Social Security Reform

Henry J. Aaron

Mr. Chairman:

Thank you for inviting me to testify before you on reform of the Social Security system. In my full testimony, which is attached, I make the following points.

What is the Current Debate About?


The current debate is not about whether Americans should have access to individual retirement accounts. Nearly everyone has such access already.

The retirement income system of the United States consists of five main elements: Social Security, which provides more than half of the income for roughly two-thirds of American retirees; employment-based pensions; voluntary tax sheltered saving, such as Individual Retirement Accounts and Keogh plans; other private saving; and welfare for the elderly through Supplemental Security Income. Couples with adjusted gross incomes under $40,000 a year (individuals with incomes under $30,000 a year) may contribute to individual retirement accounts. Roughly half of all workers are covered by work-based pension systems and a majority of those plans consist in whole or in part of individual defined-contribution accounts. The self-employed or people whose jobs do not supply a pension may set up Keogh plans. All of these accounts enjoy the benefits of tax deferral until benefits are paid. And, of course, people are free to save out of after-tax income.

If access to tax-sheltered individual accounts is now almost universal—and any remaining gaps could be filled in without reference to Social Security in ways I shall describe—the debate about reform of Social Security must be about something else. And it is. Social Security is the instrument that has provided assured basic income to retirees, the disabled, and survivors of the death of an earner for nearly sixty years. The current debate concerns how best to provide such assured basic income in the future at least cost.

The key words in the last two sentences are “assured,” “basic,” and “least cost.” I shall show that a defined-benefit pension system will yield higher returns and deliver them with greater reliability than would defined-contribution individual accounts. [A separate question concerns whether it is desirable to build up pension reserves. I believe that such reserve accumulation is desirable and that it can be done either through social insurance reserves or individual accounts.]

Why Returns Will Be Higher Under Social Security than Under Private Accounts

Advocates of private accounts have claimed that the returns to pensioners will be higher under a privatized system than under Social Security. This claim is the exact opposite of the truth for reasons that have been developed cogently in scholarly research and can be expressed simply in nonacademic terms. Furthermore, private accounts would expose individual workers to risks that are now broadly shared among workers and across generations.

The logic is straightforward. Payments to support Social Security (or Social Security and private accounts) go for two purposes:

  • to pay Social Security benefits for current or future beneficiaries that exceeds reserves (the so-called unfunded liability); and

  • to build reserves to help support benefits of future retirees.

Whether to curtail Social Security benefits for the current elderly and those soon to retire is an important issue, but it is entirely independent of the issue of privatization. Whatever decision Congress makes on this divisive issue, current workers will have to pay the same amount for those benefits, whether we retain Social Security in its current form or gradually shift to some alternative. Current workers who support benefits for the current beneficiaries derive no personal financial gain whatsoever from paying those taxes. This situation is the inescapable consequence of decisions made over the past half century to pay larger benefits to retirees than the payroll taxes paid on their behalf could justify. Paying out those taxes as current benefits rather retaining them as reserves produced what policy analysts call the “unfunded liability,” the obligation to pay benefits to current retirees and to currently active workers in excess of accumulated reserves. We can debate whether or not it was wise to pay those benefits and build up this unfunded liability, but we cannot avoid paying it except by reneging on commitments to current beneficiaries and older workers. On this score, there is no difference at all between the return to workers under a gradual transition to privatized accounts of any kind and retention of Social Security.

Any difference between the returns workers receive under a privatized system and that under Social Security must come from the taxes paid above those necessary to support current benefits and that go to accumulate reserves.

The average returns from similarly invested assets would be lower under private accounts than they would be under Social Security because private accounts would be more costly to administer and these extra costs would reduce the amounts available to support pensions.

Furthermore, the pension returns from (defined-contribution) private accounts would be subject to considerably more risk than would be the pensions provided by (defined-benefit) Social Security.

Administrative Costs and Average Returns

All individual accounts would generate administrative costs higher than those under Social Security.

