Social Security: Analysis of a Proposal to Privatize Trust Fund Reserves

Published by the Government Accountability Office on 1990-12-12.

Below is a raw (and likely hideous) rendition of the original report. (PDF)

    Analysis of a Proposal
    to Privatize Trust
    Fund Reserves


RESS’IYRICTED--Not    to be                   e the
General Accounting Office unless specilically
approved by the Office of Congressional

          United States
GAO       General Accounting OfTice
          Washington, D.C. 20548

          Human Resources Division


          December 12,lOOO

          The Honorable John E. Porter
          House of Representatives

          Dear M r. Porter:

          In response to your June 16, 1989 letter, and later discussions with you
          and your staff, we are reporting our analysis of your proposal to create
          a new system of Individual Social Security Retirement Accounts (ISSRAS).
          Under this proposal, the accumulating reserves of the Social Security
          Trust F’unds would be returned to workers and placed in ISSRAS    where
          they would be invested in the private sector and held for individuals
          until their retirement. In effect, the proposal would partially and tempo-
          rarily privatize Social Security.

          You have stated that the objectives of this proposal include:

      . unmasking the general fund deficit, that is, the shortfall in revenues for
        financing the operations of the federal government, excluding trust fund
      l increasing the national savings rate and American competitiveness; and
      9 providing individuals an opportunity to improve their overall retirement
        income by earning a higher return on their contributions in the private
        sector than under Social Security.

          At your request, we examined the potential of your proposal to accom-
          plish these objectives.’ We found that, conceptually, the ISSRAproposal
          could be integrated with the existing progressive benefit structure and,
          given favorable financial market conditions, could improve retirement
          incomes. If implemented, the ISSRAmay change the m ix of public and
          private saving but not necessarily the magnitude of national saving. The
          proposal also raises numerous administrative difficulties and policy
          issues that need to be resolved before ISSRAS could be considered a fully
          operational alternative for the use of trust fund reserves.

          As part of our analysis, we expanded on our previous work on the
          reserve accumulation issue and its relation to the budget deficit and

          ‘We found that the proposal would have unmasked the n&8ocial Security portion of the deficit as
          the federal deficit was defined before the passage of the &nnibus Budget Reconciliation Act of 1990.
          As you,,know, the act removes social security from the deficit calculation previously employed under
          the B&mced Budget and Emergency Deficit Control Act of 1986 (i.e., Gramm-Rudman-Hollings). As
          such, the 1990 act sets a path for reducing the non-Social Security deficit and potentially making the
          Social Security surplus economically meaningful by increasing national savings.

          Page 1                                     GAO/HRD91-22      Privatizing   sodal Security Trust Fund

                         national savings2 We also reviewed the literature on Social Security
                         financing and discussed ways to divert social Security taxes to private
                         accounts with public- and private-sector experts.

                         In this letter, we present the basic elements underlying the ISSRApro-
                         posal and summarize our analysis and conclusions. In appendixes I
                         through III we present additional discussion concerning: (1) a method
                         for adjusting Social Security benefits to account for diversion of payroll
                         taxes to ISSRGS  (app. I), (2) the historical rates of return in the private
                         market and how these might affect the retirement benefit levels of indi-
                         viduals under the ISSRGprogram (app. II), and (3) issues surrounding the
                         administration and implementation of the ISSRAproposal (app. III).

                         Since the 1983 policy changes that led to the reserve accumulation,
Background               many Members of Congress had become concerned about the Social
                         Security reserves concealing the worsening deficit in federal operations.
                         We testified in February 19903that the large and continuing federal def-
                         icit is limiting the nation’s ability to increase national savings and pro-
                         mote more rapid economic growth-growth           that could improve living
                         standards and lessen the economic burden of paying retirement benefits
                         to a larger elderly population. The ISSRAproposal was aimed at focusing
                         the debate on ways to ensure that the growing Social Security reserves
                         would contribute to national saving. It also renewed a longstanding
                         debate concerning the relative role of the private sector versus the
                         public sector in providing retirement income security.

The Trust Fund Reserve   The Social Security Amendments of 1977 and 1983 moved the Old-Age,
Accumulation             Survivors, and Disability Insurance (QASDI) trust funds away from their
                         traditional, pay-as-you-go financing basis toward the accumulation of a
                         substantial, though temporary, reserve. The primary purpose of these
                         changes was to correct both the short- and long-term financial problems ’ ,
                         of the OASDI trust funds. In doing so, however, the changes had the effect
                         of causing a large buildup of reserves that would continue over the next
                         four decades. Projections are that the reserves will peak in about 2026
                         at $9 trillion (about $2 trillion in today’s dollars).

                         2Social Security: The Trust Fund Reserve Aawmulatkm, the Economy, and the Federal Budget
                         (-44,               Jan. 19,1999).
                         3The Question of Rolliug Back the Payroll Tax: Umualdng the Deficit Illusion (GAO/T.-HRD90-10,

                         Page 2                                   GAO/HBDB~-~~     privatizingsodal security‘bust F’md
                           Some see the accumulation of reserves beyond contingency levels’ as an
                           attempt to deal with an abrupt and permanent shift in American demo-
                           graphic patterns expected early in the twenty-first century. When the
                           baby-boom generation starts to retire, around 2010, there will be a
                           declining ratio of workers to retirees. Whereas today 3.4 workers sup-
                           port each Social Security retiree, by 2030 each beneficiary will be sup
                           ported by only 2 workers. The reserve will be drawn down to pay
                           benefits to a continuing higher number of retirees. By around 2040, the
                           reserves will have declined to the pay-as-you-go level. Thereafter,
                           increased revenues (most likely higher Social Security payroll taxes)
                           will be needed to maintain benefit levels.

The Reserve Accumulation   The future impact on workers from the increased burden of supporting
and National, Savings      retirees depends largely on the behavior of our economy over the next
                           severaI decades. Fundamentally, future workers will have to provide
                           goods and services both for themselves and for relatively more retirees.
                           But workers can shoulder this burden while also experiencing higher
                           living standards if the economy undergoes sustained and steady growth
                           leading to higher real incomes. As we previously reported, increasing
                           our national savings rate may be the single most important step we can
                           take to achieve this result.

                           In recent years, however, the United States has had relatively low rates
                           of national saving, net investment, and productivity growth compared
                           with other major industrial countries. In the 198Os, the low rate of
                           saving has been depressed further by our very large budget deficits.

                           In principle, a higher rate of national saving could come about in several
                           different ways. The only one directly subject to government control,
                           however, is the federal budget. Just as budget deficits are a drain on
                           saving, a budget surplus adds to national saving. In this regard, the
                           Social Security reserve accumulation provides a potential source for
                           such an overall budget surplus that can raise our national savings rate
                           as long as we continue to achieve reductions in the non-Social Security
                           deficit. The 1990 budget agreement is an important step toward
                           achieving this goal.

                           4A aWingency    reserve level is an amount adequate to cover periods of recession or other relatively
                           short-term fluctuations in income or expenditures. By the mid-lQQOs,a contingency reserve of 100 to
                           160 percent of annual outlays is expected to be reached.

                           Page 3                                     GAO/HRD91-22      Privatizhq   !SocM Security Trust Fund

                     The ISSRA proposal seeks to allocate the reserve accumulation by chan-
The ISSRA Proposal   neling part of the Social Security payroll tax into private individual
                     accounts.6 An important condition of the proposal is that Social Security
                     benefits be adjusted to account for the diversion of payroll taxes in a
                     way that retains the underlying progressive structure of the system.G

                     Based on Social Security projections, we determined that from 1990
                     until about 2015,’ Social Security will collect contributions that exceed
                     benefit costs by about 2 percent of taxable payroll. Applying this figure,
                     about 24 percent of the contributions made toward retirement benefits
                     alone (not including survivor and dependent benefits) could be placed in
                     a special account created for each worker, These ISSRAS    would then be
                     managed and invested in the private sector where they would earn a
                     private market return and be held until retirement.

