Statement of Francis X. Cavanaugh, Chevy Chase, Maryland
(former Executive Director and Chief Executive Officer
Federal Retirement Thrift Investment Board)

Testimony Before the Subcommittee on Social Security
of the House Committee on Ways and Means

Hearing on "The Future of Social Security for this Generation and the Next" -
Examining Proposals Regarding Personal Accounts

June 18, 1998

My name is Francis Cavanaugh. I was the first Executive Director and chief executive officer of the Federal Retirement Thrift Investment Board (1986-1994), the agency responsible for administering the Thrift Savings Plan for Federal employees. Before that, I served in the U.S. Treasury Department (1954-1986) as an economist and as director of the staff providing advice on Federal debt management and related Federal borrowing, lending, and investment policies. I am currently a writer and public finance consultant. I represent no clients and speak only for myself.

I am happy to participate in this hearing on the administrative costs and feasibility of establishing personal savings accounts (PSAs) within the Social Security system.

Summary

I will comment briefly on the four issues your subcommittee asked me to address.

First, the effect of Social Security PSA investments on the capital markets would be slight and gradual increases in demand for corporate stocks and possibly other private investments and corresponding decreases in demand for U.S. Treasury securities. As funds are diverted from the Social Security trust fund, which is now invested solely in Treasury securities, Treasury will increase its borrowing in the private market. Treasury borrowing costs might increase slightly; such increases might be offset in part by savings to the Treasury from reducing issues to the trust fund on preferential terms. PSA investments in stocks would probably never exceed two percent of the capitalization of the U.S. stock market.

Your second issue, the cost and administration of the PSAs needs much more study. This country has no experience with a mandatory system of individual accounts dependent upon the performance of very small employers and very low-income employees. In 1994, 46 percent of workers for whom Social Security taxes were paid earned less than $15,000. Assuming two percent of a $15,000 income, or $300, were invested annually in a PSA, the earnings on such a small account would generally be more than offset by the cost of servicing the account in the first several years of the plan. The net earnings of the average PSA would probably never equal the net earnings of the funds invested in Treasury securities in the Social Security trust fund.

The administration of PSAs for the 140 million Social Security employees, if modelled after the 2.3 million member Federal Thrift Savings Plan (TSP), would require at least 10,000 highly trained Federal employees to man the telephones and answer employee questions. PSAs would require the cooperation of 6.5 million private employers, most of whom could not meet TSP reporting standards. Federal agencies generally report payroll deductions and other employee data to the TSP on magnetic tape, but over 80 percent of private employers are still reporting to the Social Security Administration on paper, an extraordinarily costly and error prone process. The cost of error correction, say for failure to make timely stock market investments, would be more than many small employers could bear.

Your third issue, investor education, is a statutory requirement for the TSP. The Office of Personnel Management has the primary statutory responsibility for TSP training, but the Thrift Investment Board conducts hundreds of training sessions each year throughout the country for personnel and payroll officers and for individual plan participants. These sessions, along with the TSP summary plan document, animated video, investment booklet, pamphlets, posters, and other materials, require extensive support from the Federal employing agencies. Such support could not be provided by most of the 6.5 million employers in the Social Security program, given their lack of resources, the relatively low income of the average private employee, and the language difficulties. Meeting TSP standards, if possible at all, could be accomplished only at a price so high as to reduce net investment earnings to unacceptably low levels.

As to your fourth issue, the role of employers, large and small, in a PSA system, large employers with competent personnel, payroll, and systems experts could be expected to perform the functions now performed for the TSP by Federal employing agencies. Yet most private employers have less than 10 employees. Also, household employers who hire part-time providers of cleaning and other domestic services are obviously ill equipped to meet the employee information needs of a PSA system.

I believe it would be impossible to establish cost-effective TSP-type PSAs for the Social Security system. That is, the net investment earnings (after administrative expenses) of the PSAs would be less than the net earnings of Social Security trust fund investments in Treasury securities. Nor would the IRA-type alternative be cost-effective, because of the relatively high administrative costs of small accounts.

The only feasible way for the Social Security system to benefit from the higher returns offered by the stock market is to invest a portion of the trust fund in stocks, which is what virtually all large public and private pension and retirement funds have already done.

I will now discuss these issues in more detail.

The Thrift Savings Plan or 401(k) Approach

The TSP has 2.3 million accounts and is the largest defined contribution plan in the nation, although small compared to any plan for over 140 million Social Security workers. The TSP record keeper maintains a highly trained staff of 150 persons to respond to telephone questions from TSP participants. If the PSA structure were modelled after the TSP, a telephone staff of at least 10,000 would be necessary, especially since PSA participants would generally have less education, income, and employer support than TSP participants.

PSAs in fact could not be modelled after the TSP, which is structured much like the voluntary 401(k) defined contribution plans offered by most large corporate employers. The TSP requires a highly complex central record keeping system, and it depends on the Federal employing agencies and their expert personnel, payroll, and systems people to handle its "retail" operations throughout the world. This includes the distribution of TSP forms and other materials, employee education programs, and individual counselling. Each agency is required to provide employee counselling on all aspects of the retirement system, including the TSP, and the Office of Personnel Management is required under the TSP statute to provide training for the agency counsellors.

