1994-96 Advisory Council

1994-1996 Advisory Council on Social Security

Social Security for the 21st Century:

A Strategy to Maintain Benefits and Strengthen America's Family Protection Plan

Robert M. Ball, Edith U. Fierst, Gloria T. Johnson,
Thomas W. Jones, George Kourpias, and Gerald M. Shea


For 60 years the United States has pursued a three-tier retirement income policy consisting of Social Security and two supplementary tiers: employer-sponsored pensions, now covering about half the work force, and voluntary individual savings. Each tier complements the others and has become a fixed feature of national policy. Social Security, covering just about everyone, is a contributory, wage-related, defined-benefit plan administered by the federal government and entirely supported by dedicated federal taxes, and the two supplementary tiers are explicitly promoted by federal tax policy.

Social Security, the key to this three-tier approach, has been a uniquely successful program by any measure. For more than half a century, it has served as America's family protection plan, providing millions of the elderly and disabled with secure incomes, guarding them against impoverishment, and relieving their children and grandchildren of what could otherwise be the unmanageable burden of supporting them throughout their old age.

No program has ever done more to prevent and alleviate poverty or to protect income against erosion by inflation. None has done more to protect children against impoverishment when a wage-earning parent dies or becomes disabled. And no social program has ever enjoyed greater public support.

Nor has any program, public or private, been more carefully managed to maintain long-term stability. Over the course of six decades, Social Security's trustees, administrators, and actuaries have guided the program in adapting to major demographic trends and economic developments -- some foreseeable, some not.

Social Security has been self-supporting; it has not added one cent to the deficit. Dedicated Social Security taxes have raised $5 trillion since 1937, when collections began, and the program has paid out $4.5 trillion, leaving $500 billion in reserve. Today nearly everyone who has been working for any substantial period has built up sizeable credits toward specifically defined Social Security benefits, and it is no exaggeration to say that just about every American has an important stake in the future of the program.

Has the United States now arrived at a place where, as some argue, the key to this three-tier system is failing and requires fundamental change? We say no. We believe that the system has been and continues to be a major success, and that with relatively minor modifications it can provide a sound basis for planning the future.

Financing Social Security

We are concerned that much public discussion of the future of Social Security is based on misinformation about its financing. Social Security is not facing a crisis. The program, as currently structured and financed, and without changing a word of present law, can pay full benefits for another 30-plus years. After that, the reserves that are now accumulating in the program's trust funds would be used up, but there would still be an income stream from ongoing dedicated taxes, which would support about 75 percent of the cost of the program. Even 75 years from now, current-level taxes would cover about 70 percent of program costs.

In other words, the program in the absence of any changes is not headed for bankruptcy: it would continue to function, although paying substantially reduced benefits. The changes that we will propose will make it unnecessary to cut benefits in any such major way, if at all, but the point is that in balancing the program for the long run we will not be starting from scratch after 2030. Most of the cost after that date is met by the provisions of present law. It is simply incorrect to assume that if the trust funds were to be used up, Social Security would be out of money, and it is similarly unrealistic to assume that the President and Congress would stand by and let the trust funds reach the point of exhaustion -- not when just about every American family has a major stake in the program.

Social Security's financing has always been carefully watched. The Trustees -- three Cabinet officers, the Commissioner of Social Security, and two representatives of the public, appointed by the President and confirmed by the Senate -- continually scrutinize the financing of the program and report on it every year in great detail. In addition, the law provides for the program to be independently examined by advisory groups composed of private citizens. And the current Social Security Advisory Council, as has been true of many others, also appointed an outside group of economists, actuaries, demographers, sociologists, and other experts to help evaluate the program's financial condition. The Council and the expert group concur in the Trustees' assessment that the program as presently financed and structured does not face a crisis but does face a long-term deficit, beginning about 30 years from now1/.

The whole point of this careful attention to long-run financing is to anticipate the need for changes far ahead of time, since a distant deficit can be eliminated with moderate measures if action is taken early. That is the case with the approaching long-range deficit: We can see it on the horizon and we can explore a variety of ways to eliminate it and continue paying full benefits after 2030. We do not have to have major cuts in benefits, major increases in taxes, or major alterations to the fundamental principles of the program.

The point bears repeating: Social Security is not about to expire and does not require heroic measures. Rather, the situation with Social Security is like that of homeowners living in a sound house that they very much like and that needs only to have its mortgage refinanced. There is no need to tear the house down, remodel it, or trade it for a different house. The need is only to improve its long-range financing.

The Council's Failure to Agree

Despite more than two years of discussion within the Advisory Council, members have been unable to agree on a single plan. Six of us support the Maintain Benefits (MB) plan; five support the Personal Security Accounts (PSA) plan; two support the Individual Accounts (IA) plan -- and the differences have proved to be profound.

The six of us who support the MB plan are opposed to any so-called "reform" of Social Security that makes major reductions in Social Security protection and substitutes compulsory savings. As we demonstrate, there is no need for this kind of radical change, and it would undermine rather than strengthen Social Security's unique role in providing multi-generational family protection.

We propose to strengthen the basic Social Security system. The IA and PSA proposals would not. The IA plan would, over time, cut average Social Security benefits by 30 percent while substituting a compulsory savings plan that is designed merely to make up for the 30 percent cut. This plan comes out at the same place as present law -- on average, and on average only -- but at the cost of increasing workers' payroll taxes, beginning in 1998, from the present 6.2 percent of wages to 7.8 percent.

The PSA plan would gradually abolish the present Social Security system, substituting a low flat benefit of $410 per month for full-time workers (well below the poverty rate for a single elderly person, now over $600) coupled with a compulsory savings plan intended to provide the rest of Social Security's retirement protection. This plan would require raising the combined payroll tax on employers and employees from 12.4 percent to 13.92 percent and, in addition, would require borrowing, at the peak, $2 trillion (in 1996 dollars) from the Treasury over the next three decades, repaying the loan for several decades more. During much of that time the federal deficit, currently being brought under control, would increase again, by hundreds of billions of dollars.

Putting it bluntly, our view is that this Advisory Council report does not offer three acceptable plans from which to choose. We believe that two of the plans are inherently and fundamentally flawed.

The Council has not only been unable to agree on a plan, we have been unable to agree on the proper criteria to use in assessing the plans. Other members of the Council rely primarily on comparing "replacement rates" (the ratio of retirement benefits to the last year of earnings before retirement) and "money's worth" (the ratio of the present value of expected benefits for contributors and their dependents to the present value of lifetime contributions). We believe that these measures, while useful for some purposes, are inade quate and indeed misleading when used as the primary basis for comparing these proposals.


Replacement rates have traditionally been used in program planning as one measure of the level of adequacy for a defined-benefit plan, and they are useful for that purpose. But both the IA and PSA proposals would to a greater or lesser degree convert Social Security to a defined-contribution plan, and both use the average estimated return as part of the replace ment rate. This is not appropriate. In a defined-contribution plan, the actual returns for individuals will vary too widely from the average -- with some falling far below it -- to make an average replacement rate meaningful. Rather than relying on an average rate, policymakers concerned with assessing defined-contribution plans need to take into account the full range of probable experience and whether such a range is likely to be acceptable to the public.

Moreover, replacement-rate comparisons can have validity only in situations where the inputs are comparable. In projecting returns from various investment approaches, for example, replacement-rate charts will inevitably misinform the reader unless the charts also show the variations in the amounts being paid in to the plans being compared. To illustrate, it makes no sense to compare the replacement rate of a plan that keeps Social Security contribution rates at 12.4 percent of payroll (the rate in present law and in the MB plan) with plans that increase the rate to 14 percent of payroll. The increase gives the higher-rate plans much more to invest, thus skewing an unadjusted replacement-rate calculation in their favor while obscuring the fact that the higher contribution rate imposes an immediate and continuing burden on contributors. Obviously if you put more in, you get more out -- but at a cost.


Comparing the plans using "money's worth" criteria (or "internal rate of return," which is another form of the same thing) can also be highly misleading. "Money's worth" comparisons have their uses, for example in comparing within a given plan the treatment of workers born on different dates and thus showing changes in contribution/benefit ratios in a pay-as-you-go plan. But in comparing the three plans proposed by Council members, the "money's worth" figures presented in this report will almost inevitably be misinterpreted. At first glance, the PSA plan appears to score highest. Yet this has nothing to do with setting up individual accounts and allowing individuals to handle their own investments. It simply reflects the fact that through raising taxes and borrowing, the PSA plan is able to put more money in the stock market sooner. When, for purposes of illustration, the same financing is made available to the MB plan that we offer for consideration, the "money's worth" results are comparable (as we show later). But simple comparisons of "money's worth" charts -- once those charts are assumed to be valid -- will lead the unsuspecting reader to conclude that a system of private savings accounts will yield higher returns than a public plan, even though this is not the case. And once that perception takes hold, "money's worth" charts will have done perhaps irretrievable damage to the cause of objectively comparing alternative proposals.

Aside from the drawbacks of "money's worth" charts, we believe that the comparison chapter of the Council report simply gives much too much emphasis to "money's worth" as a concept. It is undeniably important in showing the need to improve the prospects of younger workers, but, as we shall discuss, there are many other criteria that must also be considered in judging these three plans.

The other seven members of the Council also treat reductions in Social Security benefits as a positive criterion in comparing plans, a notion which we strongly reject (see our supplemental statement).

In the following pages, we describe the weaknesses of the IA and PSA plans before moving on to discuss our own proposal, along with some alternative approaches to long-term stability that would also build on rather than undermine the traditional Social Security program.

Individual Accounts (IA)

The Chairman of the Advisory Council and one other member advocate a two-part Individual Accounts (IA) plan:

(1) Reduce Social Security protection so that over the long run benefits are brought into balance with the current combined contribution rate of 12.4 percent of payroll, and

(2) Establish a new compulsory individual savings plan, financed by deducting an additional 1.6 percent from workers' earnings, bringing the total employee deduction to 7.8 percent of earnings beginning in 1998.

Benefits under the Social Security part of the plan would be gradually reduced, ultimately resulting in a cut of about 30 percent in average benefits. For the high-paid (defined as having earnings 60 percent above average), benefits would be reduced 32 percent; for the low-paid (defined as roughly the minimum wage earner), benefits would be reduced 22 percent.

These cuts for future retirees would result from several changes in the law. The effective date of the increase in the normal retirement age (NRA) would be speeded up. It is sched uled in present law to go to 67 for those who reach age 62 in 2022. The 2022 date would be changed to 2011 and the NRA would be increased after that, automatically and continually, by indexing it to longevity. In addition, average indexed monthly earnings would be computed over 38 years instead of 35 as under present law, and the benefit formula would be changed gradually, but substantially, to produce lower benefits.

Ultimately the IA plan, on average, is designed to bring the combined benefits of the reduced Social Security system and the new savings plan up to the level now provided for (but not fully funded) by the present Social Security system.

Individual savers would have a limited choice of investment vehicles, perhaps five to ten funds, managed and invested by the federal government as in the case of the Federal Employees Thrift Savings Plan. At retirement age, the retiree would be required to take out a lifetime annuity underwritten by the government with a guaranteed period of payment and protected against inflation by price indexing. The annuitant also would be required to take a joint and survivor's annuity to protect the annuitant's spouse unless, as is the case with the ERISA requirement for private pension plans, the spouse agrees in writing to waive this right.

On the one hand, the IA plan reduces future Social Security benefits substantially in order to be able to eliminate the existing program's long-term deficit while maintaining the current contribution rate for Social Security. On the other hand, the plan increases deductions from workers' earnings in order to fund the IAs which, on average, are assumed to make up for the benefit cuts. There is no increase in the employers' tax. Given that workers do not come out ahead under this arrangement compared to present law, the IA plan has the effect of putting on employees the full responsibility for eliminating Social Security's long-term deficit.

The Seeds of Dissolution

Looking at the two parts of the plan together through the narrow prism of individual "money's worth" calculations, the IA plan would do better than the current Social Security system, assuming pay-as-you-go increases in the contribution rates to make up for the shortfall in income. However, the residual Social Security part of the IA plan, looked at separately, would not do at all well on the "money's worth" test, as indicated in Figure 1 below (Where the residual part of the IA plan is shown as the Maintain Tax Rate [MTR] line). And as the plan developed over time, with beneficia ries doing less and less well under the reduced Social Security plan compared to individual accounts (at least those of the more successful inves tors), there would be every reason for many average and above-average earners, particularly, to press for further reductions in contributions to Social Security in order to make more available for their individual accounts. Thus the IA plan is inherently unstable, and could lead to the unraveling of the redistributional provisions that are so integral to Social Security and so crucial to its effectiveness.

Ball graph 1

Savings Versus Consumption for Low-Income Families

It is not doing most workers any favor to compel them to save, above and beyond what Social Security already requires, for the exclusive purpose of retirement. Particularly in the case of moderate- to low-income workers, earners who are living from paycheck to paycheck are likely to need to spend whatever they have on food, clothing, shelter, schooling and other immediate needs. And for many workers, protecting against the cost of health care may have a higher priority than setting aside more income for retirement. It is more in their interest to keep the deductions from their earnings at approximately present levels for retirement and to maintain approximately the level of protection provided by the traditional Social Security program.