                     Under the proposal, Social Security benefits would be reduced to com-
                     pensate for the mandatory and temporary diversion of Social Security
                     contributions to the private-sector ISSRA.At retirement, benefits would
                     then come from two sources-the reduced Social Security benefit and
                     the balance accumulated in the ISSRA.The worker’s total Social Security
                     retirement income would continue to come mostly from Social Security
                     benefits but would be supplemented by the ISSRA.

                     6You have presented the basic elements of the ISSRA proposal in the article “Let Workers Own Their
                     Retirement Funds. , .” m The Wall Street Journal, February 1,lQQO; in testimony before the Senate
                     Finance Committee, February 16,1QQO;and in a luncheon speech before the Cat0 Institute on March
                     14, 1990. Senator Symms, in S.2026, has proposed creating private retirement accounts without
                     reducing Social Security benefits, and Merrill Lynch, Morgan Guaranty, and others in the private
                     sector have also suggested diversion of funds to private-sector accounts. For example, see Morgan
                     Guaranty Trust Company, “U.S. Social Security Surpluses: Pitfall or Opportunity?“, World Financial
                     Markets, July 1988 and, Enid A. Borden and Bruce D. Schobel, “Meeting the Future Needs of the
                     Elderly”, Report to Merrill Lynch (undated).
                     “Included in the 1990 Budget Act are so-called Social Security “firewall” procedures to inhibit
                     changes in Social Security revenues and benefits arising from new legislation. Different procedures
                     are established for the House of Representatives and the Senate. The House procedures permlt mem-
                     bers to block legislation that changes the 76year actuarial balance by more than 0.02 percent of
                     taxable payroll or results in net benefit increases or net tax decreasesthat within a S-year period
                     exceed $260 million. The Senate procedures permit members to block a budget resolution that would
                     decrease the excess of Social Security revenues over outlays over a 6-year period.

                     7After about 2016, the payroll tax rate will no longer exceed the program’s benefit costs as a per-
                     centage of taxable payroll. Also note that the discussion here is in terms of the combined employer
                     and employee contributions.

                     Page 4                                     GAO/HRD-91-22      F’rivatizhg   Social Security   Trust Fund

                             The ISSRAproposal is targeted at the excess* payroll tax contributions
                             for retirement benefits (not including survivors and dependents bene-
                             fits) and not at altering the social objectives of the Social Security pro-
                             gram. It does not affect the financing and structure of the disability
                             insurance and Medicare programs. Moreover, under the proposal the
                             benefit adjustments made in the retirement component of the program
                             would continue to be progressive, affording low earners proportionally
                             higher benefits (relative to preretirement earnings) than high earners in
                             the same way the Social Security program does now.

                             In addition, to ensure the financial soundness of investments, ISSRA
                             trustees would be required to invest funds according to specified rules,
                             could not release funds until retirement, and would be liable for
                             breaches of fiduciary rules.@Within these guidelines, workers would
                             own and control their ISSRAS and could choose their own trustees, which
                             would most likely be banks, insurance companies, and other money man-
                             agement firms. ISSRAS could become part of the estate of a deceased
                             worker or retiree.

                             Using ISSRA funds, individuals would be required to purchase a Social
                             Security-like annuity upon retirement. This annuity, plus the reduced
                             Social Security benefit, would represent the individual’s retirement ben-
                             efit. Depending on the market return on the ISSRA, an individual’s com-
                             bined retirement income (Social Security plus ISSRA) might be the same,
                             higher, or lower than under the existing program.

GAO’s Analysis

Effect of Proposal on        As noted earlier, the low rate of national saving has been depressed by
National Saving Is Unclear   our very large budget deficits in the 1980s. Savings come from individ-
                             uals, or personal savings; businesses, chiefly in the form of retained
                             earnings; and government, in the form of budget surpluses. The Social
                             Security surpluses provide an opportunity to raise our saving rate if the
                             significant budget deficit reductions programmed in the Omnibus Budget

                             sin this report, reference to “excess” payroll taxes or revenues (i.e., surpluses) pertains to those
                             annual payroll tax revenues that exceed annual Social Security costs, after allowing for an adequate
                             contingency reserve in the Trust Fund. For discussion of different Social Security financing concept-s,
                             see GAO/HRD-89-44.
                             OFor further discussion of fiduciary responsibilities see appendix III.

                             Page 5                                       GAO/HRD91-22      Privatizlng   Social Security   Trust Fund

                        Reconciliation Act of 1990 are achieved, resulting in an overall govern-
                        ment budget surplus.1o

                        By itself, the ISSRA proposal may alter the mix of public and private
                        saving but not necessarily the magnitude of national saving. By
                        requiring the trust fund to pay excess revenues back to workers via
                        their ISSRAS, the proposal reduces the trust fund surplus and overall gov-
                        ernment saving. At the same time, moving the excess revenues to the
                        private sector increases personal saving.

                        Given that the general fund is in deficit, taking revenue away from the
                        Treasury increases the amount that the Treasury must borrow from
                        other sources. Indeed, if no revenue or spending changes are made in the
                        general fund portion of the budget, much of the ISSFZA money may end up
                        being loaned back to the Treasury just as if it had been kept in the trust

Eknefit Adjustment      Through its benefit formula, the Social Security program pays benefits
Retains Progressivity   that implicitly redistribute income from high earners to low earners.
                        This progressive benefit structure is essential to the social adequacy
                        function of Social Security, which aims to provide all retirees with a
                        minimum level of support. As a result, high earners implicitly receive a
                        lower rate of return on their Social Security contributions than low

                        Under the ISSRA proposal, when high earners divert some of their Social
                        Security contributions into their ISSRAS,these diverted funds are no
                        longer available for redistribution. We devised a benefit adjustment that
                        recovers the amount of the subsidy that high earners would have pro-
                        vided to low earners.” First, we determined the average implicit rate of
                        return each age group would earn from Social Security.12 We then
                        adjusted benefits by subtracting for each worker an amount equivalent
                        to the annuity that the worker’s diverted taxes would have purchased if

                        loIn the short run, this shift to a more restrictive budget policy could have effects on aggregate
                        demand leading toward lowered growth of output and employment and in turn decreasedtotal
                        saving. Other policy adjustments, such as less restrktive monetary policy may be necessary to main-
                        taln the level of aggregate demand and mitigate the effects of restrictive budget policy.
                        “To the extent that this adjustment under the ISSRA proposal makes more apparent the implicit
                        subsidy that high earners contribute, the acceptanceof the current system by high earnem might be

                        * 2The rate of return varies by age group becauseof changing tax rates, benefit rules, and actuarial
                        factors that reflect mortality rates,

                        Page 0                                     GAO/HRD91-22       Privatizing   Social Security Trust Fund

                           they had earned this age group rate of return. Thus, by applying the
                           same age group rate to everyone’s benefit reduction, even though ev-
                           eryone actually earns different rates of return under Social Security, the
                           adjustment achieves exactly the current system’s redistribution of
                           income. Appendix I explains how our adjustment preserves Social
                           Security’s current progressivity.

Advantage to Individuals   The proposal’s impact on individual retirement incomes depends largely
Dependson Market           on how the return on ISSRA investments compares with the return on
                           Social Security. Examining historical rates of return on various invest-
Interest Rate              ments, only stock market investments would offer the potential for a
                           significant improvement over Social Security. Also, any improvement
                           could vary substantially across different age groups.

                           The ISSRA proposal reflects, in part, the belief that workers could earn a
                           higher return from private-sector investments than what Social Security
                           offers. Using our benefit adjustment, workers’ ISSRArate of return must
                           exceed the Social Security age group rate of return in order for them to
                           be better off under the proposal. Our analysis suggests that average age-
                           group Social Security rates of return, adjusted for inflation, range from
                           more than 3 percent for those born in 1930 to about 2 percent for those
                           born after 1960.