  • costs of funds collection would increase if employers send funds to individual accounts

  • additional expenditures for public education would be necessary

  • costs of keeping records, correcting errors, and providing information would increase; furthermore these costs do not vary with the size of accounts and will therefore take a bigger bite out of the accounts of low earners than of high earners

  • annuitization through private insurance companies would generate additional administrative charges.

The subject of administrative cost is often treated as a dull matter best left to dull people for routine consideration after creative analysts and legislators have laid out the best pension design. For reasons I shall try briefly to indicate, this view is dead wrong. Administrative considerations are central to the design of the nation’s pension system. In particular, some plans to “privatize” Social Security that have received widespread support are simply unworkable and do not deserve serious consideration. Others, although more costly to administer than Social Security, could be implemented with sufficient effort and expenditure. Because so many privatization proposals exist, I shall consider two prototypical approaches: one modeled on Individual Retirement Accounts (IRA plans), which would be privately managed, and one on the federal governments Thrift Savings Plan (TSP plans), which would be publicly managed.

IRA Plans. Under IRA plans, employers would allocate a portion of payroll taxes collected on behalf of workers—usually 2-5 percent of covered earnings—to designated investment fiduciaries—mutual funds, insurance companies, brokerage houses, banks, or other approved organizations—selected by employees. Employees would have the right to change fiduciaries without charge at some designated frequency. Funds would be invested in approved assets of the employee’s choice. Upon reaching retirement, workers would be able to withdraw funds on a predetermined schedule or convert the balance into an annuity. Any funds remaining at the worker’s death could be transferred to heirs. This approach to privatization is embraced by five members of the last Advisory Council on Social Security, the Committee for Economic Development, by several members of Congress, the Heritage Foundation, the Cato Institute, and by Martin Feldstein.

In appraising the administrability of such a system, several key facts should be kept in mind.

  • Each year earnings are credited to 140 million workers; half earn less than $18,000 a year or $346 a week. A 2-5 percent contribution to an investment fund would run between $30 and $75 a month for the median earner. For the 30 million who earned less than $5,000, the monthly contributions would run from $8.33 to $20.83.

  • Four million employers have 10 or fewer employees.

  • 5.4 million employers file their W-2 forms on paper, which means they are not using computerized systems; about 8 percent of these forms currently fail consistency checks, necessitating error review. There is ample time under Social Security to correct errors, because benefits cannot be claimed until age 62. These errors would require prompt and more costly correction under individual accounts.

Those privatization plans that require employers to send funds to fiduciaries designated by employees would impose costly burdens on small businesses few of whom now provide private pensions or other forms of tax-sheltered saving to their employees. Furthermore, the amounts that would be deposited in individual accounts would be much smaller for tens of millions of workers than the minimum deposit limits now imposed by mutual funds in order to hold down clerical costs and funds management costs. Even with initial deposit limits, which typically run from $500 to $2,000, the most recent survey by Lipper Analytical Services reports that average annual expenses of front-end load stock funds averaged 1.253 percent of funds held on deposit in 1997 (these funds also charge an initial sales charge running as high as 8 percent of funds invested). Annual expenses of no-load funds averaged 1.214 percent of funds held on deposit. The costs associated with the much smaller accounts that would exist under IRA-type private accounts would be much higher because many administrative costs are fixed in nature and would claim a much larger share of the returns on small accounts than they do of the larger accounts that mutual funds now hold. To be sure, some mutual funds incur administrative charges as low as 0.2-0.3 percent of funds on deposit, but the simple fact is that many people consistently choose to put their money in funds with higher costs.

In addition to annual expenses, workers would face additional charges when they reach retirement age if they wish to purchase annuities. The best available study indicates that administrative charges for setting up annuities eat up about 20 percent of the cost of a $100,000 annuity.

About half of this administrative cost covers selling expenses, clerical costs, investment expenses, and company profits and half covers the cost of adverse selection—the fact that people with longer-than-average life expectancies are more likely to purchase annuities than are people with short life expectancies. The first half of those costs are largely inescapable if individuals buy annuities from private insurance companies. The second half of costs arising from adverse selection are inescapable as long as the purchase of annuities is voluntary.

