirement accounts now, retirement saving could decline even further. However, individuals might take a greater interest in providing for their retirement if they were constantly reminded of the need to save for retirement. In that case, individual saving for retirement might rise.
 

IS PARTIALLY PRIVATIZING THE SOCIAL SECURITY SYSTEM FEASIBLE WITHOUT FURTHER CHANGES?

If some of the money currently going to the Social Security trust funds were transferred to individual retirement savings accounts, something would have to be done to maintain the long-run financial integrity of the Social Security system. It clearly could not maintain the same levels of benefits with lower income for a long period of time. Although the offsetting changes in revenues or benefits are not spelled out in some of the approaches, the options are clear: benefits might be reduced to match the lower cash flow, payroll taxes might be increased (thus undermining one of the major arguments for privatization), or the system might get subsidies from general revenues.

One country--Chile--that has already taken the step of privatizing a segment of its social security system illustrates the choices that must be made. Chile has privatized pensions for new employees as well as for more than 90 percent of existing employees who chose to join the new system. The remaining existing employees continue to pay into the state system and receive benefits under it, with some slight adjustments in the structure of benefits. Because the Chilean system, like the U.S. system, was not fully funded but had a large pay-as-you-go element, the end of contributions has left the system with substantial shortfalls that must be met with subsidies from general revenues. A considerable amount of government debt will also be created in the future to meet the deficits of the old system, provide minimum benefits, and redeem the bonds that were issued to existing employees for prior service when they joined the new system.

Similarly, changes that would partially privatize the U.S. Social Security system raise the question of how benefits for current and soon-to-be retirees would be financed. Increasing payroll taxes is not likely to be acceptable, since one of the main reasons for changing the current system is concern over current levels of the tax. In fact, since the inception of the system, policy has been aimed at maintaining the appearance that Social Security pays its own way. Thus, subsidies would be likely to attract considerable opposition as well. Approaches that privatize a part of the system therefore would probably require a reduction in benefits paid out by Social Security.
 

COULD PEOPLE MANAGE THEIR OWN FUNDS BETTER THAN THE SOCIAL SECURITY SYSTEM?

Although allowing individuals to choose how to invest a portion of the contributions in IRSAs could not be relied on to encourage more saving for retirement, it might have other benefits. Some people might feel more confident that retirement funds would be available. But policymakers would like to know whether workers could manage their retirement funds as well as or better than the Social Security system.

Investments in private securities, either corporate bonds or equities, do offer relatively high returns. Whether such high returns would be realized on average would depend on how much risk workers would be willing to assume in their IRSAs, how diversified their portfolios would be, and whether future asset returns follow their historical patterns. And unless national saving is higher than it otherwise would be, incomes would simply be shifted from one group of people to another without raising the level of resources available to all.

The issue is further complicated, since the Social Security system computes benefits in a complex way that provides substantially different returns to different participants in the system. Thus, some individuals might benefit and others could lose from any privatization of the system.

Investment Choices by Individuals

Many people tend to invest conservatively in low-risk, low-return investments when they control their own retirement funds. Three existing vehicles for retirement investment--401(k) plans and the self-directed retirement funds for federal workers and for teachers and professors--tend to be heavily invested in fixed-income securities such as bonds and guaranteed investment contracts (GICs) that invest in a fixed-interest contract with an insurance company. Higher-return but higher-risk investments such as equities or stock mutual funds draw only 35 percent to 50 percent of the amount invested.

If this preference for low-risk, low-return investments also pertains to IRSA holders, individual choice is unlikely to increase the resources available to baby boomers in retirement as long as national saving does not change much. Assets in IRSAs, if conservatively invested, would probably not earn much higher returns than currently accrue to the trust funds, and some individuals might find that their market rate of return was in fact below what they had expected and planned for.

Historical Returns on Investment Portfolios

Although future returns are uncertain, the historical returns on stocks, corporate and government bonds, and Treasury bills indicate the possible range of future relative returns. The volatility of returns suggests varying degrees of risk.

Between 1926 and 1987, stocks were by far the best long-term investment on average. Common stocks earned a compound average return of 9.9 percent a year in nominal terms over that period, while the compound annual return for long-term corporate bonds was only half as high. Inflation averaged 3 percent. Among government securities, the average return on long-term government bonds lay somewhat below that of long-term corporate bonds, and U.S. Treasury bills with maturities of less than one year showed an even lower average return that beat inflation by only a small margin.