                           Our research shows that whether workers would do better with their
                           ISSRAS  than with receiving an inflation-adjusted 2 percent rate of return
                           is likely to depend on the types of investments the program would allow,
                           the actual investments made, and the particular years over which the
                           worker participates.13 (See app. II for more details on historical invest-
                           ment data and for illustrative examples of outcomes in retirement
                           incomes under various assumptions.)

                           For the purposes of our analysis, we assume that the private market
                           will offer annuities comparable to the Social Security benefit.14That is,
                           retirees would be able to use their ISSRAS to purchase annuities at unisex
                           rates that are adjusted for cost-of-living increases. Moreover, we assume
                           negligible load factors; that is, charges by the annuity provider for

                           131nour analysis, we use market rates of return before taxes. Any taxation of ISSRA income would
                           reduce the effective rate of return and the I!?SRAannuity. The same is true for fees paid to those who
                           msnage the accounts. The extent of admiitrative     costs will depend on the specific structure and
                           provisions of the ISSFtA program. If, for example, the cost of managing ISSRAs was analogous to
                           mutual funds, such costs could reduce rates of return by about 1 percentage point.

                           14As noted earlier, the ISSRA proposal excludes survivors, dependents, and disability benefits.

                           Page 7                                     GAO/HRD-91-22     Pcivatizlng   &Aal   Security Trust Fund

profit, risk, and administrative costs, Also, we calculate annuity prices
using the interest rate earned on the Trust Funds.

In practice, however, we cannot presume what annuity features the pri-
vate market would offer under the ISSRAproposal. Currently, the private
market does not offer annuities at unisex rates with cost-of-living
adjustments and does have substantial load factors.16 On the other hand,
private annuity prices reflect higher interest rates than those earned by
the Trust Funds, which would make private annuities more generous.
Any judgments about the likely net effect of these offsetting factors
would be speculative.

Since 1967, inflation-adjusted returns on the Standard and Poor (S&P)
600 stocks have averaged 6.6 percent. Over the same period, returns on
highly graded corporate bonds have averaged 1.6 percent. (See fig. 1.)
Over shorter periods, returns on these assets have varied markedly. For
example, over the decade of the 197Os, stocks and bonds earned returns
of -1.4 and -1.1 percent, respectively. In contrast, during the 198Os,
returns averaged 11.8 and 7.6 percent, respectively.

Because ISSRAS will be expected to exist over long periods of time, we
also examined the compound annual returns over rolling 20-year
periods. Again, we found that earnings varied widely with the time
period used. For rolling 20-year periods since 1967, inflation-aausted
annual returns for S&P 600 stocks ranged from about 1 percent to over
6 percent. For corporate bonds they ranged from about -2.7 percent to
roughly 3.2 percent.

This degree of variation affects workers retiring in slightly different
years. If an ISSRAearned interest at the inflation-adjusted rate that the
S&P 600 earned from 1967 to 1986, for instance, it would earn 3.7 per-
cent; but another worker’s ISSRA earning interest at the rate from the 20-
year period 1966 to 1984 would earn only 1.4 percent. Using inflation-
adjusted rates for corporate bonds, an ISSRAearning interest at the rate
from 1970 to 1989 would earn 3.2 percent; but another ISSRA earning
interest at the rate from the 20-year period starting just 6 years earlier,
1966 to 1984, would earn -1 percent.

16For information on loading factors see Benjamin M. Friedman and Mark Warshawsky, “Annuity
Prices and Saving Behavior in the United States” in National Bureau of Economic Research, Pensions
in the U.S. Economy, Ebdie, Shoven, and Wise, eds. University of Chicago Press, 1988.

Page 8                                   GAO/HRD-91-22     Privatkdng   Social Security Trust Fund

Figure 1: Compound Annual Rates of
Return, Inflation-Adjusted (1957-89)

                                       Note: Data for Standard and Poor 500 before 1957 are not available.
                                       Source: Roger G. lbbotson and Rex A. Sinquefield, Stocks, Bonds, Bills, and Inflation. Chicago: lbbotson
                                       Associates, 1990.

                                       While some deposits will accumulate interest for more than 20 years,
                                       others will accumulate it for less. ISSRAS
                                                                                will pay interest on deposits
                                       from 1 to 46+ years; that is, from deposits made in the last possible year
                                       to those made in the first year of work. But if the program exists only as
                                       long as annual revenues exceed outgo, under the current Social Security
                                       financing schedule roughly one-fourth of the accounts will accrue
                                       interest for no more than 26 years. Thus, workers’ benefits from the
                                       program depend on how long they participated and whether they partic-
                                       ipated during a period of relatively high or low investment returns.

                                       We analyzed the effects of ISSRAS   on retirement income using optimistic,
                                       pessimistic, and moderate assumptions about private rates of return.
                                       For most age groups Social Security pays a rate of return of about 2
                                       percent, adjusted for inflation. If ISSRASwere to pay, on average, a rate
                                       somewhat higher than bonds have paid historically, they would earn 3

                                       Page 9                                      GAO/HlZD-91-22    PrivatizLng Social &m&y       Trust F’und

                            percent.16We have characterized this as a moderate market-rate-of-
                            return assumption. In this case, we find that the ISSRA program would
                            improve retirees’ incomes by 2.7 percent compared with current Social
                            Security benefits.” (See fig. 2.)

                            A more optimistic assumption might be that ISSRAS would earn a
                            ‘I-percent return, which would reflect the historical performance of S&P
                            600 stocks; this would result in a 24-percent increase in retirement
                            income for an average earner. A more pessimistic assumption might be
                            that HSRASwould earn only a l-percent rate of return, which would lie
                            between the performance of S&P 600 stocks over the worst 20-year
                            holding period since 1967 and the worst such performance for bonds
                            over the same period. This pessimistic assumption yields a 2-percent
                            decrease in income for an average earner. Further, for those who are not
                            working over the entire length of the program, the advantage (or disad-
                            vantage) will be proportionally less.

Potential Losses Raise      The likely variation in market returns could result in some retirees being
Benefit Security Concerns   worse off under the ISSRAprogram than they would be under Social
                            Security. This, in turn, could generate pressure for the government to
                            guarantee that every retiree would be at least as well off under the ISSRA
                            system as the retiree would have been by remaining on Social Security.
                            Social Security was created in part to assure a floor of income protection
                            to those whose earnings cease as a result of retirement. Without a guar-
                            antee, unfavorable investment experience could erode this floor.

                            A guarantee should be viewed with great caution, however, since it
                            could encourage ISSFU holders to make risky investments. They would
                            know that a loss on their ISSRA investment would be offset by increased
                            Social Security benefits, while an investment gain would yield them
                            greater retirement incomes. While investment restrictions may mitigate
                            this problem, the possibility that losses may still occur could expose the
                            government to significant costs (see app. III).

                            16As noted earlier, any taxation of ISSRA interest income or administrative costs would reduce the
                            effective rate of return and the 1!3SRAannuity.

                            “These results are for average steady earners. See appendix I for descriptions of these illustrative

                            Page 10                                     GAO/HRD-91-22     Privatizing   Social Security   Trust Fund

Flgure 2: Total Retirement Incomes, by
Market lnterert Rate (Average Steady
Earners Born in 1960)                        1499     Monthly Rotlromont Income (1990 Dollrrr)

                                                    Currrnt Law   ISSRA Proposal *

                                                     I        ISSRA Annuity
                                                     m        Adjusted Benefit
                                                              Current Law Benefit

                                             ‘Bars show alternative market interest rate assumptions. Benefit adjustment assumes inflation-adjusted
                                             age group rate of return of 2 percent under Social Security.