* Most of the charts accompanying this text are designed to show "money's worth" (that is, the present value of the lifetime protection furnished compared to the present value of employer and employee payments plus interest earnings) for a "composite worker" under various scenarios. The "composite worker" is a theoretical construct representing, in proportion to their presence in the beneficiary population, single workers, married couples with one spouse working, and married couples with both spouses working, assumed to have lifetime earnings in the proper proportion for each demographic group. The Social Security taxes and other payments are the ones appropriate for the birth dates shown on the horizontal scale. The present value of the payments is computed assuming an ultimate 2.3 percent real valuation interest rate (the rate assumed for long-term U.S. bonds). The actual accumulation of contributions in an individual account or trust fund is at the 2.3-percent real valuation rate for funds invested in long-term government bonds and certain other conservative investments and 7 percent real for investments in the stock market. The present value of the Social Security protection furnished takes into account the probabilities assumed in the 1995 Trustees Reports for retirement, disability and survivors benefits, including eligibility for, the benefit rate and the length of payment. On the vertical scale against which "money's worth" is compared for the various plans and sets of circumstances, 100 percent equals the "money's worth" result for a fully advance funded plan with all funds invested so as to earn 2.3 percent real. Since most of the plans fall far short of full advance funding, it is not surprising that they fall short of 100 percent on this scale even though some investments are assumed to yield more than 2.3 percent. The Maintain Benefits (MB) plan with investment in equities, for example, levels off at about 96 percent for workers born after 1996, which represents a return on taxes paid of between 2.2 and 2.3 percent. "Money's worth" for specific demographic groupings, earnings levels, and a variety of other plans and circumstances are shown in the actuarial appendix to the full report.

The IA line assumes asset allocations by individual investors to be the same as in 401(k) plans, i.e., half in the stock market and half in conservative investments such as government bonds.

The Office of the Actuary of the Social Security Administration is the source for all figures and tables in this statement.

Increased Risk for the Individual Family

The IA plan shifts Social Security away from being a defined-benefit plan and toward becoming a defined-contribution plan. This is a bad idea. As previously noted, Social Security is the foundation of our national multi-tier retirement system, with private pension plans and private savings built on top. With more and more private-sector employers offering only defined-contribution pension plans, it is all the more important that the nation's basic plan be maintained as a defined-benefit system, with the amounts available in retirement determined by law rather than by the risks and relative uncertainties of individual investment. With the basic plan secure, some level of risk in defined-contribution pension plans and investment of individual savings is certainly acceptable, since Social Security's basic protection is still there regardless of what happens to the individual's investments. But to shift the base itself, or any part of it, to a retirement plan dependent on individual investment decisions seems unwise.

Moreover, substituting a savings plan designed on average to make up for the reductions in Social Security benefits means that many families -- those who get below-average returns -- would lose out. Lower paid workers are particularly likely to fall into this group. But this is not just a matter of concern for lower-income workers. There is increased risk for everyone in substituting returns on investments for part of the Social Security defined benefit plan.

It is one thing to have retirement income supplementary to Social Security, such as IRAs and 401(k) plans, tied to investments, but quite another thing to have one's basic retirement income dependent on the uncertainties of individual private investment. In that situation, there is more than one kind of risk. In addition to the general risk of picking investments that perform badly, individuals are exposed to the risk of being forced to begin or end an investment period at a bad time. Using projected average investment returns as a basis for retirement planning is treacherous: one can drown in a river with an average depth of six inches, and predicted average investment returns will leave some investors destitute and many others with much less income than expected.

The Catch in Individual Ownership

Although the intent of the IA plan is to create a nationwide system of individual retirement accounts with the principal and income available only in retirement, it is very doubtful that this objective could be preserved in practice. As with today's IRAs and 401(k) plans, there would be pressure to use individual savings accounts for medical, educational, housing, or other needs. If funds go into individually-named accounts, as provided for under the IA plan (whose sponsors even argue that the 1.6-percent wage deduction is not a tax because it continues to be individually owned), account holders will assuredly find it unreasonable to be denied access to their "personal" funds in an emergency -- or indeed for any purpose that seems worthwhile. As a result, the total amounts that would actually be available for retirement under the IA plan, particularly for low-income people, would almost certainly be far lower than predicted.

The IA Plan and the Disabled

Although increasing the age of first eligibility for benefits does not reduce disability benefits under the IA plan, the benefit formula changes do. These cuts -- resulting from a formula change from 90 percent, 32 percent and 15 percent to 90 percent, 22.4 percent, and 10.5 percent -- are particularly harsh for the disabled.

Disabled workers have lower incomes, on average, than retirees. They are less likely to have private pensions or insurance benefits to supplement Social Security and are less likely to be able to get and keep part-time work, and thus they often must rely on Social Security for a larger share of their total income. Further, disabled workers typically have much smaller asset holdings than retirees. This reflects two of the realities experienced by disabled workers: their careers are cut short before they have accumulated savings to supplement Social Security, and they face new disability-related expenses. Unlike retirees, whose regular living expenses may decline when their working-years-related expenses decline, living costs for persons with disabilities may rise when they become disabled and leave the labor force.

The IA Plan and Medicare

Social Security is not the only social insurance program supported by payroll-tax contribu tions. Of the current total combined employer-employee contribution rate of 15.3 percent, 12.4 percent goes to Social Security and 2.9 percent goes to the Medicare Hospital Insurance (HI) fund. If the nation is of a mind to increase deductions from earnings by as much as 1.6 percent, as the IA plan calls for, it would make more sense to divide the payment between employers and employees and assign most of those additional revenues to Medicare. After all, Medicare is just as important as cash benefits to the financial security of retirees and the disabled, and the HI fund faces insolvency in about five years. To ignore its immediate needs while financing a long-term redesign of Social Security strikes us as a serious inversion of priorities. To illustrate this point: If a payroll tax increase of 1.6 percent were to be enacted now with the revenues directed to Medicare, this alone would be sufficient to postpone the HI's fund exhaustion by about 15 years, thus providing a substantial planning period in which to design and implement needed cost-saving changes.

Disruption of Existing Pension Arrangements

In one way or another all private pension plans and just about all government pension plans are built on the expectation that the retirees will also receive Social Security. And pension arrangements, at least in large firms, have also usually been worked out over the years with employee representatives. Pensions are part of the overall compensation bargaining process. What would happen to these plans if Social Security is cut by an average of 30 percent with changes in the benefit formula, the computation period, and the normal retire ment age? At the very least, a long period of uncertainty and unrest could be anticipated. All the pension integration rules would need to be changed as well as many other regulatory provisions. Bargaining over pensions could once again become an important source of labor unrest. Would private plans have to make up the difference -- particularly for middle-aged and older workers -- as Social Security is reduced? Unions would certainly want to retain the equivalent of present Social Security promises in defined pension benefits, and employers would just as certainly resist. No one knows what would happen.

A Cautionary Word

Proponents of the IA plan and others have argued that direct investment of Social Security funds in equities would be undesirable because of the possibility that, in spite of legislative directives mandating neutrality in stock purchases, future Congresses might want to influence the selection of stocks for one purpose or another, such as social investing. Or they suggest that the federal government might try to influence the policies of specific industries or individual companies. One IA proponent has suggested that such investment might funda mentally alter the relationship of government to industry. We do not accept that this is a serious danger under the plan we outline for consideration (as we explain later), but it needs to be kept in mind that in the IA plan the government retains control of the investment of the funds of individual savers. If there were a real risk of departing from a policy of neutrality when the federal government controls investments, the risk would certainly apply to the IA proposal too.

In summary, we oppose the IA plan because it very substantially increases deductions from workers' earnings; because it creates an unstable situation in which many workers will be motivated to shift funds (and support) from a deteriorating Social Security system to a private savings account; because it weakens the reliability of the nation's basic retirement system by shifting it in part from a defined-benefit to a defined-contribution system; because it requires workers to set aside additional income for the sole purpose of retirement, regard less of what they may require for protection against health care costs or to meet other more immediate needs than retirement; because it calls for particularly harsh benefit cuts for the totally disabled; because it makes the problem of adequately financing Medicare more difficult; and, finally, because it is unlikely to be able to deliver fully on its promise to increase individual savings for retirement, since in all probabil ity the savings in the new accounts will eventually be made available for multiple purposes.

For all of these reasons, the IA plan could end up undermining what would remain of the traditional Social Security program, thus leading to even greater substitution of a compulsory private savings plan for social insurance.

Personal Security Accounts

Five members of the Advisory Council propose replacing the present Social Security system with a two-tier Personal Security Accounts (PSA) plan:

(1) A government system of flat monthly retirement benefits, varying only with length of time worked, and

(2) Compulsory private individual "security accounts" (i.e., savings accounts) for retirement, funded by 5 percentage points of the payroll tax now going to Social Security.

The monthly benefit payable by the government system would be $205 after 10 years of coverage (in 1996 dollars, wage indexed thereafter), rising by about $8 for each additional year of coverage until the maximum benefit -- $410 per month -- is reached for workers with 35 years of coverage. Spouses of eligible workers would receive a monthly benefit of $205, and older surviving spouses would get 75 percent of the total flat benefit payable to the couple. A disability and young survivor's program similar to the present system (but ultimately reduced by 30 percent in the case of disability) would also be part of the central government system.

The PSA plan calls for an increase of 1.52 percentage points to the payroll tax, beginning in 1998 and continuing through 2069. In addition, the plan would borrow from the federal government over 33 years - at the peak owing the Treasury about $2 trillion in 1996 dollars ($15 trillion in then-current dollars) - and then repay it from the 1.52 percent tax increase over the following 35 years as the transition costs go down. The tax increase and the borrowing are designed to enable the plan to fulfill the benefit promises of the present Social Security system for those 55 and older (although with some reductions), and to pay for past service credits from the present system to those 25 to 54. All those now under 25 would, at retirement, receive only the flat benefit and whatever the 5 percent of wages invested in the PSAs amounted to.

Individuals would be free to invest their PSAs in any generally available financial instrument, and the accumulated amounts would become available to them at retire ment age, with no requirement for annuitization nor any special provision for spouses or other dependents.

The past service credit for those between 25 and 54 would be an accrued benefit calculated at the point when the transition to the new system begins. It would be arrived at in three steps:

In the first step, a primary insurance amount (PIA) would be determined for 1998 as a disability award is determined under present law. The second step would be to index the PIA from 1998 to the year of retirement eligibility by intervening wage increases in the overall economy. The third step would be to multiply the PIA by a factor reflecting the portion of a worker's career worked under the old system. Basing the calculation on an assumed career of 40 years -- from age 22 to age 61 -- a worker who is 42 at the date of transition, for example, would have an accrued benefit from the old system equal to one half of the PIA (i.e., 20 years under Social Security, divided by a 40-year career span). Those 25 to 54 would receive this accrued benefit, the flat payment, and whatever their personal account (PSA) has produced.

If the present Social Security system's formulas for determining benefits seem bewilder ing, consider how much more so the above formula is. It would be just about impossible to explain to individuals what they were supposed to get, and it is hard to imagine what Social Security's informational pamphlets would look like. They would have to continue to explain the present system, of course, while also explaining the new system and the transitional provisions.

But explaining the benefits formula is just the beginning of the PSA plan's problems. It has all the disadvantages of the IA proposal, as previously explained -- and more.

"Money's Worth" and the PSA

As Figure 2 illustrates, "money's worth" for the PSA can look good in comparison with the IA plan and the MB plan (that is, the MB plan with investment in equities, described later). But this apparent advantage has nothing to do with setting up private accounts. It has to do entirely with the fact that the PSA plan, by increasing taxes and borrowing heavily from the federal government, is able to free up 5 percentage points of the payroll tax at the plan's inception for individual investment, about half of which is assumed to be invested in the stock market. This is much more in stocks than in the case of either the IA plan or the investment option that we describe for the MB plan.*

* Among other potential problems with the assumptions used by PSA proponents, the interaction of PSAs with other retirement income vehicles has not been explored. A spokesman for the Employee Benefits Research Institute (EBRI) pointed out at a recent Capitol Hill briefing that asset allocation under 401(k) plans might be very different under a compulsory savings plan replacing all or part of Social Security. For one thing, if there is no longer a guaranteed benefit as currently provided by Social Security (or if it is substantially reduced), individuals may be more conservative in their investments precisely because they no longer have the same guaranteed Social Security benefit to rely on. (Referred to in Pension and Benefits Reporter, Vol. 23, No. 40, pp. 2369-2425, October 7, 1996)

Ball graph 2

When the MB plan is modified by making available the same amount of money for investment in the stock market, it matches the PSA plan in the same test of "money's worth." In fact, it does better, because its administrative costs related to investment are much lower. Figure 3 compares the modified MB plan (MB+) and the PSA plan.

Ball graph 3
The way in which the PSA plan is shown by its sponsors to produce such high average "money's worth" figures is worth examining. In addition to increasing the tax rate by 1.52 percentage points, it borrows from the government at the assumed long-term average rate of 2.3 percent real, and then assumes that those with PSA accounts invest half of their funds in the stock market at an assumed rate of 7 per cent real. In other words, the PSA plan is only able to show such good returns because, in addition to increasing the tax rate, it borrows large amounts of money from the Treasury at low interest rates and then puts the money into the stock market to earn higher returns. This arbitrage on a grand scale -- and the large amount of money it makes available for investment in stocks -- is the real secret of the seemingly high "money's worth" ratios for this plan.

But there is no free lunch. The apparent "money's worth" advantages of the PSA plan come at the price of payroll tax increases, a very substantial increase in the federal deficit, and increased pressure on all other programs as the government struggles to contain the deficit. Most people, when they understand the trade-offs, are unlikely to think that the PSA plan's "money's worth" gains are worth so much pain.