Figures 1 and 2 illustrate the effect of these charges. In Figure 1 I assume that annual administrative costs average only 0.75 percent well below the current average of mutual funds. In Figure 2 I assume that annual administrative costs equal those reported for the United Kingdom by the Government actuary for accumulations that would result from 2 percent annual contributions by a worker with median covered earnings in the United States. The costs of funds management, together with the costs of converting reserves into an annuity, constitute implicit taxes of between 30 percent and 50 percent of the amounts saved. In addition, a major public education effort would be necessary to educate average workers about the pitfalls, as well as about the opportunities, of individual accounts.

If Social Security trust funds invested a portion of reserves in similar assets, it would earn the same rate of return, but would avoid virtually all of the costs of funds management because of economies of scale and it would avoid any added costs of converting savings into an annuity, because it already pays out annuities. The difficulties that the IRA approach to privatization would pose for small businesses, the crippling administrative costs of handling tens of millions of small accounts, and the formidable difficulties of enforcing compliance on millions of small businesses, each with few employees, makes the IRA approach to privatization infeasible, wasteful, unfair, and unworthy of serious consideration.

TSP Plans. Other approaches to privatization avoid many of these administrative problems, but suffer from other serious difficulties. Under individual account plans modeled on the federal employees’ Thrift Saving Plan (TSP plans), current or augmented payroll taxes would be deducted by employers from workers’ pay and transferred to the Treasury. The Treasury would allocate part of the funds to the Social Security system and part to individual accounts. The funds allocated to individual accounts could be allocated according to prior instructions. Or they could be held in a general investment account until several months after then end of the tax year, when actual data on workers’ earnings and payroll tax payments are known exactly, at which point workers could be given the option to allocate their accumulation to one of a small number of broad passively managed funds. To hold down administrative costs, the number of transfers would have to be strictly limited and workers might be permitted to allocate funds only after their balances had reached some stipulated level.

Administrative costs of the TSP approach would significantly exceed those of the current system, but the increase would be manageable.

All of the costs of individual accounts would be in addition to those of the Social Security Administration, which would continue, as all workers would continue to be eligible for benefits under this program. The Thrift Savings Plan costs an average of $23.50 per person in 1997. If individual accounts had the same cost for each of the 140 million workers who have earnings each year, administrative costs would rise $3.3 billion. For the median earner, this cost is approximately 6.5 percent of each year’s contribution, if the contribution to an individual account is 2 percent of earnings. This charge amounts to a 6.5 percent loading charge. For the 30 million workers with earnings of $5,000 a year or less, this initial charge would run 23.5 percent of more of each year’s deposits. However, there are many reasons to believe that administrative costs would be significantly higher under TSP-style individual accounts than they are under the TSP plan. These added costs include

  • education programs, which the Thrift Savings Plan does not provide but would be vital for the mass of workers who are unfamiliar with the attributes of alternative investment vehicles (a minority of workers have IRAs or Keogh plans with mutual funds); these programs could cost $50-100 per year per person, based on studies of private educational programs provided by employers.

  • the provision of detailed information on investment alternatives under the plan; this task, now performed by federal agencies under TSP, would fall to private businesses or government agencies.

  • responding to policyholder questions by phone and by mail; TSP provides no “800” number and agencies handle on-going questions.

  • correcting errors would be more costly, because they would be more frequent and harder to check out, than they are under TSP where the amounts deposited and the allocation among funds are handled wholesale through electronic communication by federal agencies that are both large and quite familiar with the program; individual accounts would be handled by millions of employers, most of whom file paper returns on employee earnings, employ few workers per employer, and would be unfamiliar with the program.

  • tracing spouses after divorce would be more costly under individual accounts than it is under TSP because the paper work routinely required of newly hired federal employees includes detailed information on spouses that private employers do not routinely collect. Such information would be necessary if individual accounts are to be divided at divorce. If individual accounts are not divided at divorce, women would be at risk, as most have much lower earnings than their husbands and would be left with significantly lower benefits.

To add to these problems, millions of workers change jobs each year or hold more than one job—140 million workers generated more than 220 million W-2 forms in the most recent year for which data are available. Monitoring the self-employed and of farmers is particularly vexatious.