                                             While the benefit adjustment overcomes a significant conceptual hurdle,
Implementation                               many serious technical and administrative problems require solutions
ChallengesNumerous                           before the ISSRA proposal can be implemented. (See app. III.) Among
and Complex                                  these problems are the following:

                                         l The rules or restrictions on investment that might be applied to ISSRAS
                                           could affect the program’s advantages. For example, following fiduciary
                                           guidelines may limit the kinds of institutions that can offer ISSRAS and
                                           impose conservative investment strategies. Similarly, legislative restric-
                                           tions on investment of ISSFU funds could limit the returns to workers
                                           under the program.
                                         . Administration of accounts may be complex. Initially, over 130 million
                                           accounts would have to be set up and maintained by Social Security and

                                             Page 11                                      GAO/HRDBl-22     Privatizing   Social Security Trust Fund


Figure 3: Effect of ISSRA Proporal on Social Security Coat8 (Retired Worker Benefits Only)
14     Pormnt OfT~Ubb         Payroll


0                        ..- _-- -,--.-                --      -     -   -    -   .- ._- .-_--     ..-_-    -     .-   -

lea0         low       a00        a05     2010      2ols           2om        2025     2020      2026      2010        2046   a50     as5        2ow     aLwlll     mo       2076

       -           Current Law Tax Rate
       D           Current Law Cat Rate
       I-I    I    Coat Rate Under ISSRA Proposal

                                                               Note: Assumes ISSRA diversions stop in 2015

                                                             the private sector; ultimately, there could be as many as 230 million.
                                                             Many accounts might be small, particularly for part-time or occasional
                                                             workers or those with small amounts of covered earnings. Some individ-
                                                             uals might not select an institution to receive their BRA deposits.
                                                           . The Social Security Administration (ss~) would have to complete
                                                             processing of employers’ annual statements and crediting of earnings
                                                             shortly after the calendar year in order to determine ISSRAdeposits. This
                                                             may be difficult and could entail a substantial lag in the timely deposit
                                                             of ISSRAfunds. Also, the Department of the Treasury would have to
                                                             make these deposits at a large number of financial institutions and the
                                                             timing of these deposits could entail some very large cash transactions.
                                                             Data would have to be reconciled between ss~ and Treasury and
                                                             between Treasury and private-sector institutions.
                                                           l Whether ISSRASshould be taxed needs to be explored. For example,
                                                             making ISSRA benefits tax exempt would result in a loss of tax revenues
                                                             because some Social Security benefits currently are taxed.

                                                               Page 12                                             GAO/HRD-M-22     Privatizhg    Social Security   Tmat Fund

l   Because the Social Security surplus is projected to end around 2016, at
    about that time ISSRAdeposits will cease. It does not appear that the
    benefit adjustments under the program will be sufficient to generate
    continued surpluses for very long after 2015. After the ISSRA diversions
    end, the benefit reductions under the ISSRA plan will continue to have a
    cost-reducing effect on Social Security for a considerable time. But even-
    tually, as the last cohorts to participate in the ISSRA program die out, all
    other things being equal, the costs of the system will return to the long-
    term cost path shown in figure 3, Continuation of the program or move-
    ment toward a larger role for privatization, in light of projected
    increased costs for Social Security, may present serious challenges for
    maintaining future benefit levels and for financing past benefit

    As arranged with your office, unless you publicly announce its contents
    earlier, we plan no further distribution of this report until 30 days from
    its issue date. At that time, we will send copies to cognizant congres-
    sional committees; the Secretary of Health and Human Services; the Sec-
    retary of the Treasury; the Director, Office of Management and Budget;
    and other interested parties.

    This report was prepared under the direction of Joseph F. Delfico,
    Director of Income Security Issues, who can be reached on (202) 275-
    6193. Other major contributors are listed in appendix IV.

    Sincerely yours,

    Lawrence H. Thompson
    Assistant Comptroller General

    Page 12                         GAO/HID-91-22   Frivatizing   social security   Trust Fund


Appendix I                                                                                                    16
The Benefit              Progressive Benefit Formula Promotes Adequate                                        16
                             Retirement Incomes
Adjustment Under the     Benefit Adjustment Retains Subsidy                                                   18
ISSRA Proposal           An Alternative to the Age Group Rate                                                 23

Appendix II                                                                                                   24
Proposal’s Advantage     Historical Rates of Return Depend on Risk and Inflation
                         Returns Vary Widely Over Rolling 20-Year Periods
to Individuals Depends   Effect of ISSRA Plan on Typical Beneficiaries                                        29
on Market Interest
Appendix III                                                                                                  36
Considerations in        Treasury Transfer of Funds to ISSRAs
                         Management and Investment of ISSRAs
Implementing an
ISSRA Program
Appendix IV                                                                                                   40
Major Contributors to
This Report
Tables                   Table I. 1: Replacement Rates and Implicit Rates of                                  17
                             Return, by Income Level (Workers Born in 1960,
                             Single Earner Families)
                         Table 1.2: Proposal’s Impact on Retiree Incomes When                                 23
                             ISSRAs Earn Social Security Age-Group Rate of
                             Return (By Income Level for Workers Born in 1960)
                         Table II. 1: ISSRA Proposal’s Impact on Retiree Incomes by                           29
                             Market Interest Rate Assumption (Average Steady
                             Earners Born in 1960)
                         Table 11.2:ISSRA Proposal’s Impact on Retiree Incomes                                31
                             Using Moderate Market Rate Assumption (by Age
                             Group for Average Steady Earners)

                         Page 14                       GAO/HRDBl-22   Privatizing   Social Security   Trust F’und

Figures   Figure 1: Compound Annual Rates of Return, Inflation-
              Adjusted (196739)
          Figure 2: Total Retirement Incomes, by Market Interest
              Rate (Average Steady Earners Born in 1960)
          Figure 3: Effect of ISSRA Proposal on Social Security
              Costs (Retired Worker Benefits Only)
          Figure I. 1: The Adjusted Social Security Benefit and the
               ISSRA Annuity
          Figure 1.2: Total Retirement Incomes, by Earnings Level,
              When ISSRAs Earn Social Security Age-Group Rate of
              Return (Workers Born in 1960)
          Figure II. 1: Compound Annual Rates of Return, Inflation-                            26
              Adjusted (1967-89)
          Figure 11.2:Compound Annual Returns for the Decades,                                 26
              Large Company Stocks (1926-89)
          Figure 11.3:Compound Annual Returns for the Decades,                                 27
              Long-Term Corporate Bonds (1926-89)
          Figure 11.4:Compound Annual Returns, Inflation-                                      28
              Adjusted, for 20-Year Holding Periods
          Figure 11.6:Total Retirement Incomes, by Market Interest                             30
              Rate (Average Steady Earners, Born in 1960)
          Figure 11.6:Workers With ISSRAs by Number of Years                                   32
              With Deposits
          Figure 11.7:Total Retirement Incomes, by Age Group,                                  34
              Using Moderate Market Interest Rate Assumptions


          AIME        Average Indexed Monthly Earnings
          IRA         individual retirement account
          IRS         Internal Revenue Service
          ISSRA       Individual Social Security Retirement Account
          OASDI       Old-Age, Survivors and Disability Insurance
          PIA         Primary Insurance Amount
          SSA         Social Security Administration

          Page 16                         GAO/H&D91-22   Privatizing   Social Security Trust Find
Appendix I

The Benefit Adjustment Under the

                      The Individual Social Security Retirement Account proposal specifies
                      two important features regarding Social Security benefit levels: (1)
                      because individuals diverting contributions to ISSRAS would pay less into
                      the Social Security system, their Social Security benefits would be
                      reduced, and (2) the resulting benefit adjustments would retain the pro-
                      gressive redistribution of income embodied in the current system. We
                      examined a variety of benefit adjustment approaches and developed one
                      that would maintain the system’s progressivity. In appendix II we use
                      this approach to analyze the effect of the ISSRA proposal on the benefit
                      levels of representative individuals.

                      The current benefit formula is constructed so that individuals who are
Progressive Benefit   less able to provide for their retirement privately are more likely to
Formula Promotes      receive a “socially adequate” public benefit. This progressive benefit
Adequate Retirement   formula yields benefit levels that are proportionally higher for low
                      earners than for high earners and that, in effect, redistribute income
Incomes               from high earners to low earners.