Paying Twice

In shifting from the present system to individually funded plans, proponents have to meet, at the same time, the cost of paying ben efits to present retirees and those who are nearing retirement plus the cost of establishing the new system. This is often described by proponents as a "transition" cost, but it is a very long "transition" with a very high cost. In the case of the PSA plan, it requires increasing the payroll tax immediately and substantially -- from the present 12.4 percent to nearly 14 percent -- and then keeping it at that level for about the next 70 years while, in addition, borrowing heavily from the Treasury, with the debt to be fully repaid only in the latter half of the 21st century.

To fund this "transition," the first generation or two of workers have to pay twice: first, for the benefits payable to those already retired and to workers with accrued rights under the present system (all those 55 and older, with partial benefits owed to those between the age of 25 and 54 who have earned credit under Social Security); and, second, to build, at the same time, individual funded savings accounts for themselves. And there is no escaping this "transition" cost when moving from a pay-as-you-go system like Social Security to a funded compulsory savings system.

The PSA plan arrives at the transitional cost numbers described because instead of meeting the full costs in the early decades when the transition is most expensive -- over 4 percent of payroll -- the plan averages the cost over about 70 years, and makes up for the lack of sufficient income in the first part of the period by borrowing. It later pays back what it has borrowed from the proceeds of the tax which, after about 33 years, would be higher than necessary on a pay-as-you-go basis. This approach may make the PSA plan more politically palatable at the outset, but there is no avoiding the basic fact that it raises taxes and borrows money for many decades.

Taxing the PSA

The PSA plan includes a tax proposal that has no precedent in the United States tax code, and one that requires much more expert scruti ny than could be provided by the Advisory Council.

PSA proponents argue that earnings on the investments in the personal savings accounts should be tax-exempt because the initial investment s are made with after-tax income. In contrast, IRAs and 401(k) plans are financed with income on which no tax has been paid, and all the proceeds (both the original investment and the earnings on it) from such ac counts are subject to taxation. This approach in present law has the advantage of making clear that there is a tax loss -- that is, a loss of federal tax revenues -- involved in setting up tax-favored plans. Under the proposed PSA ap proach, on the other hand, contributions to PSAs would be taxed, so there would be no tax loss at first. But the loss would be spread out over time, as the untaxed earnings on investments were paid out.

The approach chosen by PSA proponents makes it possible to present the plan to Congress as not involving any initial loss of tax revenues -- a strategy that has been used over the years in connection with many proposals that do not affect the federal budget adversely at the time of proposed implementation but that have major effects later on. This may help to generate initial support for the PSA plan, but support might well erode as tax losses began to be felt in later budgets and as it became necessary to raise taxes on wages and salaries to protect the tax-exempt status of unearned income. If that were to be the case, the PSA plan, after being promoted as a way to shelter investment returns from taxation, might end up with contributions continuing to be taxed at first, as proposed, but with the returns also taxed when paid out (to the extent they exceed what the worker paid in), just as defined contribution retirement plans are treated now. After all, every other type of invest ment of after-tax dollars requires a tax on the return. In short, this approach to the taxation of individual accounts does not make clear that there is a cost attached to the new tax shelter and that it might prove unsustainable over time.

Whatever the arguable merits of the PSA tax-treatment proposal, however, our view is that the Social Security Advisory Council is not the right place in which to consider such a major change in tax policy. Neither the members of the Council nor its staff were chosen for their ability to contribute to tax policy, nor has the Council devoted much time to consid ering the implications of this proposal, which is offered as part of both the IA and PSA plans (with sponsors of the IA plan taking the position that either this proposed approach or the existing arrangement for taxation of IRAs and 401(k)s would be an acceptable way to tax individual accounts).

It bears noting that although a majority of the Council believes that current beneficiaries should share in the cost of bringing the Social Security system into long-range balance, the PSA plan would do the opposite, reducing present taxes on Social Security benefits by limiting the amount of benefit taxed to 50 per cent instead of 85 percent. This would mean an immediate loss of income to the Medicare HI fund, bringing the date of the exhaustion of the HI fund even closer. In addition, the PSA plan would eliminate the retirement test at the NRA. If this plan were approved, present beneficiaries would be more favorably treated than at present while workers and future beneficia ries would have to bear the full cost of paying for the transition and bringing the program into long-range balance. It is doubtful that such an inequitable approach could survive scrutiny in the Congress.

Increased Risk

Among the PSA plan's many drawbacks in comparison to the present approach to Social Security, perhaps the most disturbing is the increased risk to the individual family. In con siderable part, we would be trading a defined-benefit plan with statutory benefits enforceable in the courts for whatever individuals could secure from the investment of 5 percentage points of the present payroll tax. The only remaining guarantee would be the flat benefit payment of $410 a month for full participation, an amount 35 percent below the $627-per-month poverty level for an elderly person in 1996.

As we have noted in our discussion of the IA plan, it is one thing to have retirement income supplementary to the Social Security system based on income from investment, such as IRAs and 401(k) plans, but quite another to make the basic system depend on the uncer tainties of individual private investment. There are many risks in such a shift, but two stand out:

(1) A high proportion of people required to seek investments for their retirement income may choose what appear to be the safest investments, which are also the lowest-yielding, and are thus un likely ever to achieve the predicted high returns on investment.

(2) Others may end up at the same place but for the opposite reason. Urged on by a greatly enlarged investment advi sory industry, and besieged by salespeople to invest in this or that, some will take very large risks and will lose out -- or pay more in investment expenses than they earn.

Risk for individuals investing on their own is quite a different matter from the shared risk of investing by the Social Security system or indeed by any other pension system. We do not believe that the nation's basic retirement system should require everyone -- the knowledgeable and the inexperienced, the lucky and the unlucky, the rich and the poor -- to bear investment risk as isolated individuals. As we discuss later in describing the MB plan and the option of investing Social Security funds in equities, a central fund broadly indexed to the stock market and investing regularly in good times and bad is at far less risk, and that is the investment approach that we recommend considering for Social Security.

Will PSA Investments Be Held for Retirement?

As with the IA plan, the PSA plan depends for its success on the invested funds being held to retirement, and PSA sponsors say that this will be required. But, as is true of the IA plan, it is much more likely that legislative exceptions would soon be made. If the money is seen as belonging to the individual as it builds up during the worker's career, he or she will feel aggrieved if access to the funds is denied in situations such as paying medical expense, educating children, building a starter house, meeting expenses when a worker is unemployed and unemployment insurance has run out, and so on.

As with IRAs and 401(k)s, exceptions will undoubtedly be sanctioned, and in many cases the individual's PSA funds will have been reduced or exhausted before retirement, with the individual then left to rely on the low-level flat benefit, augmented perhaps by recourse to a poverty program such as Supplemental Security Income (SSI).2/

Unintended Consequences

In assessing any plan, it is important to look at how an initial proposal may change as it moves through the Congress and, perhaps even more importantly, as it makes an impact over time. The PSA plan appears particularly likely to produce consequences that should be scrutinized now rather than later.

Figure 4 compares "money's worth" ratios for the MB plan with and without investment of funds in stocks, the present-law pay-as-you-go program (paygo), and the PSA plan without individual accounts -- that is, with only the flat governmental benefit paid for by the remaining lower tax rate (

Ball graph 4
The flat basic benefit is so clearly a bad deal for all but low-wage workers that, even if it could be enacted, it probably could not be sustained. Under the PSA plan, the same basic benefit is paid to all who have 35 years under the program, but its cost is met from a percentage of earnings, up to the maximums provided in present law. The highest-paid wage earners get from the basic system only the same as the lowest-paid -- including part-time wage earners -- who contribute much less. The only requirement for a full year's credit under the plan is to earn just over $2,500 a year (in 1996, and rising as national wages rise). Those who earn many times as much will pay many times more, but the benefits stay the same.

Many questions will be raised about this whole approach: Why pick this method of financing a benefit unrelated to contributions? Why should contributions stop at an annual maximum earnings level? Why not finance from a portion of the income tax? And other questions will arise. But basically the problem is that participants are already concerned about the return under Social Security for what average and above-average earners pay in, and this new flat-benefit approach is obviously a much worse deal. How could it be expected to last? Above-average earners in particular will be discon tented by what they are paying for the flat below-poverty benefit in the central government plan, and an influential part of the voting public is likely to press for dropping it. Or the higher-paid might press for means-test ing the basic system in order to subsidize only the "truly needy." We might well end up trading Social Security for an enlarged SSI program plus a compulsory savings plan. All semblance of the relationship of benefits to contributions is lost in the PSA plan, and most people will have little stake in maintaining a flat $410-per-month benefit that is paid for by a proportional wage tax.

The Annuity Principle

Under the PSA plan, participants at retirement are not required to annuitize their accu mulated funds. This creates yet another area of risk. Inevitably, many retirees will underestimate the amount of money they need to last through their retirement and will use their funds for other purposes. Others will err on the side of caution. Under privatization, there is no easy way out of this dilemma.

Private annuities can help, but private annuities typically do not offer a fair return for the average person. Based on experience, insur ance underwriters assume that those buying annuities will tend to be those who live longer than average. They are required, therefore, to charge a high premium for individually sold annuities because of adverse selection in addi tion to the usual need of private insurers to charge an extra premium to protect against the general risk that performance will differ from expectation. Currently available annuities also have the disadvantage of lacking Social Security's complete inflation protection, which makes it additionally difficult to plan for adequate payments over the entire retirement period. And defined contribution pensions dependent on private annuities have comparatively high operating costs.*

The PSA plan offers no solution to these drawbacks, instead leaving retirees to seek the best annuities that the market offers. The lack of an assured, inflation-protected annuity is particularly hard on women, since on average they live so much longer than men, and will have to spread the returns from any PSA they have (or inherit from their spouse) over a lon ger period. Indeed, given the proposed tax treatment of PSAs and the absence of any requirement for annuitization, the PSA plan becomes a powerful tax-avoidance tool for high-income people.

Treatment of Spouses

One of the more troubling aspects of the PSA plan is its treatment of non-working spouses. Unlike current law, which provides an additional spouse and survivor benefit pay able to those either married to a worker or divorced from one after 10 or more years of marriage, the PSA plan provides no assured benefit for a spouse or survivor (unless there is a child of the wage-earner in the care of a surviving spouse, which the plan covers as under present law) except for the flat benefit paid by the residual government program. In other words, non-working spouses would be entitled to only $205 per month, with this amount increasing after the worker's death to as much as $461, or up to $615 if the spouse had earned a benefit of his or her own with up to 35 years of coverage (75 percent of their combined government benefit). These amounts are generally far too low to protect survivors against poverty. As a result, it can be assumed that the taxpayers would frequently be called upon to provide additional assistance through the Supplementary Security Income program3/.

*A 1994 study found that the operating cost was $368 per covered worker per year for private pension plans assuming purchase of an annuity in the individual annuity market and $177.60 assuming access to a group annuity. When costs are computed on the same basis for Social Security, the cost per worker is $19. ("Administrative Changes in Pensions in Chile, Malaysia, Zambia and the United States," Policy Research Working Paper 1372, World Bank, 1994, Salvador Valdez-Prieto; referred to in "Proposals to Restructure Social Security," Peter A. Diamond, Journal of Economic Perspectives, Summer 1996, pp. 67-88)

As for the PSA account, a spouse would have no inherent right to any part of the funds in the worker's account, nor evidently even the right to information about the account balance. This creates, at the very least, needless tension between spouses, and, at the worst, new opportunities for financial abandonment. It contrasts sharply with private pension plans covered by the Employee Retirement Income Security Act (ERISA) and with public plans covering feder al employees, both of which require workers to give their spouses a survivor benefit unless the spouse specifically waives it. Since the PSA plan leaves spouses with no guaranteed access to funds, the only solution would be for spous es to negotiate (or litigate) binding agreements -- and in divorce situations any division of the account would be subject to case-by-case ne gotiations and court decisions, an approach that would ultimately be of greatest benefit to divorce attorneys.

The substantial weakening of protection for widows and widowers seems particularly ill-advised. This is the group in American society most vulnerable to poverty. Twenty-two percent of women not currently married are living below the poverty line. The PSA proposal would make their plight even worse.

Treatment of the Disabled

Similarly, the long-term totally disabled are treated badly under this plan. Present disability benefits are reduced across the board as the normal and early retirement ages are raised. The across-the-board reduction would be over 13 percent by 2014. By 2038, the cut is estimated to be 20 percent, and ultimately reaches a cap at 30 percent.

As we noted in our discussion of the IA plan, the disabled are a particularly vulnerable group and particularly dependent on Social Security. Since disabled workers, like others, would not under this plan have access to their private savings accounts until they reached retirement age, the mandated savings are of no help. And if a disabled worker dies before retirement age -- a high probability -- the assets in the savings plan go to the estate.

PSAs and the Investment Industry

The PSA plan would link millions of work ers to the investment industry for the first time. On January 1, 1998, the date when the plan is assumed to go into effect, there will be about 127 million workers under age 55 covered under Social Security. Assuming that 5 percent of their wages is made available for PSAs, they would need to find ways to invest about $150 billion a year (at the outset, with this amount rising annually as the percentage of the workforce eligible for PSAs increases).

Workers with average and below-average earnings -- that is, below about $28,000 a year -- as a group have little investment expe rience; they have been largely left out of 401(k) plans and have generally not made other investments on their own. Evidence suggests that lower-paid workers will, on average, approach investment very conservatively -- pro ducing lower returns than assumed by PSA proponents -- but, in any event, the amounts they have available for their accounts will be small, averaging $1,000 a year or less, and investment expenses will have a considerable impact on how their accounts perform.