The riskiness of each of these broad categories of investments, however, varies with the average compound rate of return. In other words, categories with higher average total returns are more likely to show rates of return in any one year that are substantially above or below the average for the period. Investors obtain more reliable--though lower--rates of return when they invest in short-term securities of the federal government than when they invest in common stocks.

Corporate securities are riskier than government securities in another respect--they are far more heterogeneous. Treasury securities differ only by their maturities and coupons, but corporate securities require a whole range of additional considerations related to competitiveness and management. These factors affect dividends, stock prices, and the prices of bonds, and thus the return of a particular portfolio of securities. The impact of this risk can be reduced by diversification--by holding a widely spread portfolio of securities with different characteristics so that gains in one area will offset losses in another.

To illustrate the possible range of returns for individual investors who invest in common stocks through mutual funds, CBO calculated compound total returns on the top 10 percent and the bottom 10 percent of mutual funds that invested in common stocks over the 1984-1993 period.² During that particular period, the compound total return for those funds with yields in the top 10 percent of stock mutual funds was 18.4 percent a year, while the return for those funds with returns in the bottom 10 percent of stock mutual funds was 5.2 percent.

Such a range of returns means that investors could have had very different experiences had they invested their Social Security contributions in stock mutual funds during the 1984-1993 period. A lucky investor, who earned an average annual return of 18.4 percent, would have seen an initial $1,000 investment burgeon to $5,414. An unlucky investor, who earned just 5.2 percent a year on average, would have had an initial $1,000 investment grow to only $1,660. Moreover, the unlucky investor would have had a much greater probability of negative returns in any one month.

Clearly, the experience of workers whose retirement funds are invested in IRSAs is likely to be widely disparate. Even among those who use mutual funds to diversify their investments, some will do very well and others will not keep up with inflation. Moreover, the outcomes for some could be even worse than the weakest experience described here, since the mutual funds that had the worst performance over the past 10 years probably went out of business and are thus not even reflected in the data. Workers who purchase securities directly will be exposed to yet greater risk.

Ways do, however, exist to minimize the risk--for instance, rules might specify that IRSAs could be invested only in very broad index funds. But such rules would also reduce the possibility of high returns and would give individuals little latitude in managing their retirement assets, thus undermining one of the goals of privatization proposals.

Other Types of Risk Associated with Individual Retirement Savings Accounts

Benefits received from an individual retirement savings account would depend not only on the rate of return earned on the account but also on the value of the annuity that could be purchased at the time of retirement and on the administrative costs associated with maintaining the accounts.

Unlucky market timing at the point that the IRSA is converted into a stream of payments could present additional risk. If the market for stocks suffered a setback at the time of a person's retirement, for example, a significant share of the gains earned over many years could be lost. Careful management, together with rules allowing partial conversion of the IRSA into an annuity over a period of time, could ameliorate the risk from unlucky market setbacks.

Administrative burdens could be severe, especially for the smallest accounts. The many investment choices among individuals, the small sums of money per person, and the obligations to keep individuals informed about their accounts could all be costly. Under some approaches, firms need not pay into an individual's account until the accumulated savings reach some limit. But how those funds are invested in the meantime, and what happens if the firm ceases to exist or fails to make good on the payments, is not at all clear.
 

WOULD INDIVIDUAL RETIREMENT SAVINGS ACCOUNTS BE INCONSISTENT WITH SOME OF THE GOALS OF THE SOCIAL SECURITY SYSTEM?

Benefits under the Social Security system are determined by a complex system that seeks a balance between equity and ensuring an adequate level of benefits even to the poorest recipients. Moving to IRS As would alter this balance. The amount of retirement income derived from funds invested in an IRSA would be determined by the amount invested and the rate of return on that investment. That payout scheme would be equivalent to moving to a defined contribution plan, instead of a defined benefit plan.

The benefit formula itself exemplifies the mixture of goals in the current Social Security system. To help achieve equity, a link exists between the earnings on which individuals pay taxes while working and what they subsequently receive when they retire or become disabled--regardless of their needs. In general, workers with more Social Security covered earnings will receive more Social Security benefits. However, the formula is progressive rather than proportionate, so that replacement rates--the ratio of retirement benefits to preretirement wage incomes--are generally higher for workers with low-wage histories than they are for well-to-do workers. That progressivity reflects the goal of providing benefits to people in need of assistance.

A major concern raised by opponents of approaches that include IRSAs is that individual accounts could undermine the progressive benefit structure of the Social Security system.