                      When workers retire, Social Security uses their lifetime earnings records
                      to determine each workers’ Primary Insurance Amount (PIA), on which
                      the initial benefit and auxiliary benefits are based.’PIA is the product of
                      two factors-the Average Indexed Monthly Earnings (AIME) and the
                      benefit formula. The AIME is determined by taking the lifetime earnings
                      record, indexing it, and taking the average. To determine PIA, AIME is
                      then applied to a step-like benefit formula, shown here for 199OC ’

                            PIA =          0.90 X     (AIWE   up to $366)   ’
                                       +   0.32 X     (AIME   between $367 and $2,146)
                                       +   0.15 X     (AIME   over $2,145).

                      The percentages in this formula replace a smaller proportion of income
                      as incomes rise and make it progressive. That is, a high percentage
                      (90%) of AIME is replaced for the first several hundred dollars of AIME. A
                      lower percentage (32%) is replaced for the next bracket and a still lower
                      percentage (16%) is replaced for the highest bracket of AIME.

                      ‘PIA is the key element used in determiniig not only the retired worker’s benefit but also the amount
                      of dependent and survivors benefits. However, as noted earlier in thii report, the ISSRA proposal
                      applies only to retired worker benefits.
                      2AIME brackets are updated snnually (i.e., indexed) based on the change in annual average covered

                      Page 16                                    GAO/HRD91-22      Privathing   !Social Security Trust Fund
                                          Appendix I
                                          g~~g~uetrnent              Under the

                                          The ratio of PIA to the worker’s earnings in the year before retirement is
                                          referred to as the “replacement rate.” For example, a worker who
                                          always earned the average wage and retired in September 1990 at age
                                          66 would have an AIME of $1,424. The worker’s PIA would, therefore, be
                                          $720.70, and the replacement rate would be 40.2 percent.

                                          Replacement rates provide one way of illustrating the formula’s
                                          progressivity. Table I.1 shows that replacement rates are higher for low
                                          earners than for high earners. The table shows three earnings levels for
                                          hypothetical steady earners born in 1960 and retiring in 2027:

                                          Low earners have credited earnings equal to 46 percent of the average
                                          steady earner’s earnings; in 1990, this earnings level would be $9,692
                                          per year.
                                          Average earners have credited earnings in each year equal to the
                                          average of all wages covered by Social Security for that year; in 1990,
                                          this earnings level would be $21,537 per year.
                                          High (maximum) earners have credited earnings equal to the maximum
                                          earnings subject to Social Security taxes in each year; for example, in
                                          1990, the first $6 1,300 of annual earnings is subject to Social Security

Table 1.1: Replacement Rate8 and
hllpkit Rat08 Of ROtUrn, by hICOme Levd   Figures are percentages
 Workers Born in 1960, Single Earner
L amilies)                                                                                                            Income level
                                          ~_                                                               Low         Average     Maximum
                                          Replacement rate                                                54.95           40.98        27.08
                                          Implicit rate of return (Inflation-adjusted)                     4.14            2.98         1.24

                                          Implicit rates of return paid by Social Security also illustrate the
                                          formula’s progressivity. Although workers do not actually have their
                                          own Social Security accounts in which their contributions earn interest,
                                          they do implicitly earn a rate of return based on the benefits they
                                          receive relative to the contributions they made (including those made by
                                          employers on their behalf). In effect, workers “purchase” their Social
                                          Security benefits at retirement with their lifetime contributions. Table
                                          I.1 shows that these implicit rates of return are higher for low earners
                                          than for high earners.

                                          Thus, as a result of the progressive benefit formula, higher earners sub-
                                          sidize the benefit levels of lower earners by effectively receiving lower
                                          rates of return on their Social Security contributions and, at the same

                                          Page 17                                        GAO/HRl%91-22   Privatizbg   Social Security Trust Fund
                     Appendix I
                     The Beneflt Adjustment    Under the

                     time, lower replacement rates. Conversely, lower earners receive higher
                     rates of return and higher replacement rates to help them achieve ade-
                     quate retirement incomes.

                     Because the ISSRAproposal affects only retired workers’ benefits (not
                     dependents’ or survivors’ benefits), the relevant implicit rates of return
                     reflect the value of these benefits relative to the contributions for these
                     benefits. In practice, however, Social Security combines the contribu-
                     tions for retired workers’, dependents’, and survivors’ benefits. To
                     reflect the limitation to workers’ benefits, our analysis assumes that
                     contributions for dependents’ and survivors’ benefits are separate from
                     those for workers’ benefits and collected on a pay-as-you-go basis3 This
                     leaves the entire reserve accumulation in the worker benefit portion of
                     the program. The appropriate implicit rate of return reflects the value
                     of workers’ benefits alone relative to the contributions for them alone,
                     including the reserve accumulation.

                     Under the ISSRAproposal, Social Security would pay reduced benefits to
Benefit Adjustment   compensate for the contributions diverted to workers’ ISSRAS.   At the
Retains Subsidy      same time, Social Security would preserve the subsidies that workers
                     either give or receive under the current system’s progressivity. After
                     exploring several approaches, we devised a workable benefit adjustment
                     that meets these requirements and uses the concept of Social Security’s
                     implicit rate of return.

                     Because Social Security benefits reflect workers’ contributions and
                     implicit interest earnings, the benefit adjustment should also reflect
                     both the diverted contributions and the implicit interest they would
                     have earned from Social Security. The amount of the annuity that these
                     diverted contributions and interest earnings would have “purchased”
                     from Social Security would be the amount of the benefit reduction. How-
                     ever, to preserve the subsidies, the adjustment calculates the implicit
                     interest earnings using the rate of return for the age group as a whole.
                     Thus, the adjustment applies the same implicit rate of return for each
                     worker in an age group even though everyone earns different rates, as
                     discussed above.

                     3The implicit rate of return on contributions for dependents’and survivon’ benefits depends on
                     household configuration as well as income, reflecting Social Security’s redistribution of income smong
                     various household configurations.

                     Page 18                                    GAO/HRD-91-22     Privatizing   Social 8ecurlty   Trust Fund

Appendix I
The Benefit Adjustment     Under the

In effect, instead of depositing their diverted contributions with Social
Security, workers would deposit them in their private-sector ISSRA.
Figure I. 1 diagrams the combined effect of the benefit adjustment and

4To illustrate the effect of the benefit adjustment, we are assuming that people will be able to and do
purchase annuities with their I!SSRAsthat have exactly the same features as the Social Security
annuity, especially unisex pricing with cost-of-living adjustments. See appendix II for further

Page 19                                     GAO/H&D91-22      Privatizing   Social security   Trust Fund
                                           Appendix I
                                           The Jheflt  Adjustment        Under the

Figure 1.1:The AdJuated Social Security Benetlt and the ISSRA Annuity

                                                                           7-        Adjusted Final

                                   I                 Adjurted Interest                                                          r


I-                                                                                                                I


                                                     ISSRA Deposits


                                                                                      ISSRA Final     l-J
                         b       ISSRA

I                                                                                                       I         I                           1
Working Llfe                                                                                                       Retirement

                                           This benefit adjustment would give everyone an equal opportunity to
                     Y                     benefit from the ISSRA program. To be better off with the ISSRA program,
                                           everyone would have to earn rates on their ISSRASbetter than those of
                                           their age group as a whole will earn from Social Security; they would be

                                           Page 20                                        GAO/HID-91-22     Privatizing   Social Security Trust Fund
Appendix I
The Benefit Adjustment     Under the

worse off if they did not earn at least as much as the average Social
Security return for their age group.

To demonstrate that the benefit adjustment retains progressivity, con-
sider what happens if ISSRA rates of return are exactly the same as the
implicit Social Security age group rate. The correct benefit adjustment
should leave retired workers as well off as they would be under the cur-
rent program. If the benefit adjustment yields this result, the current
program’s redistribution of income must be intact.