Still, the enormous size of the overall pot of money that would be shifted into PSAs means that we can expect an explosion of marketing efforts and of the business of giving advice, much of it dedicated to urging people to switch from what they have to something else.* It is easy to see why some representatives of mutual funds and other segments of the investment industry are strongly supportive of the idea of shifting part of Social Security into compulsory savings accounts. Considering all of the drawbacks, however, it is hard to see how this approach to retirement security will serve the interests of most workers.

* The financial markets are now largely regulated by disclosure requirements, but the effectiveness of this approach depends upon the ability of the investor to understand what is being disclosed. The regulatory challenge posed by the opportunities for fraud and abuse that would attend the creation, virtually overnight, of a whole new class of inexperienced investors should not be underestimated.

Probable Effect on the Present Elderly

The sponsors of the PSA plan maintain that the cost of the transition is an accepted part of their plan and that those already retired would continue to get what they are entitled to under present law, with the implication that they need not concern themselves with the Social Security changes that would affect younger people. Realistically, this is unlikely to be the case. If such a plan were to be seriously considered, those already retired would undoubt edly be asked to bear part of the transition cost. When faced with the need to enact large tax rate increases and authorize long-term bor rowing from the Treasury, the Congress would very likely see a good rationale for asking present beneficiaries to share in the cost. And why not? Since this plan abandons the ratio nale of Social Security for the future, why expect Social Security protection to be fully maintained for present retirees?

After all, earlier proposals along the same line, such as those offered by Senator Kerrey and some of his colleagues -- the plans collo quially known as Kerrey-Danforth and Kerrey-Simpson -- would cut benefits substantially (in particular by reducing the COLA) for those already retired and those nearing retirement. And that is in fact the only way to hold down the double payments required of workers during a long transition period. It will certainly be argued that in many cases the benefits now being paid to retirees are higher than could have been paid for by their own and their employer's contributions, and it will be tempting to limit benefit payments to what can be bought with what has been paid in.

Another possible line of attack on present benefit levels could come from the Concord Coalition and its allies, arguing for reduction or elimination of Social Security payments to those who can get along without them. And it is true, of course, that some people who have earned rights under Social Security from past work and contributions do not need a monthly benefit check to avoid poverty, because they have other income which would keep them above some minimal standard even without Social Security. It is hard to imagine that such people would be exempted from having to help pay for the transition costs of moving from traditional Social Security to PSAs. Thus, if the PSA approach were adopted, means-testing might be perceived as a logical next step.

Indeed, if the nation were to decide to phase out Social Security in favor of another ap proach to retirement income, the cost of the transition would be so high that sooner or later all generations would be expected to share the burden. No group would be exempted except, perhaps, those with very low incomes. Those 55 and over thus have a major stake in the argument over substituting compulsory private savings for Social Security, not only because of what would happen to their children and grandchildren and other members of their families but because of what would happen to their own Social Security protection.

Disruption of Existing Pension Arrangements

As in the case of the IA plan, current private pension arrangements would have to be rede signed. What would happen to these plans if Social Security is abolished in exchange for a low flat benefit plus individual savings accounts? All the pension integration rules would need to be changed as well as many other reg ulatory provisions. Bargaining over pensions could once again become a major source of labor unrest. Would private plans have to make up the difference -- particularly for middle-aged and older workers -- as Social Security is reduced? What about Social Security's unique features -- its inflation protection, dis ability protection, and dependents protection? Unions would certainly want to retain Social Security or its equivalent in defined benefits. How would employers react? Many, perhaps most, might want to take the opportunity to reduce pension costs by shifting to defined-contribution plans. Conflicts would be inevitable.

The PSA Plan and Medicare

Finally, the PSA plan is even more detrimen tal to Medicare than is the IA plan. The PSA plan, like the IA plan, uses up a major increase in the payroll tax to support a compulsory savings plan for cash retirement benefits and also adds to the federal fiscal burden by extensive borrowing. Both of these proposals make it all the more difficult to achieve Medicare balance. Also, in a rather peculiar wrinkle, as the result of rescinding the 1993 tax rate in crease, the PSA plan would eliminate, in 1998, the tax on Social Security benefits now going into the Hospital Insurance (HI) part of Medicare, which has the effect of accelerating (al though not dramatically) the projected date of HI trust fund exhaustion.

Administering the PSA Plan

It would not be an easy task for the government to make sure that workers have the op portunity to invest pretty much as they wish while at the same time making sure that 5 percent of workers' earnings are actually invested and stay invested. Large employers or others with 401(k) plans might have little diffi culty in getting employees to add to or substitute for what is now being done voluntarily, but it is difficult to envision how the compulsory plan would work with smaller employers.

Presumably the employer would withhold 5 percent of wages each payday and transfer it to a broker, bank, or other investment institution designated by the employee. The recipient agency would then presumably report the receipt and its disposition to the Internal Revenue Service, also reporting all changes in in vestment and investment earnings from then on. The complexity of all this activity would obviously create opportunities for abuse, and part of the cost of administering the plan would be the cost of attempting to ferret out collusion. But no allowance for such administrative cost to the government has been included in the plan. In fact, the only administrative cost explicitly included in the plan's specifications and assump tions is the 100 basis points allowed for the cost of financial transactions.

A Cautionary Word

As noted earlier, proponents of the PSA plan are likely to use "money's worth" and "replacement rate" tables and charts to argue that a system of compulsory private individual savings accounts offers workers better protection than a public plan. This is simply not so.

The PSA plan is able to show better "money's worth" figures than the version of the MB plan that invests in stocks for one reason only: because it increases taxes, borrows money, and assumes that individuals will invest 50 percent of their funds in equities, and thus starts out right away with a much larger pool of money to invest in the stock market. As one would expect, when the MB plan is beefed up so that it has the equivalent amount of money for stock investment and improves benefits to match the additional money, its "money's worth" results are roughly the same -- in fact a bit better because it has much lower invest ment-related administrative costs (less than one basis point compared to 100 basis points assumed for the PSA plan). The moral, clearly, is that if you put more in you get more out. By contriving to put more in, the PSA plan is made to look good, at least by this one criteri on. But the outcome has nothing to do with private savings and investment: the results of putting more in are roughly the same whether the plan is public or private.

And it must be stressed again that averages are only averages. Reliance on averages obscures what really happens in a plan in which only the contribution, and not the benefit, is specified. Using average "replacement rates" to indicate retirement income that is in fact de pendent on market returns is misleading in the extreme. In reality, there is no given amount that individuals can count on and thus there is no "replacement rate" that is valid.

This problem is exacerbated for the low-in come investor and for inexperienced investors generally (See our supplemental statement). The amounts available for the low-paid to invest will be small; the high administrative costs of individual investment will cut into or even exceed investment income; and large numbers of the inexperienced will elect the safest, lowest-return investments and for that reason, too, will fail to achieve an "average" return. Their real "replacement rates" are thus likely to fall far short of what the plans advertise.

Another doubtful claim is the extent of the wealth-enhancing effect of the PSA proposal. One economic theory would support such claims, translating every dollar of new taxes as a cut in consumption and an increase in savings as long as the savings are not offset by additional government spending. It seems more likely, however, that as the more successful investors received reports of their growing PSA balances, many would reduce their contributions to IRAs, 401(k) plans and other sav ings vehicles. A life-cycle theory of saving behavior would call for a dollar offset of old savings for every dollar of new saving; the truth probably lies somewhere in between these theoretical constructs.

In addition to whatever new net savings arise from investment of taxes, there may be some effect on wealth accumulation from the higher return on equities as compared to government bonds since individuals would save some of the higher return. There would be no effect on savings, however, if all that happened was a shuffling of portfolios -- equities into individual accounts and government debt sold to pri vate savers rather than to the Social Security trust funds. Under these circumstances, a higher possible return from part of the portfolio held by individuals would be offset by a lower return on another part. It is only a greater total return that would add to savings. Moreover, any benefit coming from this route is also available simply by having the trust funds invest in equities, without having individual ac counts. Indeed the effect is likely to be larger in this case, since a higher return to the trust funds is less likely to induce offsetting consumption than is a higher return on individual accounts.

For all of these reasons, the estimates of increased wealth from PSAs are probably greatly overstated. What is clear beyond doubt, however, is the very large negative effect of the PSA plan on the unified budget of the federal government, as shown in Figure 5. The plan would more than undo all recent progress toward balancing the budget, adding an esti mated $2.4 trillion to the deficit between 1998 and 2017. (In comparison, the MB plan has virtually no effect on the unified budget from 2000 through 2014 because the increase in the Social Security buildup and the return on in vesting a part of that buildup during this period just about offsets the effect of reduced investment in government bonds and the cost of buying equities. Thereafter, the MB plan has a very positive effect, as Figure 5 shows.)*

* The alternative versions of the MB plan (discussed later) produce slightly different results in the 2000-2014 period. Changing the computation period from 35 to 38 years has a slightly negative effect, while a small increase in the contribution rate has essentially no effect in 2000-2005, and an even smaller negative effect than the alternative in 2006-2014 (see Figure 5).

Ball graph 5

The specifics of the PSA plan offer much to criticize, but it is the thinking behind the plan that is most troubling. In abandoning Social Security, we would be shifting from a time-tested national strategy of protecting against individual risk to a scheme involving increased exposure to risk for the aged, those with disabilities, and young one-parent families -- groups that are singularly ill-equipped to deal with risk. Above all, we would be substituting a plan based on an extraordinarily high degree of go-it-alone individualism for our most successful expression of community.

In summary, we oppose the PSA plan because it trades Social Security's defined-benefit approach, with its benefits determined by law, for a defined-contribution approach in which benefits are determined by investment returns; because it trades the relatively manageable set of problems facing Social Security today for a wide range of new problems, not the least of which is the need for workers to pay twice in order to fund present benefits while simultaneously funding their own PSAs; because it raises payroll taxes substantially to fund the redesign of America's basic retirement plan at a time when such tax increases, if warranted at all, should be reserved for Medicare; because it substantially increases the deficit for decades to come; because it sets up a residual govern ment plan which would be unpopular and unlikely to survive; because it exempts present beneficiaries from making any contribution to bringing Social Security into long-range bal ance; because PSA funds are unlikely to be fully retained for retirement; because this approach does not provide for inflation protection or for annuities; because it presents unprecedented tax and administrative issues; and be cause spouses and the disabled are left inadequately protected.

The PSA plan is not a good deal for workers or retirees, now or in the future. It undermines our national commitment to protect the most vulnerable. It undermines the commitment to each other that is at the heart of the Social Security program.

Our Maintain Benefits Proposal

As we have noted, Social Security does not face a crisis, but it does face a long-term defi cit, and that is the first of the challenges that should be addressed in designing a proposal to build on rather than replace the existing system.

Social Security's trustees anticipate that the program, as currently structured and funded, faces an eventual revenue shortfall, beginning about three decades from now. Between now and 2013, the income from Social Security taxes alone will exceed expenses each year. From 2013 until 2020, income from taxes plus interest on accumulated funds will continue to exceed expenses each year. Then, after 2020, if no changes have been made in the meantime, it will become necessary to draw down the accumulated funds to help meet the rising cost of benefits resulting from the baby-boom generation's retirement. In their 1995 report, the Trustees estimated that beginning in the year 2030 -- again, assuming that no corrective steps have been taken -- the reserve funds will have been entirely exhausted, and the program's revenues from payroll taxes and other current income will cover only about three-fourths of the costs of benefits.

Expressed in the conventional way -- that is, as a percent of all payrolls that are subject to Social Security taxes -- the long-term deficit shown in the 1995 report of the Trustees is 2.17 percent. To put this deficit in perspective, if we were to decide to increase payroll taxes dedicated to Social Security by just over 2 percentage points (raising the contribution rate from the present 12.4 percent of earnings to 14.57 percent, with the increase split at just over one percentage point each for employees and employers), and if we were to implement this change now, we could entirely eliminate the long-term deficit and keep the program in balance over the course of the entire 75 years for which Social Security's long-range estimates are made.

This is not to suggest that such a tax in crease is in order, nor that it would be the best way to solve the long-term deficit problem. The point is simply to demonstrate that action taken now on a relatively modest scale can bring the deficit under control without requir ing major benefit cuts or compromising the program's basic principles. And this payroll tax increase -- which, again note, we are not advocating -- would increase deductions from workers' earnings less than the IA plan (from 6.2 percent to 7.29 percent in the example above compared to 7.8 percent for the IA plan) and would be much less burdensome than the PSA plan's combination of a tax increase and large-scale borrowing from the Treasury.

It is important to begin making a major reduction in the 75-year deficit as soon as possi ble, but without major tax increases or benefit cuts and without sacrificing the basic principles and traditional advantages of the program. This can be done without much difficulty by making several relatively minor changes to the Social Security program -- all of them entirely consistent with tradition:

1. Beginning with new hires, extend Social Security coverage to the 3.7 mil lion state and local workers (about one-fourth of all state and local employees) not now covered by the program.*

2. Change the policy governing taxation of Social Security benefits so that the income tax is applied individually to benefits in excess of what the worker paid in (as is now done with other con tributory defined-benefit plans), with revenues from the tax deposited in the Social Security trust fund as under present law.

3. Eventually redirect, to Social Security's Old-Age, Survivors, and Disability Insurance (OASDI) trust funds, the reve nues from taxation of Social Security benefits that are currently going to Medicare's Hospital Insurance (HI) trust fund.