Peter Diamond, who is the only person I know to try to measure the administrative costs of a realistic TSP type plan, reports that the total cost of administering it would be equivalent to a loading charge levied on initial investments of approximately 8.8 percent. Such a charge also reduces benefits that can ultimately be paid by 8.8 percent. These costs and the attendant bureaucracy are probably at least as great as those of the current Social Security Administration. Fortunately, they are unnecessary for reasons set forth below. While large in total, they would not be disabling when averaged over all workers. The total returns earned under TSP-type accounts would be marginally lower for the average pensioner than those under Social Security if the trust fund is invested in comparable assets.

While the average returns under TSP-type individual accounts would be only marginally lower than the average returns under Social Security with similar reserves and investment policies, individual workers would face far greater risk. Figure 3, which was prepared by my colleague Gary Burtless, shows the variation of returns under an individual account invested exclusively in what amounts to a total-stock, passively-managed index fund. It shows the variation in replacement rates for an average male worker who entered the labor force at age 20, works to age 62, and converts his accumulation into an annuity without incurring the roughly 20-percent fee mentioned above. The first group of workers enters the labor force in Burtless’ simulation in 1870, the last in 1954.

The most striking feature of Figure 3 is the enormous variation in replacement rates over relatively brief period of time. Workers reaching age 62 in 1976, for example, would have had replacement rates approximately 60 percent smaller than did workers who reached age 62 in 1969. Elected officials may recall the enormous pressures to which they were subject from the so-called “notch babies” after replacement rates dropped a few percent in 1977. They can appreciate the political firestorm that would result from some future drop of replacement rates by percentages that the historical record indicates are common. The fundamental pension system of American workers should not be subject to such uncertainty. If such uncertainty were necessary to achieve the higher returns that can be earned from investments in a diversified portfolio including private securities, it might be worth bearing. But, as indicated above, average returns from individual accounts will actually be lower—a little bit lower or a lot lower, depending on administrative arrangements—than they would be from investing Trust Fund reserves in a similar portfolio.

Restoring Financial Balance in Social Security: A Menu

Table 1 contains a lengthy menu of ways to restore balance in Social Security without fundamentally changing the attributes of the current system that are vital to preserve:

  • the defined-benefit pension structure which spreads broadly the risks inherent in any pension system,

  • the political linkage of social assistance to low earners and large families with middle-class pensions, and

  • full and automatic indexation of benefits, and automatic annuitization.

This menu is more than sufficient to close the projected long-term deficit. Most of the elements of the menu are familiar. Today, I shall focus only on two features of this menu.

  • Without using tax increases or changing the fundamental structure of the Social Security system, it is possible to entirely close Social Security’s projected long-term deficit.

  • Investment of Trust Fund reserves in a diversified portfolio of private and public securities is a powerful instrument for closing the projected long-term deficit. The importance of this policy change depends on the willingness of Congress to take steps to sustain current Social Security surpluses, projected to total more than $1.5 trillion over the next decade under current policy.

Trust Fund Investment in a Diversified Portfolio

Some policymakers and analysts have concluded that it is time to permit Social Security to invest in a diversified portfolio of that would contain private assets as well as government-guaranteed securities. Managers of private pension funds would be roundly criticized and might even by sued for breach of fiduciary responsibility if they invested exclusively in federal government bonds. The reason is that common stocks have generated far higher yields than bonds, public or private, for decades. While the value of a diversified portfolio that includes common stocks and private bonds is subject to some variability, the risk (relative to a portfolio consisting entirely of special Treasury issues) is amply compensated by the added returns. To deny the mass of American workers, who have few financial assets other than Social Security, the returns from investments in common stocks is unfair because it means that they receive less per dollar of pension saving than is available to those with sufficient assets to capture these returns. Nonetheless, proposals to invest Social Security reserves in private securities have been criticized on both economic and political grounds.