If IRSAs were implemented, the impact on the distribution of benefits would depend on the percentage of benefits based on assets in the IRSA. If the current benefit structure was simply replaced with one in which retirement benefits were determined by the amount of money in one's IRSA, then its progressive character would be eliminated. All else being equal, a retired worker who consistently earned one-half of the earnings of the average worker would expect to receive half of the average benefit.³ By comparison, under current law, that low-wage worker receives roughly 65 percent of the average benefit. Similarly, a retired worker whose lifetime earnings were 50 percent higher than those of the average worker could expect to receive 50 percent more than the average benefit under a pure defined contribution plan, rather than about 30 percent above the average benefit under the current system.

Deviations would occur if the rates of return on investments differed. But low-wage workers might not invest their IRSAs in assets that would yield higher returns than those of other workers. Indeed, lower-wage workers might well be more risk-averse and less sophisticated in their investment behavior and therefore obtain lower returns.

However, the goal of redistribution need not be abandoned if IRSAs were adopted. Other federal programs used in conjunction with Social Security benefits based strictly on the value of IRSAs could achieve minimum levels of income or redistributional objectives. For example, the Supplemental Security Income (SSI) program already provides cash assistance to poor elderly or disabled participants. That program could be expanded or modified. One problem that would need to be addressed, however, is whether the existence of SSI or other means-tested programs would encourage individuals to choose riskier investment strategies for their IRSAs or withdraw them early because taxpayers would, in effect, bear a portion of the risk.

Alternatively, a progressive benefit structure could be maintained by using a mixed strategy in which only a portion of the payroll tax would go to the IRSA. If the benefit structure for the remainder was left in place, then the overall system would still be progressive, albeit somewhat less so. One way of doing this would be to lower the replacement rates within each of the three brackets in the benefit formula proportionately, while setting the bend points as under current law. Another approach (which could be combined with the mixed strategy) would be to adjust either the Social Security benefits that would be paid or the percentage of earnings put into an IRSA for individual workers to preserve the current degree of progressivity.

Other aspects of the Social Security system would be more difficult to preserve. For example, the specific treatment of benefits for disabled workers and for survivors and spouses of workers in the present benefit structure might not be easily replicated in a defined contribution plan. And the indexed annuity feature of the current Social Security system, which ensures that benefits are indexed to keep pace with inflation and are paid as long as a person lives, might not be available in private markets.

Indeed, how IRSAs are converted into benefits for retirees would have a large bearing on the well-being of the elderly and the amount of public assistance that is needed to support a minimum standard of living. If IRSAs were paid out as lump-sum distributions at the time of retirement, people who did not invest wisely or who lived unexpectedly long lives might find themselves without sufficient financial resources toward the end of their lives. Public assistance might be required for some. If people chose to convert IRSAs into annuities, any price charged other than the actuarially fair price would deliver windfalls to either the issuers or the recipients, whether people purchased annuities from the government or from private entities.

In sum, moving to a system in which a portion of one's Social Security benefits was based on investments in an IRSA would require decisions to be made about which elements of the current benefit structure should be preserved, modified, or eliminated. Mechanisms could probably be devised that would preserve the long-established principle of providing higher earnings-replacement rates for retired workers with low lifetime earnings. But maintaining the myriad other features of the system would be much more of a challenge.
 

CONCLUSION

No easy fixes to the funding problems of the Social Security system exist. Although redirecting the assets of the Social Security trust funds at first seems to offer some relief to the long-term funding problem, closer examination of such approaches shows that little would change. Social Security benefits must be financed using resources from the economy. Whether those resources are obtained from current taxes or from earnings on assets does not matter much.

Letting individuals take control of their Social Security accounts might generate more interest in saving for retirement, but it would also introduce a host of changes in the redistributive nature of the Social Security system as it exists today. Moreover, better rates of return are uncertain at best. In some cases, individuals could unexpectedly find their retirement goals undermined by the volatility of market rates of return.

1. See Congressional Budget Office, "Implications of Revising Social Security's Investment Policies," CBO Paper (September 1994).

2. Morningstar Mutual Funds OnFloppy (Chicago, 111., March 1994). The Congressional Budget Office weighted the returns by the mutual fund's net asset value so that relatively small funds receive less weight in calculating the average rate of return.

3. This calculation assumes that all contributions would accumulate until retirement. Early withdrawals, of course, would reduce benefits further.