4. Adjust the assumptions underlying Social Security's long-term forecast to reflect the downward correction to the Consumer Price Index (CPI) announced by the Bureau of Labor Statistics in March 1996.

5. Either increase, from 35 to 38 years, the period over which average indexed wages are computed and which in turn forms the basis for determining benefit amounts for future retirees, or, alternatively, increase contribution rates by 0.15 percent of earnings on both employers and employees.**

*Although three members of the Advisory Council who generally support the Maintain Benefits (MB) plan oppose this proposal, it is supported by the other ten members of the Council. See footnote #8 of the main report. ** Advisory Council member Edith U. Fierst, who generally supports the MB plan, opposes both increasing the averaging period and increasing the contribution rate. See her supplementary statement.

These changes alone will have sufficient impact to reduce the long-term deficit from 2.17 percent of payroll to 0.80, postponing the projected date of trust fund exhaustion by two decades. And, with the exception of the last, all of these changes improve the equity of the program and thus are highly desirable in their own right:

  • Making the program universal: Over time, almost all occupational groups have been brought into the Social Security program, and coverage is now nearly universal. But while most full-time state and local employees are already covered, about a fourth are not. This is the last large group of workers not covered. There are several advantages to them and their families to be part of the universal system while retaining supplementary protection under their own plans, including greater freedom in changing jobs; improved protection against inflation; frequently better survivor and disability protection; and reliable coverage for spouses (which is not required under some state and local government pension plans). And in any event it is only fair for all state and local employees to be protected by and contributing to the nation's social insurance system, which benefits all taxpayers by preventing what otherwise could be a large tax burden of welfare and relief. The effect of this change, partly because it increases revenues for many years before benefits must be paid out, is to reduce the long-term 2.17-percent-of-payroll deficit by 0.22 percent of payroll4/.
  • Changing the taxation of benefits: It is fair to ask present beneficiaries to help bring So cial Security into long-term balance, as long as the burden is fairly distributed among those most able to help. The most equitable way to do this is to apply the progressive income tax rules to Social Security benefits in the same way they are applied to other contributory defined-benefit plans, so that Social Security benefits are taxed individually to the extent they exceed what the worker paid in. Also as a matter of fairness, Social Security beneficiaries should not be given a special exemption from income taxes solely because they are beneficia ries, and thus the current exemption for those with annual incomes below $25,000 for single persons and $32,000 for couples should be repealed. Beneficiaries at the lower end of the income scale -- about 30 percent of all beneficiaries -- would be unaffected because their benefits would remain untaxed under general income-tax rules protecting low-income people. With the income from such taxes going into the Social Security trust funds, this change in policy reduces the long-term deficit by 0.31 percent of payroll5/.
  • Reallocating revenues: Purely because of an anomaly that arose in 1993, when the poli cy of taxing Social Security benefits, first adopted in 1983, was broadened by the Congress, some of the revenues from taxation of Social Security benefits are currently being directed to Medicare's Hospital Insurance (HI) fund rather than to Social Security's Old-Age, Survivors and Disability Insurance (OASDI) trust funds6/. In recommending that this be corrected, we do not propose to deny Medicare of needed current income. We propose, rather, to correct this anomaly when Medicare is refinanced, as must necessarily take place at the latest between 2010 and 2020, when the baby-boom generation begins to receive Medicare benefits. This change reduces Social Security's long-term deficit 0.31 percent of payroll. (Again it should be noted that in contrast to this approach, the PSA plan would immediately eliminate the support now going to Medi care from taxation of Social Security benefits, thus accelerating Medicare's approaching trust fund exhaustion.)
  • Adjusting for the effect of correcting the Consumer Price Index: There is unanimous agreement within the Council that it is only fair for any correction by the Bureau of Labor Statistics (BLS) in the way that the Consumer Price Index (CPI) is computed to be reflected in Social Security's Cost of Living Adjustment (COLA) to the extent that the change is not offset by other factors. The Council as a whole also agrees that it would be unwise to change the COLA by legislation. Whether or not a specific change could be justified on grounds of correcting an inaccuracy in the CPI, we believe that such legislative actions would constitute a dangerous precedent that could be used to cut benefits unfairly. We are united in believing that the COLA is one of the most important features of Social Security and that the integrity of the computation must be pro tected. Therefore, we believe that the fair way to proceed is to follow the changes made by BLS and its advisers, inasmuch as BLS is the government agency charged with the technical task of designing and maintaining the CPI. The adjustment announced by BLS in March 1996, when applied to the COLA, reduces Social Security's long-term deficit by 0.31 percent of payroll.

In addition to the changes described above that improve the overall equity of the Social Security program, our proposal allows for ei ther a small reduction in future benefits or a small increase in the contribution rate if needed. We would prefer not to have to rely on either of the options outlined below, but we believe that they are preferable to proposals such as increasing the age of first eligibility for full benefits.

Option 1: reducing benefits by changing the averaging period

Social Security retirement benefits are currently based on computing average indexed wages over 35 years. Lengthening the averag ing period from 35 to 38 years has the effect of reducing benefits by an average of 3 percent for future beneficiaries. It can be argued that since present law provides for gradually increasing the normal retirement age -- that is, the age of first eligibility for full retirement benefits -- from 65 to 67, it would be compatible with this provision to phase in an increase in the period used to calculate average earn ings. (It can also be argued that doing so would create an additional incentive to work until the age of first eligibility for full benefits.)

The effect of this proposal is to reduce benefits more for those with intermittent wage records than for those with histories of consistent employment over the full 38 years.* (The reduction would rarely exceed 6 percent, with no reduction for those with earnings which when indexed to current wages are level or rising over the 38-year period.) This reduc tion in benefits reduces the long-term deficit by 0.28 percent of payroll.

* As a practical matter, this change will affect more women than men, since women are more likely to leave and then re-enter covered employment. Nevertheless, a majority of the Council supports the change because Social Security in terms of returns on contributions will still substantially favor those who have moved in and out of covered employment. Women workers, of course, draw Social Security benefits longer on average than do men. Men and women workers pay the same tax rates but women workers will live 4 years longer in retirement, on average.

Option 2: a contribution rate increase of 0.15 percent

Increasing the contribution rate in the next few years by 0.15 for workers, matched by employers, has approximately the same effect on the long-term deficit, on "money's worth" calculations and on the size of the fund ratio at the end of 75 years as extending the averaging period, discussed above. Replace ment rates under this version of the MB plan are slightly higher because benefits are fully maintained at the rate in present law and the slight increase in the budget deficit under the other version of the MB plan is reduced.

Those who prefer this option make the point that under the MB plan present and future beneficiaries in any event contribute important ly to solving the long-term deficit problem because of the proposed increase in the taxation of benefits, and this change ensures that wage-earners also contribute -- although not to the point of creating significant hardship. The increase in taxes amounts to $15 per year for each $10,000 of wages, or about $1.41 per month for a minimum-wage earner and $4 per month for those earning the maximum wage covered by the program.

Most of these proposed changes have broad support within the Council.* Their combined effect is to considerably improve Social Security's financial situation, as shown in Table 1.

* With the exception of the proposal to eventually transfer income from taxation of benefits from HI to OASDI, these changes have the support of a Council majority.

Table 1

Impact of Specified Changes in Coverage, Calculation of Benefits, Taxation of Benefits, and Adjustment for Correction of the Consumer Price Index
(Shown as a Percent of Payroll)

Long-term (75-year) Deficit   2.17

Impact on Deficit:

Increase taxation of benefits -0.31  
Redirect taxes on benefits from
HI to OASDI (effective 2010-2020)
Extend coverage to new hires in
excluded state and local positions
Adjust for Correction of CPI -0.31  
Increase length of computation period
from 35 to 38 years or increase payroll
tax 0.3 percent
Adjust for interaction among proposals* +0.06  



Remaining deficit


*The impact of a proposed change may increase or decrease somewhat in interaction with others. For ex ample, taxation of Social Security benefits will produce less income if benefits are reduced, and increasing the wage-averaging period reduces benefits an average of 3 percent.

Options to Eliminate the Remaining Long-Term Deficit

The long-term deficit that remains after the above changes are implemented is 0.80 percent of payroll. This may sound substantial, but in fact it repre sents an approximation of "close actuarial balance" as defined over the years. Few people associated with administering Social Security have ever believed that 75-year estimates could be made with such accuracy that the system should be declared out of balance and a legisla tive solution sought simply because of a deviation from exact balance. The tolerance rule has been that the program would be considered within close long-range actuarial balance if estimated income over the 75-year period was with in 5 percent of the estimated cost of the program. At the present time, a 5 percent deviation equals 0.75 percent of payroll -- so the remaining deficit, at 0.80 percent of payroll, is very close to being within the accepted deviation.

We make this point not to imply that nothing more needs to be done to eliminate the long-term deficit but to underscore the fact that if the changes sum marized in Table 1 -- most of which have the support of a majority of this Advisory Council -- are accepted and implemented, there will be ample time to consider further changes.

Part of the remaining deficit could be eliminated by providing for further in creases in the normal retirement age (NRA), the age of first eligibility for full benefits. The NRA, now 65, is scheduled under present law to increase gradually to 67. Eight members of the Advisory Council propose to accelerate the scheduled increase and to provide for additional increases by indexing the NRA to longevity. Five of the six of us who advocate the Maintain Benefits plan strongly oppose this proposed change.* We believe that the nation is ill-prepared for the possible effects of even the scheduled NRA in crease and that further increases cannot be justified until there has been an opportunity to assess the impact of this change7/.

It would also be possible to eliminate the remaining deficit in the traditional way by raising contribution rates and reducing benefits. Raising the rate by one-fourth of one percent on employees and a like amount on employers ($25 for each $10,000 of annual earnings) and cutting benefits by an average of 6 percent (rather than the 3 percent suggested earlier) would be sufficient. But this would only worsen, rather than improve, the benefit/contribution ratio for younger workers, and the payroll tax increase and benefit cut are greater than seems either necessary or desirable.

Another way to eliminate the deficit is to maintain Social Security as a pay-as-you-go program and institute a schedule of pay-as-you-go tax rate increases in the law. Table 2, below, shows the tax increases that would be needed based on present benefit and financing provisions (that is, without any of the proposed changes outlined in Table 1).

* As previously noted, Council member Edith U. Fierst, while supporting the Maintain Benefits plan generally, prefers to see the NRA increased rather than changing the wage-averaging period or increasing the contribution rate.

Table 2
Contribution Rates Required to Maintain Social Security in Long-Term Balance
(Rates shown are for deductions from workers' earnings, to be matched by the employer)

1996-2012 6.20*
2013-2019 6.40
2020-2029 7.05
2030-2049 7.60
2050-2059 7.85
2060-2070 8.20**

*present rate
**and continuing to rise

This approach is feasible but not, in our view, desirable8/. The main disadvantages are the decline in "money's worth" ratios for people born in the future (see Figure 6 below) and the need to implement, beginning in about 2024, either considerably higher contribution rates than projected under present law or substantially lower benefits. Either way, the end result of this approach is to needlessly undermine public support for the program, and thus its effectiveness.

There is still another possibility: a hybrid approach based on partial advance funding for approximately the next 45 years and then pay-as-you-go after that. Under this approach, changes along the lines of those summarized in Table 1 would be adopted in the near future, reducing the long-range deficit to .80 percent of payroll and making it possible to maintain the present contribution rate of 6.2 percent for several decades. When the ratio of the reserve to the next year's outgo dropped to the contingency-fund level of about 100 percent (in 2040, assuming that the proposals in Table 1 are implemented), financing would continue on a pay-as-you-go basis.

Figure 6 shows "money's worth" ratios under two scenarios: present-law pay-as-you-go (paygo) and a plan that incorporates the Maintain Benefits proposals summarized in Table 1 but continuing on a pay-as-you-go basis (paygo).

Ball graph 6
As shown in Table 3 below, the necessary pay-as-you-go rate would begin to exceed the PSA rate in 2040, would equal the deduction from workers' earnings in the IA plan in 2060 (under pay-as-you-go employers also would be paying the increased amount) and in 2070 would reach more than 16 percent combined, continuing to climb thereafter.

Table 3
Contribution Rates Required to Maintain Social Security in Long-Term Balance Under a Hybrid Approach Combining Partial Advance Funding and Pay-As-You-Go
(Rates shown are for deductions from workers' earnings, to be matched by the employer)

1996-2039 6.20*
2040-2059 7.40
2060-2069 7.80
2070- 8.10**
*present rate
**and continuing to rise

The consequences of pay-as-you-go financing, although long postponed under this hybrid approach, are still inescapable: a decline in the "money's worth" ratios for younger workers and a higher than necessary contribution rate in the long run. We believe that it is important, therefore, for Social Security to depart from pay-as-you-go financing in favor of partial advance funding. Developing and maintaining a substantial reserve fund reduces the need for future increases in contribution rates and helps the baby-boom generation to cover more of its retirement costs rather than requiring a higher contribution rate from the smaller generation that follows.

The current Social Security reserve of about $500 billion is equivalent to about 140 percent of current annual expenses. This is appropriate as a contingency reserve for a pay-as-you-go system (normally 100 to 150 percent of the next year's outgo). Under present law, however, and with the changes that we have recommended, this reserve will begin to accumulate to the point where, if it is maintained, earnings on the fund can contribute in a major way to the long-term stability of the program while markedly improving its "money's worth" value for younger workers.