Table 1
Closing the Projected Long-term Social Security Deficit: A Menu

Deficit or change in Deficit
As Percent of Payroll
Proportion of
Long-term Deficit Closed
Projected long-term deficit—1998 Trustees Report 2.19 n/a
Effects of correcting the Consumer Price Index -0.45 n/a
Adjusted long-term deficit 1.74 n/a
Program Changes
1. Benefit cut through increasing “normal retirement age”—eliminate 12 year hiatus in increase to age 67 and index age at which unreduced benefits are paid -0.49 28.00
2. Increase the initial age of eligibility from 62 to 64 -0.23 13.00
3. Increase the averaging period from 35 to 38 years -0.25 14.00
4. Invest in common stocks, reaching 40 percent of reserves in 2015 -1.20* 69.00
5. Cover all newly hired state and local employees 0.21 12.00
6. Cut spouses benefits from 50 percent to 33- percent of workers benefits and raise benefits for surviving spouses to 75 percent of the couple’s combined benefit +0.15 -14.00
Sub-Total -2.23 128.00
Tax Increases
7. Tax Social Security benefits the same as contributory private pensions -0.36 21.00
8. Raise Payroll Taxes 0.1 percentage points on employees and employers -0.19 11.00
9. Increase maximum earnings base from 2000 to 2010 to cover 90 percent of earnings -0.58 34.00
Total – 3.36 193.00

* Estimate assumes inclusion of item 7.

Source of estimates: Department of the Actuary, Social Security.

Administration Economic Concerns. The economic criticism holds that investing additions to the Trust Funds in private securities rather than government bonds would have no direct and immediate effect on the nation’s saving or investment or on the capital stock or production. The Trust Funds would earn higher returns because they would hold assets other than relatively low-yielding government bonds, but private savers would earn somewhat lower returns because their portfolios would contain fewer common stocks and more low-yielding government bonds—those that the Trust Funds no longer purchased. These portfolio shifts would cause some significant long-term effects, an increase in national saving if Trust Fund reserves grow by the added return from diversified investments. If one takes account of international capital movements, the full effects cannot be reliably forecast, given the contradictory predictions of open-economy macroeconomic models. This comment is basically correct, but sustaining a policy of limiting the Trust Funds to investments in low-yielding assets perpetuates a distortion that is fundamentally unfair to the majority of workers with few assets other than Social Security.

Investing trust fund reserves in shares would have an important political and economic side-effect that arises from budget accounting rules. Trust Fund purchases of common stocks would be recorded as government expenditures. Thus, Trust Fund surpluses would not appear as surpluses in the unified budget to the extent that they are invested in common stocks. For this reason, additions to Social Security reserves would substitute for added expenditures or tax cuts elsewhere in the federal budget—all of which would directly lower national saving. If the on-budget accounts were in balance, the unified budget would also be in balance if Social Security were invested in assets other than government bonds. Investing in private equities would thereby increase the likelihood that cash-flow surpluses in the Trust Funds translate into added national saving.

Political Concerns. The principle objection to Trust Fund investments in private securities is based on a political concern—the possibility that government officials might use Trust Fund holdings to influence private business decisions. If this concern had substance, the idea of allowing the Trust Funds to invest in private securities should be dismissed and buried.

But these concerns are groundless.

  • To begin with, a Congress that wants to influence private business has ready access to better instruments than the Trust Funds. It can tax, regulate, or subsidize private companies in order to encourage or force them to engage in or desist from particular policies. Manipulation of Trust Fund investments is extraordinarily cumbersome, and under procedures outlined below, the political cost would also be extraordinary.

  • Second, the government now manages billions of dollars of investment in private securities and no problem has arisen. Several government trust funds, including the Thrift Saving Plan for government workers and the pension plans of the Federal Reserve Board, and the Tennessee Valley Authority now invest in private securities. Managers of these funds have not exercised any control over the companies in which they invest. They have pursued only financial objectives in selecting portfolios.

  • A leading advocate of permitting the Trust Funds to invest in private securities, Francis X. Cavanaugh, is a former Treasury official in the Reagan administration, who argued for years against this policy. Then, for his pains, he was placed in charge of the Thrift Savings Plan, the part of the federal government pension system that supervises investments of pension funds in private securities. Cavanaugh found that he c