Figure I.2 and table I.2 illustrate this hypothetical case. They show the
ISSRA proposal’s effect on retirement incomes for low, average, and high
earners.6 Under the ISSRA program, a worker’s combined Social Security
and ISSRA retirement income would be as follows:

      Total income =            Current law Social Security benefit
                                - Benefit adjustment
                                + Annuity on ISSRA.

The figure and table show that if the private-sector rate exactly
matches the benefit adjustment rate the ISSRA annuity offsets the benefit
adjustment and no one’s retirement income changes. Thus, the benefit
adjustment accomplishes the same redistribution of income that current
law provides.

6These examples are only for households with one wage earner. Households with two earners may
have two ISSRAs but may collect benefits based on only one spouse’s earnings. In this event, the
benefit reduction would have to reflect the ISSRA diversions of both wage earners to retain the cur-
rent system’s redistribution of income among various household configurations. However, as dis-
cussed earlier in this report, the benefit adjustment always applies to the benefits only of the retired
worker, not of survivors or dependents.

Page 21                                     GAO/HRD91-22       Privatizing   Social Security   Trust Pund
                                              Appendix I
                                              The Beneflt Adjuetment     Under the

Figure 1.2:Total Retirement Incomes, by
Earning8 Level, When ISSRAr Earn
Social Security Age-Qroup Rate of             2400 Monthly Retirement Income (IBBO Dollars)
Return (Workers Born in 1960)                 2200


                                                       LOW                     Avsragr                  Maximum
                                                       Earnlngr Level

                                                               ISSRA Annuity

                                              The following numerical example shows how the adjustment preserves
                                              the subsidy when

                                          l   an age group’s implicit rate of return under Social Security is 2 percent,
                                              adjusted for inflation;
                                          l   high earners implicitly earn a l-percent return from Social Security;
                                          l   low earners implicitly. earn a 3-percent return from Social Security.

                                              Using the age group rate, the benefit adjustment calculates interest on
                                              the diverted contributions as if all workers would have earned 2 percent
                                              from Social Security. In fact, high earners only would have earned 1 per-
                                              cent; thus, the adjustment takes away more interest than these earners
                                              actually would have earned, and they continue to subsidize the system.
                                              Conversely, low earners would have earned a full 3 percent; thus, the
                                              adjustment takes away less interest than these earners would have
                                              earned, and they continue to receive their full subsidy. Thus, by
                                              applying the same age group rate, even though everyone actually earns

                                              Page 22                                    GAO/HRDSl-22   Privatizhg   Social Security Trust Fund
                                          Appendix 1
                                          The Benefit Adjustment    Under the

                                          different implicit rates from Social Security, the adjustment achieves the
                                          current system’s redistribution of income.

Table 1.2: Proposal’s Impact on Retiree
Incomes When ISSRAs Earn Social                                                                                      Income level
Securl Age-Qroup Rate of Return (by       Monthly dollar results (1990 dollars)                      Low               Average            Maximum
income Y eve1 for Workers Born in 1960)
                                          Current law benefit                                  $660.91                   $I,09530          $1,730.82
                                          Benefit reductiona                                   -53.47                     -118.82           -283.87
                                          ISSRA annuity                                        -i-53.47                   +I 18.82          +283.87
                                          New retirement income                                $660.91                   $1,095.30         $1,730.82
                                          Percentage results
                                          Benefit reduction                                          8.09%                   10.85%F           16.40%
                                          Change in income                                           0.00%                    0.00%             0.00%
                                          GRA as a percent of income                                 8.09%                   10.85%            16.40%
                                          aAssumes inflation-adjusted age group rate of 2 percent.

                                          As an alternative, the benefit adjustment could use the interest rate
An Alternative to the                     earned on the Social Security Trust Fund in calculating interest on the
Age Group Rate                            diverted contributions. This alternative would have the same distribu-
                                          tional properties as those discussed above. If ISSRAS were to earn exactly
                                          the Trust Fund rate, the ISSRA annuity would exactly offset the benefit
                                          adjustment; retirement incomes would not change.

                                          Under this alternative, however, if the Treasury rates differ signifi-
                                          cantly from an age group’s implicit rate of return on Social Security,
                                          there would also be a cost effect to the system. If the Treasury rate is
                                          greater than the age group rate, benefits on average will be reduced too
                                          much and the Social Security system will have lower benefit costs.
                                          Alternatively, if the Treasury rate is less than the age group rate, bene-
                                          fits will be adjusted too little and the system will incur higher costs.

                                          Page 23                                    GAO/HID91-22          Privatizing     Social Security Tmat Fund
Aooendix II

l?roposal’sAdvantage to Individuals Depends
on Market Interest Rate

                      Whether the Individual Social Security Retirement Account proposal
                      will improve retirement incomes depends entirely on whether the
                      investments made by individuals yield higher rates of return than the
                      implicit rate of return to Social Security. More specifically, if everyone
                      under the ISSRA program earns a higher rate of return in the private
                      sector than the rate used in the benefit reduction (that is, the age
                      group’s rate), then the ISSRAannuity will more than offset the benefit
                      reduction. As a result, ISSRAS will contribute to higher retirement income.
                      ISSFUS could also yield lower rates of return, however, resulting in lower
                      retirement incomes compared with current law Social Security benefits.

                      For this analysis, we use estimates of the age group return ranging from
                      more than 3 percent for those born in 1930 to 2 percent for those born in
                      1960 or later.1 The rate of return varies by age group because of
                      changing tax rates, benefit rules, and actuarial factors that reflect mor-
                      tality rates. We focus our analysis on the difference between private
                      market rates and the age group rates, as detailed below.

                      To analyze how individuals might fare with ISSRAS, we examined rates of
Historical Rates of   return earned over the last 70 years on typical investments that would
Return Depend on      be available to retirees.2 As a rule, the market offers higher rates of
Risk and Inflation    interest on riskier investments. Rates of return will, therefore, depend
                      on how retirees invest their ISSRAS. If the ISSRA proposal precludes riskier
                      investments to enhance the security of private retirement income, it also
                      limits the gains individuals can expect from the program (see app. III).

                      Figure II. 1 shows the compound annual rates of return3 for various
                      types of investments since 1957,4adjusted for inflation. As government
                      securities are traditionally considered the least risky investment, their
                      returns are historically the lowest. Annual returns on private securities
                      are more volatile than those on Treasury securities. However, on

                      ‘Our age group rate estimates are broadly consistent with estimates in the academic literature. See
                      Michael D. Hurd and John B. Shoven, “The Distributional Impact of Social Security,” in Pensions,
                      Labor, and Individual Choice, David A. Wise, ed., National Bureau of Economic Research, University
                      of Chicago Press, 1986, pp. 193-221. Comparable results are found in, M. Ho&in, L. Kotlikoff, D.
                      Puffer-t, and J. Shoven, Social Security: A Financial Appraisal Across and Within Generations,”
                      National Tax Journal, Vol. XL, No. 1, March 1987, pp. 19-34.
                      2Data on historical rates of return are from Roger G. Ibbotson and Rex A. Sinquefield, Stocks, Elands,
                      Bills, and Inflation. Chicago: Ibbotson Associates, 1990.
                      3Compound annual rates reflect the return on an investment over a number of years, figured on a
                      constant annual bash this is not the same as the arithmetic average of rates for each year.

                      4The first year for which data are available on the Standard & Poor 600 stocks is 1967.

                      Page 24                                    GAO/HlUb91-22     Privatizbg   Social Security Trust Fund
                                       Appendix II
                                       P~~pod’e Advantage to IndivlduaJe       Dependa
                                       on Market Interest Rate

                                       average, the market compensates for this greater risk by offering higher
                                       returns on investment in private securities.