Before moving to a discussion of this point, however, we want to reiterate that if the changes summarized in Table 1 are implemented, there will be ample time to consider proposals to eliminate Social Security's remaining long-term deficit. And there is good reason to proceed cautiously.

First, there is the real possibility that the modification of the Consumer Price Index announced by the Bureau of Labor Statistics in March 1996 has not fully corrected for what is widely perceived as an upward bias in the CPI. On December 4, 1996, an advisory commission to the Senate Finance Committee suggested that the upward bias amounted to as much as 1.1 percentage points a year. Although the Bureau of Labor Statistics is highly unlikely to agree that the correction should be this large, accuracy may call for a greater adjustment than contemplated in BLS's March 1996 announcement. It would be doing a disservice to Social Security participants, to say the least, to enact benefit cuts or tax increases only to discover later that they were not necessary because CPI corrections resulted in reduced Social Security Cost of Living Adjustments.*

Another equally important reason to proceed cautiously, however, is that one of the most promising ways to bring Social Security into long-range balance would represent a departure from present policy and, as such, requires careful consideration by policymakers and the public. Investing a portion of Social Security's trust funds directly in equities tied to an index of the broad market would improve the return on the funds that are presently accumulating and would improve the "money's worth" contribution/benefit ratios for younger workers. Investment in equities is standard practice for other pension systems, both public and private, but it would represent a new concept for Social Security. We believe that the idea has merit -- but we also think it requires careful study, public debate, and perhaps even the convening of an expert commission to explore the pros and cons. We do not advocate near-term enactment of this proposal as we do for the proposals summarized in Table 1. We do, however, consider it worthy of serious consideration, and we propose to describe how this idea might work as a basis for proceeding with further study.**

*Like other members of the Council, we oppose changing the CPI through legislation because of the precedent-setting nature of such an approach, but technical corrections to the CPI by BLS, if not otherwise offset, should certainly be reflected in the Social Security COLA.

** To provide a clear picture of how direct investment in equities would work in comparison with the approach chosen for the IA and PSA plans — investment through a system of compulsory private savings accounts — the charts and other comparisons used in this report assume that the MB plan as implemented will incorporate di

rect investment.

A Public-Private Investment Strategy

When Social Security had only a relatively small contingency reserve, how it was invested did not make much difference. But as the system shifts toward partial advance funding, how those funds are invested becomes important -- and a departure from tradition may be warranted.

In Social Security, the government is the administrator and fund manager of an enormous pension and group insurance plan. It collects dedicated taxes which are the equivalent of the premiums in a private insurance plan and the payments into the defined benefit plans of a private corporation or state retirement system. And it administers benefits which, like the dedicated tax contributions, are spelled out in detail in the law, and which if denied may be appealed to the courts (although the terms of the defined-benefit plan may nevertheless be changed legislatively, as indeed they have been in the past and doubtless will be in the future). There are, of course, important differences between Social Security's defined-benefit plan and the defined-benefit plans managed by the private sector and the states, but the broad characteristics are similar.

Currently it can be argued that the government is not performing its role as fund manager for Social Security as well as it might -- not because of any failure on the part of those managing the system but because Social Security by law is allowed to invest only in the most conservative of investments: long-term, low-yield government bonds. Trustees of private pension systems and managers of state pension systems, who have the authority to invest much more broadly, would surely be castigated if they pursued such an ultra-conservative investment policy, and it can be argued that Social Security should have the same freedom to invest part of its funds in the broad equities market representing practically the entire American economy.*

The increased revenues to Social Security from this change in investment policy can be expected to close the remaining 75-year deficit (0.80 of payroll), thus obviating the need for further increases in payroll taxes or for benefit cuts greater than those included in the changes outlined above in Table 1.

* In connection with this change in policy it would also be desirable to change, over time, the way Social Security is accounted for in the finances of the federal government. Social Security income and outgo should not be included in the budget used in determining the balance of federal obligations and income. The budget is a device for allocating government income among programs during the budget period, while Social Security, with its benefit commitments and fiduciary provisions spread over decades, requires a long-range framework. After all, people are paying today for benefits that may not be payable until 40 or 50 years from now. Social Security, like the retirement systems of the states, should be outside the allocation goals of the short-term budget cycle (while remaining subject to review, of course, as part of the government's long-range planning).

The validity of this argument has been recognized by the Congress, which by law has placed the income and outgo of Social Security "off-budget." Yet in reporting on the budget, the various executive branch and congressional agencies concerned with budgeting have continued to include Social Security in a "unified budget," which is the one used in budget negotiations. This creates confusion about the nature of Social Security and can result in unsatisfactory policy decisions affecting not only Social Security but other programs as well.

The discipline of Social Security is in the long-range balance of income and outgo (as illustrated by the fact that, working within a 75-year time frame, we are currently concerned with an imbalance that does not arise until some 30 to 35 years from now). In contrast, the discipline of the general budget is undermined by continuing to treat balances in Social Security — which by law may be used only to pay Social Security benefits and to meet other Social Security expenses — as though these funds are available for other purposes. With Social Security off-budget in actual practice, we would see more clearly the challenge of finding ways to pay for what the nation needs, and it would also be easier to ensure that the buildup in Social Security assets is saved.

We recognize that Social Security cannot be taken off-budget immediately without disrupting the budget process and creating undue pressure to cut other programs. But as a long-term goal, we believe it is the right one. Indeed, with Social Security's uniquely long-range financing thus clarified, the logic of investing part of the long-range buildup in higher-yielding private equities becomes even more persuasive.

Partial advance funding and investment in equities, in addition to improving the rate of return, greatly improves the "money's worth" ratio for younger workers, as shown in Figure 7, which compares the MB plan including investment in equities with the previously discussed paygo and policy options. As noted, the MB-with-investment plan levels off at about 96 percent for workers born after 1996, while the "money's worth" ratios of the pay-as-you-go version of the MB plan and the present-law pay-as-you-go system continue to decline for each new birth cohort.

Ball graph 7

The MB-with-investment plan would gradually invest up to 40 percent* of the accumulated funds in private equities (reaching the 40-percent range in about 2015), with the balance to be invested in long-term government obligations as before. Forty percent is high enough to close the remaining deficit, assuming a 7-percent real return on investment in equities, while allowing a balanced public/private investment mix9/.

Investments would be indexed to the broad market, and the objective of investment neutrality would be established in law by requiring investment solely for the economic benefit of Social Security participants and not for any other economic, social, or political objective. As in the case of the Federal Retirement Thrift Investment Board, which administers the Thrift Savings Plan for federal employees, an expert investment board would be established. It would have three functions:

*See supplemental statement by Council member Edith U. Fierst

First, it would select an appropriate passive market index (such as the Russell 3000 or the Wilshire 5000) to govern Social Security's equity investments.

Second, it would select, through bidding, several of the leading passive equity index portfolio managers with experience in managing large institutional accounts. In addition to managing the portfolio, the managers would be responsible for maintaining the investment ratio of Social Security's accumulated funds at roughly 40 percent in stocks.

Third, it would monitor portfolio management and establish mechanisms to impartially review the overall operation of the plan, report to the Congress and the public, and periodically consider changes in the index or in portfolio managers as appropriate.

The investment of Social Security funds would have only a limited effect on the market. Although Social Security represents a large part of the government's total operations, it does not represent a large part of the entire $7.5-trillion U.S. economy. Social Security's annual stock transactions should probably involve less than one percent of the value of U.S. equities, and, even at the peak in 2014, should account for considerably less than 5 percent of the value of all stocks (See our supplemental statement).

In addition to pursuing a policy of investment neutrality, it is important to ensure that Social Security's holdings have a neutral effect on stockholder voting on company policy. It may be sufficient to simply prohibit, by law, the voting of any stocks held by Social Security. Alternatively, it might be desirable for the voting of Social Security stocks to automatically be scored in the same proportion as other stockholder votes or the exercise of proxy voting could be delegated to the portfolio managers as is the case with the federal Thrift Savings Plan. Whatever the approach, the point is that it is entirely feasible to ensure the neutrality of Social Security investments in matters of corporate policy.

Some critics of this investment strategy argue that politicians would be tempted to tamper with the index of government investments in order to steer investments toward preferred social objectives. In reality this is unlikely to be a problem. Once the objective of investment neutrality is set, we can be reasonably confident that our competitive political system will furnish the necessary checks and balances to protect this principle. Efforts by one party to undermine neutrality would provide a major point of attack for the other party, with the result that future Congresses would be reluctant to interfere with an established investment arrangement in which nearly every American family would have a stake. (The same principle of political balance has thus far protected Social Security from radical change.)

Social Security would be the largest but not the first federal defined-benefit retirement system to invest a significant proportion of its assets in the stock market. In 1995 the Tennessee Valley Authority had about 40 percent of its $3.8 billion in assets invested in stocks; the Federal Reserve System had about two thirds of its $2.9 billion in assets in stocks; and the systems covering the Army/Air Force exchanges had about 80 percent of their $1.9 billion in assets in stocks.

None of these federal systems has been politically influenced in selecting investments. Nor has the Federal Retirement Thrift Investment Board. In 1986 the Congress authorized the board to make indexed stock investments. As Francis X. Cavanaugh, a former high-ranking Treasury official and the first executive director (1986-94) of the Board, has stated:

"I encountered no significant problems as we selected an index (the S&P 500), obtained competitive bids from large index fund managers, and established a highly efficient stock fund with minimal administrative expenses. By 1996, 1 million federal employees had elected to invest $13 billion in the board's stock index fund. I see no reason why the Social Security trust fund should not have the same stock investment advantage as the Thrift Savings Plan." *

Coupled with the other changes previously described, investment of part of Social Security's accumulated funds in the stock market has the great advantage of leaving the essential principles of the system undisturbed while restoring long-term balance and offering Social Security participants the same stock investment benefits that are enjoyed by participants in other large retirement plans -- state, local, and private. The investment risk is kept manageable and affordable by investing as a group rather than as individuals, and the administrative costs are, of course, very low in comparison to buying stocks and mutual funds retail and managing millions of relatively small individual accounts (not to mention regulating these accounts and reporting them to the Internal Revenue Service). And any lingering concerns about what might go wrong with a centrally managed fund should be balanced against considerations of what might go wrong in a system such as the PSA plan requiring more than 127 million compulsory individual savings accounts.**

It is very important not to confuse the idea of investing in an indexed, passively managed portfolio of equities with proposals to move toward a system of compulsory individual savings accounts. Under the first approach, Social Security remains a defined-benefit plan; under the second it is converted in part to a defined-contribution plan, with benefits determined not by law but by individual experiences with investment.

* *Francis X. Cavanaugh, The Truth about the National Debt, Harvard Business School Press (1996).

* **Although initially only workers under 55 — about 127 million people in 1998 (when the PSA plan is proposed to go into effect) — would have individual accounts, eventually the entire labor force would be involved and at risk.

Investment in stocks by a defined-benefit plan is not, of course, a new idea, nor does it fundamentally alter the nature of a public plan. State retirement systems with defined benefits invest heavily in stocks -- but are still public plans. As Social Security's trust funds build up in anticipation of future costs, it simply makes good sense to consider investing part of the buildup in stocks in order to obtain a higher return than is possible under a policy of investing only in low-yield government obligations. And this approach to restoring long-term balance to the program, unlike payroll tax increases and benefit cuts, greatly improves Social Security's "money's worth" for younger workers.

Increasing the return on Social Security funds can also be justified on the grounds that large-scale investment of Social Security funds in government bonds results in more of a return to the nation's economic well-being than Social Security is now receiving. When Social Security invests a dollar in government bonds, it frees a dollar for investment in the private sector. Thus, a full reflection of Social Security's contribution would be closer to the real return on all private capital (roughly 6 percent) than to the government long-term bond rate. Although investing 40 percent of Social Security accumulations in the stock market nearly doubles the real rate of return on the total of Social Security accumulated funds -- raising it from an estimated 2.3 percent on long-term government bonds after inflation to 4.2 percent after inflation -- it properly goes only part of the way toward what might theoretically be justified.

We reiterate that except for this change in investment policy, Social Security's principles and structure would remain unchanged under this approach. Social Security continues as a defined-benefit plan, with the amount of benefits and the conditions under which they are paid still determined by law rather than by individual investments. In contrast, the plans proposed by other Advisory Council members establish compulsory savings plans, with workers investing their savings individually and, in retirement, getting back whatever their investments may yield. These are fundamentally different approaches.

Risk for individuals investing on their own is quite a different matter from the shared risk of investing through the Social Security system or indeed via any other pension system. We do not believe that the nation's basic retirement system should require everyone to bear investment risk as isolated individuals, as would be the case with a system of compulsory savings plans. A central fund broadly indexed to the stock market and investing regularly in good times and bad is at far less risk. With Social Security's investments tied to the performance of the entire U.S. economy, there would be ups and downs in returns but only very long-range trends would matter. And the assumed rate of return, while important, would be secondary to the fact that benefits would remain defined by law rather than by the relative uncertainty of individual investment decisions.

There is no denying that Social Security funds could be earning a higher return than is possible under present law, and that doing so would help stabilize the program for the long run while making it a better investment for younger workers in terms of "money's worth" criteria. If changes in investment philosophy are to be considered, the approach outlined above would appear to have many advantages over the much more radical changes that the IA and PSA proposals call for.

Correcting for the Tendency to Drift Out of Balance Over Time

There is one more matter to be considered. Even with the elimination of the entire 75-year deficit, as would be accomplished with the MB plan investing in the stock market, Social Security would still be vulnerable to drifting out of balance simply because of circumstances related to the passage of time.