Employers are also responsible for the timely transmission of data to the TSP record keeper each payday for each employee's contributions, investment choices, interfund transfers, loans, loan repayments, withdrawals, and other essential information to ensure prompt and accurate investment and maintenance of employee accounts, including the restoration of employees' lost earnings because of delayed deposits or other employer error. While PSA proponents may not contemplate emergency loans or withdrawals, 401(k)s and the TSP permit them. I believe that it would be politically impossible to deny emergency access to funds once their ownership is vested in the names of individual account holders.

Private employers are now required to report individual Social Security tax information only once a year. Surely there would be millions of small employers who would be unwilling or unable to assume the additional administrative burden of PSAs and the corresponding financial liability, for example, to make up for lost stock market earnings resulting from employer failure to process an employee's interfund transfer request on time.

Even if the 401(k) approach were made workable for PSAs, perhaps by adopting (politically unpopular) measures such as exempting small employers or limiting the earnings or options of very small investors, net investment earnings would probably still be much less than would have been earned from Social Security fund investments in Treasury securities. According to the Social Security Administration, 46 percent of Social Security workers, including part-time and temporary workers, earned less than $15,000 a year in 1994. Servicing such small accounts would entail unacceptably high expense ratios.

A PSA deposit of two percent of a $15,000 income would produce contributions of $300 in the first year. Assuming the annual cost of servicing an account is $30⁽¹⁾ (or $4.2 billion for 140 million accounts), then the expense ratio would be ten percent, or 1000 basis points, compared to the TSP net expense ratio of 7 basis points in 1997.⁽²⁾ That ratio would clearly exceed the real (after inflation) returns from PSA investments in a balanced portfolio of stocks, bonds, and other instruments in the first year of the plan. Moreover, since individuals with incomes below $15,000 tend to be risk averse and thus avoid stocks⁽³⁾ in favor of lower yielding fixed-income investments, their net earnings (after expenses) would likely be negative for several of the early years of the plan.

By contrast, and contrary to popular belief, the Social Security trust fund now receives a relatively attractive net return on its investments in special issues of Treasury securities. The average annual interest rate on such issues over the past 30 years has been approximately 8.3% (about 3% after inflation). The trust fund is given preferential treatment, compared to private investors in Treasury securities: it is not required to pay any brokerage or security transaction costs, it receives the (higher) long-term interest rate on its short-term investments, and it is insulated from market interest rate risk by being guaranteed par value redemption on securities redeemed before maturity.⁽⁴⁾ These securities are safer and more liquid than short-term market instruments such as Treasury bills or bank CDs which pay substantially lower rates.

The IRA Approach

An alternative suggested by some PSA proponents is to require employees to set up IRA-type accounts at private financial institutions selected by the employees. Employers could then be required to send the prescribed percent of pay to the various financial institutions chosen by each of their employees. This IRA alternative has the advantage of being much less burdensome on small employers than the 401(k) approach. Yet IRAs are generally much less cost-effective than 401(k)s because the 401(k)s have the advantage of professional fund management, bargaining power in financial markets, and other economies of scale. The average annual expense ratio for stock mutual funds over the past decade has been estimated by Vanguard at approximately 200 basis points, including transaction costs,⁽⁵⁾ and the PSA accounts would be much smaller and thus relatively more costly to maintain.

As indicated above, a typical PSA might have an expense ratio of about 10 percent in the first year of the account. It would take many years before such an account would earn a reasonable net return after administrative expenses. Over the past 30 years, the average annual real (after inflation in excess of 5 percent) return was approximately 3 percent for Treasury bonds, 6 percent for common stocks, and from 0 to a minus 1 percent for Treasury bills and various other short-term instruments, including bank CDs and money market accounts.

Yet many "risk averse" low-income PSA investors would undoubtedly seek the apparent safety and simplicity of a CD or money market account at their local bank or credit union, which would have provided over the past 30 years no net return after inflation (compared to a net 3% return from the Treasury bonds in the Social Security trust fund). Even under the very optimistic assumption that PSA investors would in time allocate their accounts on average one-third to stocks (at 6 percent), one-third to bonds (at 3 percent), and one-third to CDs (at 0 percent), for an average return of 3 percent after inflation (but before administrative expenses), those investments could never catch up with the 3 percent return of the Social Security trust fund.

The suggestion by some that competition would force financial institutions to lower costs substantially is doubtful. The market for personal savings and investments is already well established and highly competitive. More aggressive competition for small accounts would add substantial marketing, promotion, advertising, and high pressure sales costs.

Also, given the likely concerns about exploitation of small investors by the sharp practices of many financial advisers and investment managers, Congress would likely impose new regulatory burdens which would add to administrative costs.