Flgure 11.1:Compound Annual Rates of
Return, Inflation-Adjusted (1957-89)
                                       6   PmontRoturn

                                       Note: Data for Standard and Poor 500 before 1957 are not available.
                                       Source: Roger G. lbbotson and Rex A. Sinquefield, Stocks, Bonds, Bills, and Inflation. Chicago: lbbotson
                                       Associates, 1990.

                                       Figures II.2 and II.3 show the compound annual rates of return by
                                       decade for stocks and corporate bonds, in both nominal and inflation-
                                       adjusted terms. Nominal rates reflect the actual dollar payout of an
                                       investment. Because inflation reduces the v’alue of those dollars, how-
                                       ever, we focus on inflation-adjusted rates to reflect the payout in terms
                                       of real purchasing power.

                                       Page 25                                     GAO/HRD-91-22     Privatizing   Social Security   Trust Fund
                                      Appendix II
                                      Proposal’s Advantage to Individuala     Depends
                                      on Market Interest Rate

Figure 11.2:Compound Annual Returns
to; the Decades; Common Stocks
                                      22 Puoanl Rotum
( 1926-89)

                                           L-l      Inflation-Adjusted

                                      Note: 1920s data begin with 1926.

                                      Data before 1957 are for Standard and Poor 90; after 1957, they are for Standard and Poor 500.
                                      Source: Roger G. lbbotson and Rex A. Sinquefield, Stocks, Bonds, Bills, and Inflation. Chicago: lbbotson
                                      Associates, 1990.

                                      Page 26                                     GAO/HRBBl-22       Privathhg   Social Security Trust Fund

                                       Appendix II
                                       Propoeah Advantage to lndividunls       Depends
                                       on Market Interest Rate

Figure 11.3:Compound Annual Returns
for the Decades, Long-Term Corporate   22   Puwnt   Return
Bonds (1926-89)


                                       Note: 1920s data begin with 1926.
                                       Source: ROgar G. lbbotson and Rex A. Sinquefield, Stocks, Bonds, Bills, and Inflation. Chicago: lbbotson
                                       Associates, 1990.

                                       As the figures demonstrate, rates of return fluctuated considerably over
                                       time. While the experience of the 1980s suggests that ISSRASmight per-
                                       form very well, the figures also show that, historically, these rates are
                                       atypically high. They cannot necessarily be expected to persist over sev-
                                       eral decades.

                                        Ultimately, workers will hold ISSRAS  over different years and for dif-
Returns Vary Widely                    ferent lengths of time. For instance, some people will retire just a few
Over Rolling 20-Year                   years after their first ISSRA deposit. Others, who enter the work force
Periods -                              just as the excess revenues diminish around 2016, will retire 40 or more
                                       years after their last ISSRA deposit. Roughly one-fourth of the ISSRAS will
                                       collect interest for no more than 26 years.

                                       A retiree’s advantage from the program will depend entirely on the
                                       returns earned by the individual ISSRAS from diversion to retirement. As
                                       a result, this advantage may vary considerably from worker to worker,

                                       Page 27                                     GAO/HRDBl-22      Privatizhg   Social Security Trust Fund
                                                Appendix II
                                                Froposal’s Advantage to Individuala     Depends
                                                on Market Interest Rate

                                                even for those holding accounts over periods just a few years apart.
                                                Figure II.4 shows that inflation-adjusted compound annual returns vary
                                                widely over rolling 20-year periods. Specifically, since 1957, they have
                                                ranged from about 1 to over 5 percent for Standard & Poor 500 stocks
                                                and from about -2.7 percent to roughly 3.2 percent for corporate bonds.

Figure 11.4:Compound Annual Returns, Inflation-Adjusted,         for 20-Year Holding Periods

 1026          1930             lQS5          1940            1945             1OW                1965            1980              lQ65             1970
 Flmt Year ot Rolling 20-Yom Holding Period

     -        U.S. Treasury Bill8
     l ml n   Long-Term Corporate Bond8
     m        CommonStock
     m--m     SmallCompanyStock8

                                                Note: Data for common stocks before 1957 are for Standard and Poor 90; after 1957, they are for Stan.
                                                dard and Poor 500.
                                                Source: Roger G. lbbotson and Rex A. Sinquefield, Stocks, Bonds, Bills, and Inflation. Chicago: lbbotson
                                                Associates, 1990.

                                                Such wide variation could have significant effects on the income of indi-
                                                viduals retiring in slightly different years. For example, if an ISSRA
                                                earned interest at the rate the Standard & Poor 500 earned from 1967 to
                                                1986, it would earn 3.7 percent; but, another worker’s ISSRA that earned
                                                interest at the rate from 2 years earlier, 1965 to 1984, would earn only
                                                1.4 percent. For corporate bonds, an ISSRA earning interest at the rate
                                                from 1970 to 1989 would earn 3.2 percent; but, another ISSFUearning at
                                                the rate from just 5 years earlier, 1965 to 1984, would earn -1 percent.

                                                Page 28                                     GAO/HlZD91-22      privatizing   Social Security Trust Fund
                                          Appendix II
                                          Proporurl’s Advantage to Individunla    Depends
                                          on Market Interest Rate

                                          We explored different assumptions concerning the market rate of
Effect of ISSRA Plan                      return. If ISSRASearn a moderate return, like that available historically
on Typical                                for corporate bonds, small improvements in total retirement income are
Ekneficiaries                             likely. If ISSRAS
                                                          could obtain returns comparable to those historically
                                          earned in the stock market, substantial improvement in workers’ retire-
                                          ment incomes is possible. Conversely, holding ISSRASduring an unlucky
                                          period, such as has occurred from time to time for both stocks and
                                          bonds, could actually reduce total retirement income compared with
                                          that available from Social Security. In addition, contributing less than
                                          the maximum number of years (26) moderates the effect of the plan on
                                          retirement incomes.

Results Vary by Market                    Using various assumptions about the rate of interest ISSRAS   earn, figure
                                          11.6compares the Social Security benefit under current law with the
Rate of Return                            combined income from the ISSRA proposal’s adjusted benefit and ISSRA
                                          annuity for an average earner born in 1960. Table II.1 provides numer-
                                          ical results, including the percentage increase in retirement income and
                                          the ISSRAannuity as a percent of total retirement income.

Table 11.1:ISSRA Proposal’s Impact on
Retiree Incomes by Market Interest Rate                                                  Interest rate assumption (inflation-adjusted)
Assumptiona (Average Steady Earners       Month1 dollar resultsb                          Pessimistic        Moderate          Optimistic
Born in 1960)                              (199idollars)                                         (1%)             (3%)               (7%)
                                          Current law benefit                                $1,095.30               $1,095.30          $1,095.30
                                          Benefit reduction                                   -118.82                 -118.82            -118.82
                                          ISSRA annuity                                        + 94.18                +148.05            +377.58
                                          New retirement income                              $1,070.66               $1,124.53          $1,354.06
                                          Percentage results                                                                             .___
                                          Benefit reduction                                       10.85%                 10.85%             10.85%
                                          Change in income                                       -2.25%                   2.67%             23.63%
                                          ISSRA as a percent of income                             8.80%                 13.17%             27.89%
                                          aAssumes inflation-adjusted age group rate of 2 percent under Social Security.
                                          bNumbers may not add due to rounding.

                                          Page 29                                    GAO/IiRD91-22       Privatkhg     Social Security Trust Fund
                                          Appendix II
                                          Proposal’s Advantage to Individuals        Depends
                                          on Market Interest Rate

Figure 11.5:Total Retirement Income& by
Market Intereet Rate (Average Steady
                                          1400      Monthly Rollromrnt Incomo (1990 Dollrrr)
Earners, Born in 1960)

                                                 Currant Law    ISSRA Proposal ’

                                                            ISSRA Annuity
                                                            Adjusted Benefit
                                                            Current Law Benefit

                                          %ars show alternative market interest rate assumptions. Benefit adjustment assumes inflation-adjusted
                                          age-group rate of return of 2 percent.