As the outer limit of a 75-year projection shifts further into the future with each new 75-year projection, it captures years in which beneficiaries will be living longer, thus increasing the ratio of beneficiaries to those paying in, so that the cost of the system in each new 75-year period is higher than the last. In the absence of a corrective mechanism, a forecast for the 75-year period from, say, 2010 to 2084 will produce a higher deficit estimate than a forecast for the period from 1995 to 2069. To correct this problem it is necessary to end the traditional 75-year estimating period with a stable ratio of assets to the next year's expenditures. If this ratio is going down, it means that a deficit will shortly appear. But if the ratio is stable, the further passage of time, for many decades at least, would not in itself cause a problem with maintaining long-range balance. To bring about such a stable ratio, we include in our proposal an increase in the contribution rate of 0.8 percent for the employee with a matching rate for the employer, to go into effect about 50 years from now.*

This increase of a combined 1.6 percent of payroll may not be needed when the year 2045 is reached, but we would build it into the law now so that it can be used if needed at that time to maintain a stable ratio between the trust funds and the next year's outgo, or to serve as a fail-safe provision in the event that the inevitable changes in the long-range cost estimates show a higher cost than at present (a lower cost is equally possible under the range of estimates). It should be kept in mind, however, that under our Maintain Benefits plan the elimination of the traditional deficit over a 75-year period is accomplished entirely without this tax increase -- and that when and if this increase is implemented, 50 years hence, it will only then bring the tax rates for our proposal up to the higher levels that will have to be charged all along under the two alternative plans supported by other Council members.

** Council member Edith U. Fierst does not support this proposal; see her supplementary statement.

Understanding the Full Value of Social Security

It is undeniably important for Social Security participants to see themselves as being treated fairly regardless of age or income. Younger workers now face the prospect of making contributions to the system throughout their working lives at rates higher than those required of workers in the past, and higher-paid workers, because of Social Security's redistributional formula, get back relatively less in benefits than lower-paid workers. Thus it is not surprising that increasing attention is being focused on "money's worth" -- the ratio of contributions to benefits that workers can anticipate. But in considering whether to build on or depart from the traditional Social Security program, policymakers and Social Security participants alike should keep in mind the traditional program's unique characteristics and contributions to society as a whole rather than comparing proposals primarily in terms of how they score on individualized tests of "money's worth."

To begin with, Social Security keeps some 15 million people above the poverty line and millions more from near-poverty. Thirty years ago, the poverty rate among the elderly was 28 percent, more than twice as high as the rate for the population as a whole. Social Security has been the key factor in bringing the poverty rate for the elderly down to about 12 percent today, on a par with rates for other adult age groups. The fact is that without Social Security, very large numbers of the elderly would outlive any assets they might have at retirement and would end their lives trying to subsist on incomes below the federal government's rock-bottom definition of poverty.

But Social Security is much more than an anti-poverty program. Because Social Security benefits are not means-tested, the program encourages saving. Participants are free to add other income to Social Security and thus to work toward the goal of providing a level of living in retirement that is not too far below what they achieved while working. This is a major objective of Social Security as the basic tier in a three-tier system of retirement income that includes pensions and savings.

Social Security is just about universal. Among Americans who are 65 and over, 92 percent are receiving benefits, with another 3 percent who are still working eligible to begin receiving benefits when they retire. Social Security is portable, following covered workers from job to job, and the protection before retirement rises as wages rise. And benefits, once they begin, are inflation-proof -- a guarantee that no private pension plan offers.

Social Security is an earned right growing out of past work and contributions, self-financed entirely by dedicated deductions from workers' earnings matched by the employer, by taxes on Social Security benefits and by earnings on the trust funds. Administrative expenses, also financed entirely from the above sources, take less than one cent out of every dollar paid in. The fact that Social Security is contributory and self-financed reinforces the concept of an earned right, and the fact that the entire cost of benefits and administrative expenses is met without support from general government revenues reinforces the importance of planning the program for the long run. This has been accomplished. As noted previously, Social Security since 1937 has collected approximately $5 trillion in income and paid out $4.5 trillion in benefits and other expenses, leaving $500 billion in reserve. The program is not adding a penny to the federal deficit.

Social Security is more than a method of saving for the future. Just as importantly, it is also a mechanism for sharing among all working families the burden of caring for the current elderly and those with disabilities. Because of Social Security and Medicare, no one family has to bear alone what could be the huge cost of caring for parents who are sicker than average or who outlive their savings.

Social Security functions as our national family protection program. Because Social Security spreads the risk and because of its nearly universal coverage, relatively few older retired people have to move in with their children, and millions of married couples are spared the pain of having to decide how much of the family's income should go to supporting elderly parents and grandparents versus paying for their own and their children's living expenses, health care, and education.

Social Security also offers current family protection by providing ongoing insurance against loss of family income due to death or disability. Among the nearly 44 million Americans receiving monthly benefits, 3 million are children under 18, mostly children of deceased workers, and 5 million adults are being paid benefits because of disabilities.

Social Security provides more than $12 trillion in life insurance protection, an amount that exceeds the combined face value of all private life insurance policies currently in force. No one is turned down for this insurance because of health problems, and protection for young workers is very substantial. In a typical example -- a 27-year-old couple, both working at average wages, with two small children -- survivors' protection is worth $307,000. Disability protection for the same family amounts to $207,000.

Comparisons of the value of Social Security and private protection are never exact, because there is no close approximation of Social Security's protection available in the private insurance market, but Social Security has some inherent advantages -- including its universality, its ability to follow the worker from job to job with a defined benefit, its inflation protection, and its low administrative cost (less than 1 percent of income, compared to more than 11 percent, on average, for private life insurance) -- that make it a uniquely good buy from generation to generation10/ . These values, easy to ignore when focusing narrowly on individualized "money's worth" investment comparisons, need always to be kept in mind.

Social Security is a blend of reward for individual effort and, at the same time, perhaps our strongest expression of community solidarity. Social Security is based on the premise that we're all in this together, with everyone sharing responsibility not only for contributing to their own and their family's security but also to the security of everyone else, present and future.

There is nothing sentimental about this approach; it is neither liberal nor conservative; it simply makes sense. Unable to know in advance who will succeed and who will struggle unsuccessfully, who will suffer early death or disability and who will live long into retirement, in good health or ill, we pool our resources and are thereby able to protect against the average risk, at manageable cost to each of us. Social Security's redistributive benefits formula, feasible only in a system in which nearly everyone participates, not only helps to protect us all against impoverishment but, because it is part of a universal system, does so at much lower administrative cost than private insurance and without the stigma of a welfare program.

All in all, it would be hard to imagine a more broadly valuable or cost-effective investment than Social Security. Our Maintain Benefits proposal improves the "money's worth" of Social Security for individual workers. But it is still the full value of Social Security -- across generations -- that should be kept uppermost in mind when comparing the pros and cons of proposals that would fundamentally alter the traditional Social Security program. Our proposal, alone among those offered by members of the Advisory Council, fully preserves the principles that have made Social Security so valuable for so many generations of Americans 11/.


Social Security fulfills what Lincoln described as the legitimate objective of government: "to do for a community of people whatever they need to have done but cannot do at all or cannot do so well for themselves in their separate and individual capacities." It is extremely important that Social Security, as the foundation for all retirement planning, continue in the form of a defined-benefit plan -- that is, a plan promising specified benefits that are not at risk of being affected by investment decisions.

With Social Security as a base, those who can afford to add other retirement income to Social Security will continue to be free to do so, with assistance from the tax code and without being penalized by a means test. And with adequate Social Security protection in place, supplementary pension plans and private savings vehicles can seek investment returns that, while comparatively risky, may pay off at relatively high rates.

This argues for retaining our traditional three-tier retirement system with Social Security as the foundation -- a foundation that is not threatened by the failure of a business or the decline of an industry, and with benefits defined by law. Basic protection that one can count on is particularly important in a dynamic, risk taking economy such as ours, in which long-established businesses, even whole industries, may fade even as new ones are springing up. More than most, our economy rewards rapid adaptation to changing conditions. That is one reason why it functions well at the aggregate level -- but the more dynamic the economy, the greater the need to protect against major economic hazards. In short, we need the fundamental security that Social Security uniquely supplies regardless of downsizing, mergers, bankruptcies, and the volatility of the job market.

Over time, of course, Social Security will change and adapt, just as it has in the past and even as we now recommend. But the system that has met every challenge for 60 years is soundly bottomed on principles that have broad, enduring appeal for Americans. Whenever Social Security's long-term stability has been threatened by circumstances requiring a legislative response, strong public support for the program has prompted political leaders to seek bipartisan solutions that have built on its inherent strengths.

That is the approach we recommend now -- to build on rather than replace the program that works so well for so many.


1. The report of the Advisory Council on Social Security is based on the findings of the 1995 Trustees Report. Although the trustees' 1996 report was issued while the Council was still in session, the Council opted to continue using the 1995 report. The differences between the findings of the 1995 and 1996 reports are minor in consequence, but conforming the Council's report throughout would have required extensive additional work and time, for little purpose.

2. That is, if SSI is redesigned to keep up with the rising level of living in the rest of the community. In its present form, SSI benefit levels are increased only by prices while the flat benefit goes up by wages, so that if nothing is changed the flat benefit would ultimately rule out SSI as a supplement for full-time (35 year) workers. However, it seems to us that sooner or later SSI will have to be changed to recognize not only price increases as at present but changes in the standard of living that affect all Americans, including the poor.

3. See the previous endnote.

4. Coverage of this group can be phased in by first covering new hires. This approach, which was used when Social Security coverage was extended to federal employees, gives the employing entity time to adapt to any increases in cost. If provision is made for implementing this change a year or two after enactment, there will also be time to work out the integration of Social Security with state and local retirement systems, minimizing new costs to employers while maintaining the desired level of protection for employees.

The present arrangement is inequitable. Although not contributing to Social Security in their present jobs, these state and local employees derive benefits as general taxpayers from Social Security's effectiveness in reducing reliance on means-tested public assistance programs such as Supplemental Security Income (SSI). Moreover, most of these workers will receive benefits under Social Security because of covered employment that they will have had at some point during their working careers -- and, since not all of their employment would be included for contribution and benefit purposes, under the weighted benefit formula their benefits will be high in relation to contributions paid (notwithstanding the fact that the law now contains an offset provision that meets part of this problem). Thus this change improves the overall equity of the Social Security program while simultaneously reducing the long-term deficit.

5. Social Security benefits are partly taxable today for those whose total incomes exceed specific thresholds, but we propose several changes in this tax treatment that would bring Social Security into line with other contributory defined-benefit plans by taxing benefits individually to the extent they exceed what the worker paid in -- with low-income beneficiaries protected by the general income tax provisions as noted.

6. Directing the additional revenues from broadened taxation of Social Security benefits to the OASDI trust funds -- the logical step -- would have required, under Senate rules, a super-majority (60 votes), while only a simple 51-vote majority was needed to direct the revenues to the HI fund. Congressional leaders had sufficient support to pass the tax measure but lacked 60 votes. Not wanting the additional income from taxation of Social Security benefits to go into general tax revenues, they settled for directing it to the HI fund. There is no other rationale for this arrangement and no logical reason to perpetuate it.

7. Today the NRA is 65. Present law provides for a gradual increase (affecting all those who reach age 62 in or after the year 2000) to age 66 (for those who reach age 62 in 2005) and then, after ten years at that level, it gradually increases to 67 (for those who reach age 62 in 2022). A majority of the Advisory Council would accelerate this timetable so that the NRA reaches 67 in 2011, and would provide for automatically increasing the NRA still further thereafter by indexing it to longevity.

The effect of these changes is, of course, to reduce benefits as compared to present law. For the person who would have retired at 65 and is successful at working until 67, the earnings replacement rate will be about the same as it would be under current law, but benefits will be received for two fewer years than if the law had not been modified. On the other hand, the person who still retires at age 65 after the NRA has increased will receive benefits for as long as before, but the benefit is reduced by more than 13 percent.

Of even more concern is that this provision will reduce benefit levels substantially for those who are forced to stop working before 65. The law provides today for a 20-percent reduction for those who retire at 62, but as the NRA moves to 67, the reduction will gradually change to 30 percent. And as the NRA moves above 67 when indexed to longevity, as recommended by the Council majority, there is a real dilemma about what to do about paying benefits to those who retire before reaching the NRA.

On the one hand, if the first age of eligibility for reduced benefits is continued at age 62 and the benefit amount is actuarially reduced, the benefit amount becomes more and more inadequate, resulting in an impoverished retirement (and an increase in the number of people who have to apply for SSI). On the other hand, if the first age of eligibility for reduced benefits is raised in tandem with the NRA, people who have to retire at 62 will get no benefits at all until later.
There is no satisfactory solution to this dilemma. And there is a special problem for widow's benefits, which are now payable as early as age 60 with a reduction in benefits of up to 28.5 percent for benefits that start before the NRA. It would be difficult not to increase this reduction or start benefits at a later age if the early retirement benefit for workers continues to be reduced. (It should be noted that the proposals to raise the NRA offered by various Advisory Council members choose different responses to this issue. The IA plan does not change the 28.5 percent reduction of benefits payable at age 60; the PSA plan keeps the reduction at 28.5 percent but raises the age at which it takes effect as the earliest age of eligibility is raised.) But widows, even at present benefit levels, are the most needy group of Social Security beneficiaries. In short, any increase in the NRA triggers undesirable consequences.