Congress specifically rejected IRA-type proposals when it designed the TSP:⁽⁶⁾

Because of the many concerns raised, the conferees spent more time on this issue than any other. Proposals were made to decentralize the investment management and to give employees more choice by permitting them to choose their own financial institution in which to invest. While the conferees applaud the use of IRAs, they find such an approach for an employer-sponsored retirement program inappropriate.

. . . .

The conferees concur with the resolution of this issue as discussed in the Senate report (99-166) on this legislation:

As an alternative the committee considered permitting any qualified institution to offer to employee[s] specific investment vehicles. However, the committee rejected that approach for a number of reasons. First, there are literally thousands of qualified institutions who would bombard employees with promotions for their services. The committee concluded that employees would not favor such an approach. Second, few, if any, private employers offer such an arrangement. Third, even qualified institutions go bankrupt occasionally and a substantial portion of an employee's retirement benefit could be wiped out. This is in contrast to the diversified fund approach which could easily survive a few bankruptcies. Fourth, it would be difficult to administer, Fifth, this "retail" or "voucher" approach would give up the economic advantage of this group's wholesale purchasing power derived from its large size, so that employees acting individually would get less for their money.

The conferees' concern about giving up "wholesale purchasing power" is very relevant here because investments by individual accounts, rather than by the Social Security trust fund, either in bonds or stocks, would be an enormous sacrifice of wholesale purchasing power.

Of course, the conferees' comments were from the perspective of the Federal government as an employer; it is not clear whether Congress would take a more or less paternal view in the case of Social Security.

The insurmountable problems with the PSA proposals are that (1) they shift Social Security from central financing to small individual accounts, thus losing economies of scale, and (2) they shift the investment risk from the group to the individual, thus violating the first principle of insurance.

Both economically and administratively, Social Security taxpayers would be much better off if any stock or other security investments were made by the collective Social Security fund, rather than by individual investments. Based on the assumptions in the 1997 report of the Advisory Council on Social Security, a gradual investment in stocks of up to 40 percent of the Social Security trust fund would produce a stock portfolio of an estimated $1 trillion (1996 dollars) in 2014. Yet the rapid development and growth of a variety of index funds in the United States and abroad should provide ample opportunities for substantial diversified investments of Social Security funds with minimal market impact. The capitalization of the U.S. stock market today is approximately $12 trillion, and at the Council's assumed growth rate it would be close to $40 trillion in 2014. The Council also contemplated investment in foreign stocks, which would reduce the estimated impact of Social Security stock investments on the U.S. stock market to less than 2 percent. (PSA investments of just two percent of incomes would of course have a much smaller impact on the stock market.) The Council's assumed 40 percent allocation to equities is quite modest -- a 50 percent allocation would be more in line with the portfolio mix of other retirement funds. The TSP currently has 51 percent in equities, and Pensions and Investments (January 26, 1998) reports that the top thousand defined benefit plans hold 62 percent of assets in equities and that the top thousand defined contribution plans hold 65 percent in equities. Based on the Advisory Council's investment return assumptions, a 50 percent allocation to equities in the Social Security fund would slightly more than double the investment earnings of the fund.

To those who say that an individual account approach is needed to increase real savings in our economy I would say that such real savings would be significantly reduced by the high administrative expenses associated with small individual accounts -- greater real savings would be realized by channeling any increased Social Security taxes into centralized investment in the Social Security trust fund. To those who say that an individual account approach is needed to change the income redistribution or generational effects of Social Security financing I would say the first priority should be to enlarge the total Social Security pie, through more rational investment policies, so that we may better deal with any equity issues -- a rising tide lifts all boats. Then those who would change the distribution of shares, by income or generation, could do more for some without hurting others so much.

Even if some sort of PSA is added to the Social Security system, a large portion (I would suggest up to 50 percent) of the remaining Social Security trust fund clearly should be invested in equities, which is what virtually all large public and private pension and retirement funds now do.

1. According to the "Report of the 1994-1996 Advisory Council on Social Security, Volume 1" (Washington, D.C.), 100, $30 per year is typical of charges levied for IRAs for flat dollar account maintenance fees.

2. The net expense ratio is the gross expense ratio minus forfeitures and is the administrative charge to TSP participants. For example, in 1997 the gross expense ratio was .09, and the net expense ratio of .07 represented a charge to participants of $0.70 for each $1,000 of their TSP account balances. The expense ratios have declined steadily since 1988, when the gross ratio was .67 and the net ratio was .34.

3. In 1995, only 6 percent of families with incomes less than $10,000 and only 25 percent of families with incomes from $10,000 to $25,000 had any direct or indirect stock holdings. Arthur B. Kennickell and Martha Starr-McCluer, "Family Finances in the U.S.: Recent Evidence from the Survey of Consumer Finances," excerpt from Federal Reserve Bulletin, January 1997, 12.

4. Francis X. Cavanaugh, The Truth about the National Debt: Five Myths and One Reality (Boston: Harvard Business School Press 1996), 158.

5. The Vanguard Group, "In the Vanguard," Summer 1996 (Valley Forge, PA), 10.

6. H.R. Rep. No. 99-606, at 137-38. reprinted in 1986 U.S.C.C.A.N. 1508, 1520-21.