The rationale for increasing the NRA is that as life expectancy increases, so should the length of the work life. In 1935, when Social Security was enacted and the NRA was set at 65, average life expectancy at age 65 was about 12 years for men and 13 for women. It has since increased to 15 for men and 19 for women. Obviously, if in fact people were also working longer on average, increases in the NRA would make sense. But the retirement-age trend has been the other way. In recent decades, the average age of retirement has been declining rather than increasing.

It is risky to assume that just because people are living longer, they can work longer and will want or have the opportunity to do so. Nor is it at all clear that private pensions are likely to make any concessions to the argument that people should work longer. Workers themselves -- and certainly employers -- seem committed to retaining the option of early retirement, and among those for whom the decision to retire is voluntary, the age-65 NRA requirement of Social Security seems to be much less of a determinant than the availability of private pensions at earlier ages.
Consequently, we have grave doubts about the advisability of the increase in the NRA that is already scheduled to take place under present law. Its impact falls disproportionately on lower-paid and blue-collar workers, who are the least likely to be able to extend their working lives. Many cannot work until age 65 today. If they stop work before age 67 after the NRA rises, however, they will receive considerably reduced benefits. As noted, a worker who retires at 62 today gets benefits 20 percent lower than at full retirement age, and as the NRA rises to 67, under present law the reduction for retirement at age 62 will rise to 30 percent.

This constitutes a significant penalty for retiring before reaching the NRA. It is one thing to be penalized for a decision made voluntarily. But most workers cannot voluntarily choose whether or not to extend their working lives. Early retirement may not, in fact, be voluntary for those who are in physically demanding jobs, or who are downsized, or who have limited job market skills or are partially disabled. Older women appear to be particularly disadvantaged in the labor market. In short, large numbers of workers have, for whatever reason, little choice about when they will retire. If they cannot adapt to an increase in the retirement age, however, they will be penalized by receiving substantially reduced Social Security benefits.

We are concerned about considering further changes in the NRA for Social Security before the nation has had any experience with the change already scheduled in law. In the 13 years since the provision for raising the NRA was adopted, the employability of older workers has decidedly not increased. If anything, they are less in demand now than in the past. In short, the nation seems unprepared to cope with or cushion the impact of the change already scheduled.
Moreover, we believe that a high proportion of the workers who will be affected by the change in present law are unaware of it. As they become more aware that an increase in the retirement age has been scheduled, concerns will be raised. Four years from now, when the increase begins to take effect (with the NRA initially rising in two-month increments, affecting workers who reach age 62 in or after the year 2000), it will almost certainly come as a shock to many workers and their families. In addition to those facing real hardship because of the change, the average near-retiree may well feel aggrieved upon discovering that he or she is to be treated differently from those who came before. Although it may be possible to retain the present-law provision in the face of increasing dissatisfaction (and we are not now proposing a rollback), it seems to us that it will be very difficult to retain a provision that automatically increases the retirement age in the future.

It is easy to argue today for automatically increasing the NRA tomorrow by indexing it to longevity, because this arrangement will have no immediate impact but can be used to reduce the long-term Social Security deficit by 0.4 of payroll. Accelerating the present-law provision for going from 65 to 67 cuts another 0.1 percent. Thus these two postponement-of-retirement measures can close about a fourth of the long-term deficit -- on paper. But bringing the program into long-range balance by relying on long-deferred benefit cuts that may ultimately prove unacceptable to the public strikes us as an exercise of doubtful value.
Our position is that before considering an extension of present law, we ought to see what happens as the age-67 provision takes effect. Perhaps workers in large numbers will find that they can indeed work longer; perhaps more jobs will open up for older workers; perhaps private pensions will be modified; and perhaps the age-67 provision can be implemented with solid public support. But we should wait and see. Further retirement-age changes, if warranted, can certainly be made later.

8. Although private insurance systems and private pension systems need to be on an advance-funded basis because of the risk that any particular insurance company or business may cease to exist, the continued existence of the United States government may be assumed, and therefore pay-as-you-go financing relying on the ongoing taxing power of the government is always a theoretical possibility.

9. Forty percent is not intended to be taken as some sort of magic point above which the risks of investing Social Security funds in private equities are thought to become unacceptable. An analysis conducted for the Advisory Council by HRA and Associates (and appearing in Presentations to the Council, Vol. II, pp. 341-345) using stochastic modeling showed almost no increase in risk as higher proportions of the trust funds were invested in equities. Regardless of the percentage of funds chosen for private investment, with the approach that we recommend -- investment year in and year out and indexed to virtually all of U.S. industry -- the investment risks are kept low, especially in comparison to the risks of individual investment in consumer-chosen financial instruments and to the individual investor's risk of having to buy or sell at unfavorable times. In short, this approach protects Social Security funds against all but the risk of another full-blown Depression -- that is, a very long-term decline affecting the entire economy. And in a defined-benefit plan, even if actual income were less than anticipated, policymakers would have to make new choices between additional financing and benefit cuts. In a defined-contribution plan, lower-than-expected investment income is accepted as one of the risks of the plan.

10. In their 1995 report the Trustees reported that Social Security's administrative costs were 8/10ths of 1 percent of income. In contrast, the administrative costs of the U.S. life insurance industry were 11.2 percent of income in 1995 (American Council of Life Insurance, 1996). As pointed out by Annika E. Sunden and Olivia S. Mitchell (Pension Research Council Working Paper Series 94-7, Wharton School, 1994), Social Security's administrative costs may be slightly understated, since they may not include a sufficient allocation of Internal Revenue Service general overhead costs related to collecting Social Security taxes. On the other hand, Social Security performs work for the IRS (in the reconciliation of W-2 data) that is not explicitly recognized.

11. Our Maintain Benefits plan follows the same guiding principles that have made Social Security by far the nation's most successful social program.

In the midst of the Great Depression of the 1930s, the founders of today's Social Security system in the United States took the bold step of establishing a new institution which they expected to be slow-growing but permanent. They wanted to make a decent retirement attainable for millions of Americans who would otherwise become wholly dependent on their families or on public assistance when they grew too old to work or could no longer find employment. The founders wanted to protect workers' dependents by making the death of a breadwinner at least financially manageable. They wanted to put an end to the poorhouse by distributing program income so as to provide at least a minimally adequate benefit for everyone regularly covered by the program. And, foreseeing the inevitability of change -- including the eventual need to insure against other major risks such as disability and the cost of medical care -- they sought to design an institution based on sustainable principles.

Accordingly, they took the long view. They gave major emphasis to estimating program income and expenses over a much longer period than was customarily done in other countries, and this is still true today. The time frame of 75 years that is now used for Social Security is much longer than that used in other contexts, from foreign social insurance programs to federal budgeting. The point, then and now, was not to try to pretend that anyone could really know precisely what would be happening in 75, 50, or even 25 years; the point was that the planners of Social Security wanted always to be looking far enough ahead to anticipate necessary improvements and to make needed changes while protecting and preserving the integrity of the program.

That approach has served well. The legislation adopted in 1935 and 1939 created the basic framework of Social Security, and all major legislation since then can be seen as filling in that framework: extending coverage to more and more workers, improving the level of protection, adding protection against loss of income from long-term and total disability, providing protection for the elderly and disabled against the increasingly unmanageable cost of medical care, protecting against the erosion of income by inflation, and abolishing all statutory differences in the treatment of men and women.

These and many other lesser accomplishments and adjustments have taken place within a framework consisting of nine basic principles. Social Security is universal; an earned right; wage-related; contributory and self-financed; redistributive; not means-tested; wage-indexed; inflation-protected; and compulsory. As with any framework, the stability of the entire structure depends on the contribution made by each part, so it is useful to review these principles and how they work together.

(1) Universal: Social Security coverage has been gradually extended over the years to the point where 96 out of 100 jobs in paid employment are now covered, with about 142 million working Americans making contributions in 1996. The goal of complete universality is now within reach -- and, as previously discussed, the last step should be taken to reach that goal.

(2) Earned right: Social Security is more than a statutory right; it is an earned right, with eligibility for benefits and the benefit rate based on an individual's past earnings. This principle sharply distinguishes Social Security from welfare and links it, appropriately, to other earned rights such as wages, fringe benefits, and private pensions.

(3) Wage-related: Social Security benefits are related to earnings, thereby reinforcing the concept of benefits as an earned right and recognizing that there is a relationship between one's standard of living while working and the benefit level needed to achieve income security in retirement. Under Social Security, higher-paid earners get higher benefits -- while, at the same time, the lower-paid get more for what they pay in.

(4) Contributory and self-financed: The fact that workers pay earmarked contributions from their wages into the system also reinforces the concept of an earned right and gives contributors a moral claim on future benefits above and beyond statutory obligations. And, unlike many foreign plans, Social Security is entirely financed by dedicated taxes, principally those deducted from workers' earnings matched by employers, with the self-employed paying comparable amounts. The entire cost of benefits plus administrative expenses (which are only 8/10ths of one percent of income) is met without support from general government revenues.

This self-financing system has several advantages. It helps protect the program against having to compete against other programs in the annual general federal budget -- which is appropriate, because this is a uniquely long-term program. It imposes fiscal discipline, because the total earmarked income for Social Security must be sufficient to cover the entire cost of the program. And it guards against excessive liberalization: contributors oppose major benefit cuts because they have a right to benefits and are paying for them, but they also oppose excessive increases in benefits because they understand that every increase must be paid for by increased contributions. Thus a semi-automatic balance is achieved between wanting more protection versus not wanting to pay more for it.

(5) Redistributive: One of Social Security's most important objectives is to pay at least a minimally adequate benefit to workers who are regularly covered and contributing, regardless of how low-paid they may be. This is accomplished through a redistributional formula that pays comparatively higher benefits to low-paid than to high-paid earners. The formula and the idea behind it make good sense. If the system paid back to low-wage workers only the benefit that they could reasonably be expected to pay for on their own, millions of retirees would end up on welfare even though they had been paying into Social Security throughout their working lives. This would make the years of contributing to Social Security worse than pointless, since the earnings deductions would have reduced their income throughout their working years without providing in retirement any income above what would be available from a welfare payment. The redistributional formula solves this dilemma, and, in so doing, reduces the burden of welfare for everyone: those who would otherwise end up on it and the rest who must pay for it.

(6) Not means-tested: In contrast to welfare, eligibility for Social Security does not depend on the beneficiary's current income and assets; nor does the amount of the benefit. This is a crucial principle. It is the absence of a means test that makes it possible for people to add to their savings and to establish private pension plans, secure in the knowledge that they will not be penalized by having their Social Security benefits cut back as a result of having achieved additional retirement income security. The absence of a means test allows Social Security to provide a stable role in anchoring a multi-tier retirement system in which private pensions and personal savings can be built on top of Social Security's basic, defined protection.

(7) Wage-indexed: Social Security is portable, following the worker from job to job, and the protection before retirement increases as wages rise. Benefits at the time of initial receipt are brought up to date with recent wages, reflecting improvements in productivity and thus in the general standard of living. Without this principle, Social Security would soon provide benefits that did not reflect previously attained levels of living.

(8) Inflation-protected: Once they begin, Social Security benefits are protected against inflation by periodic Cost of Living Adjustments (COLAs) linked to the Consumer Price Index. Inflation protection is one of Social Security's greatest strengths, and one that distinguishes it from other (except federal) retirement plans: no private pension plan provides guaranteed protection against inflation, and inflation protection under state and local plans, where it exists at all, is capped. Without COLAs, the real value of Social Security benefits would steadily erode over time, as is the case with unadjusted private pension benefits. Although a provision for automatic adjustment was not included in the original legislation, the importance of protecting benefits against inflation was well understood, and over the years the system was financed to allow for periodic legislation to bring benefits up to date.

(9) Compulsory: Social Security compels all of us to contribute to our own future security. A voluntary system simply wouldn't work. Some of us would save scrupulously, some would save sporadically, and some would postpone the day of reckoning forever, leaving the community as a whole to pay through a much less desirable and less efficient safety-net system. With a compulsory program, problems of adverse selection -- that is, individuals deciding when and to what extent they want to participate, depending on whether their individual circumstances seem favorable -- are avoided (as are the problems of obtaining adequate funding for a large safety-net program serving a constituency with limited political influence).

In the midst of the Depression it took courage to enact a system based on these principles. Although the Depression was a time of enormous and immediate needs, Social Security was designed to be a slow-growing tree, one that could not provide much shelter in the near term and that would outlive all those involved in planting it. The point, however, was that, once grown, it would be strong enough to weather bad times as well as good.

A contributory retirement system takes a long time to develop, since by definition those who are already retired are not eligible for benefits. Fifteen years after the program was set up, only 16 percent of the elderly were receiving benefits, and it was not until the 1950s that politicians began to see much advantage in championing improvements in Social Security. And it was only in the 1960s, three decades after enactment, that Social Security began having a major impact, paying benefits that were high enough and universal enough to significantly reduce poverty among the elderly, the disabled, and the survivors of beneficiaries. In fact, it was only after the amendments of 1972 substantially increased benefits and introduced automatic inflation protection that Social Security fully assumed the role planned for it as the all-important base of a multi-tier retirement system in which private pensions and individual savings are added to Social Security's defined protection.

Today Social Security is the only organized retirement system -- the only assured source of retirement income -- for at least half of the total workforce. Social Security has become the most popular and successful social program in America's history because it works exactly as its founders intended it to work: as America's family protection plan.