                                          For our moderate case, we assumed that ISXAS would earn an inflation-
                                          adjusted 3 percent, which is 1 percentage point more than the age-group
                                          rate of 2 percent.6 In nominal terms, this translates to roughly 7 percent,
                                          using the Alternative II-B assumption of 4-percent inflation. This case
                                          approximates the historical performance of highly graded corporate
                                          bonds, which have earned 2 percent since 1960, and 4 percent during
                                          the unusually good bond market during the 1980s. Under this assump-
                                          tion, the ISSRA proposal increases retirees’ income by 2.7 percent, and the
                                          IssR4annuity constitutes 13 percent of total retirement income.

                                          ‘As noted earlier, these private rates of return are before taxes and are not adjusted for administra-
                                          tive fees. Such adjustments would reduce the effective rate of return and the ISSRA annuity.

                                          Page 30                                      GAO/HRD91-22     Privatizing   Social Security Trust Fund
I           .

                                                             Appendix II
                                                             Proposal’s Advantage to Individuals     Depends
                                                             on Market Interest Rate

                                                             For our optim istic case, we assumed that the ISSRAS would earn 6 per-
                                                             centage points more than the age group rate, or an inflation-adjusted
                                                             rate of 7 percent. This rate approximately reflects the performance of
                                                             the Standard and Poor 600 stocks since 1967. For this assumption, the
                                                             figure and table show that the ISSRAproposal results in a percentage
                                                             increase in retirement income of 24 percent. The ISSRAannuity repre-
                                                             sents 28 percent of total retirement income.

                                                             For our pessimistic case, we assumed that the IBRASwould earn 1 per-
                                                             centage point less than the age group rate, or 1 percent. We selected this
                                                             spread to illustrate what could happen if a particular age group holds
                                                             ISSRAaccounts over an unlucky period; the above discussion of historical
                                                             returns over rolling 20-year periods suggests that such unlucky periods
                                                             m ight well occur for some age groups. An inflation-adjusted rate of
                                                             slightly better than 3 percent reflects the performance for stocks over
                                                             the worst 20-year period since 1967; but for corporate bonds, a rate of
                                                             -1 percent reflects the worst such period. For this pessimistic assump-
                                                             tion, figure II.6 and table 11.1show that the ISSRAproposal results in a
                                                             decrease in retirement income of 2.3 percent. Here, the ISSRAannuity
                                                             represents 9 percent of total retirement income but would not make up
                                                             for the benefit reduction.

Table 11.2:ISSRA Proposal’s Impact on Retiree Incomes Using Moderate Market Rate Assumptlon (by Age Group for Average
Steadv Earners)
                                                                                                   Workers born in
Age - group
_...__         attributes
       -_-- .--_-    _ ..- .-..-~-. --.-              1930               1940           1950              1960           1970             1960             1990
Number of years with ISSRA
    .-... ..-.. _-_ .-..       -~---..-___.-             5          --      16             26                  26           26               16                6
Inflation-adjusted age group return
     .._l-l-. - _._..
                   -- . _.- .__..__..
                                 ____--...-           3.51%               2.64%          2.32%             2.00%          2.00%            2.00%            2.00%
Monthly dollar results.-- -.-_~.-
                           (1990 dollars)
Current -._law
i.______...       benefit
             ..-..--  - .-~-.                   $744.31                $848.20      $960.30           $1,095.30      $1,240.29       $1,404.34        $1,590.14
Benefit reduction                                -13.43                 -51.95       -98.23            -118.82        -142.12         -100.71            -43.56
ISSRA annuity                                    +13.59                 +53.84      +106.80            +148.05        +193.06         +140.73            -t-63.44
        -.__. -..--._ _Income                   $744.47
                          - -- .- . -.- __... ---~                     $850.09      $968.88           $1,124.53      $1,291.23       $1,444.36        $1,610.02
Percentage        results
..-_-._ -.- ---- __.....--- .-..-.-.-...--_____~.-.
__.-. -_reduction
           ...-.. ___._    _..._- ~-                  1.80%               6.12%         10.23%            10.85%         11.46%            7.17%            2.74%
Change    in income
-____.-.--_-   -..- -..----          --               0.02%               0.22%          0.89%             2.67%          4.11%            2.85%            1.25%
ISSRA as percent of income                            1.83%               6.33%         11.02%            13.17%         14.95%            9.74%            3.94%
                                                             aNumbers may not add due to rounding.

                                                             Page 31                                    GAO/HRDBl-22    Privathing   Social Security Trust Fund
                                     Appendix II
                                     Pmpod’s   Advantage to Individuals     Demmds
                                     on Market Interest Rate

Results Vary by Age                  After about 2016, there will be no more excess revenues to divert to the
                                     ISSRAS. Table II.2 shows that, as a result, the number of years that any
Group                                individual will contribute to an ISSRA will vary. Workers very near
                                     retirement when the diversions begin might contribute only 1 or 2 years
                                     to the ISSRA.Figure II.6 shows that 37 percent of the eligible workers
                                     would contribute the full 26 years, specifically those born between 1960
                                     and 1970. Those entering the work force after the diversions begin
                                     would contribute for fewer years, but their ISSRMwould accumulate
                                     interest for a long period until their retirement. As a result, the ISSRA
                                     program phases out over a very long period of time, many years after
                                     the last diversion. For example, a newly entering worker in 2016 would
                                     contribute only 1 year, under our simplifying assumptions, but the
                                     (small) ISSRAwould be held and accumulate interest for a working life of
                                     46 or more years.

Figure 11.6:Workers Wlth ISSRAs by
Number of Year5 Wlth Deposits                                                               26 years

                                                                                            l-5 years

                                                                                            6-l 0 years

                                                                                     -      1 l-l 5 years

                                                                                 -          16-20 years

                                                                                            21-25 years
                                     Note: Based on simplifying assumption that everyone retires at normal retirement age

                                     Table II.2 also shows that the inflation-adjusted age group rate of return
                                     on Social Security used in our benefit adjustment ranges from more than

                                     Page 32                                    GAO/HID-91-22     Privatizing   Social Security Trust Fund
Appendix II
Propods   Advantage to Indh-iduale     Depends
on Market Interest Rate

3 percent for those born in 1930 to 2 percent for those born after 1960.6
These rates of return imply that the impact of the ISSRA proposal on indi-
vidual retirement incomes would vary considerably by age group.

Figure II.7 and table II.2 show this variation using our moderate
assumption that ISSRAS will earn a 3-percent inflation-adjusted interest
rate. Age groups born before 1960 do no better under the proposal
because their benefits are reduced at up to a 3-percent rate for their age
group rate of return. Later groups do better because their benefits are
reduced at a lower age group rate, but the size of their benefits decline
because they contribute for fewer and fewer years.’

‘The implicit rates of return vary by age group because(1) payroll taxes have increased, so later age
groups pay relatively more during their careers; (2) benefit levels and eligibility requirements have
changed; and (3) the actuarial value, or “price,” of these benefits changes as mortality rates and
interest rates change.

7This analysis does not reflect the additional risk that earlier age groups run becausethey have held
their ISSRAs over shorter periods. As discussed earlier, the combined returns earned over the life
span of the ISSRAs ultimately determine the impact on an individual’s income. For example, those
retiring ln the year 2000 will have been holding their ISSRAs for 10 years, but compound annual rates
of return over lo-year holding periods fluctuate more than the 20-year perlods examlned above.
Thus, this age group runs a greater risk of holding their ISRAs over an unlucky period than those
retiring ln later periods.

Page 33                                   GAO/HRD91-22      Privatizing   Social Security   Trust F’und
                                               Appendix II
                                               PropmaI’s Advantage to Individualsbepends
                                               on Market Interest Rate

Figure 11.7:Total Retirement Incomer, by Age Group, Ushg Moderate Market Interest Rate Assumptions
1900 Monthly Retlromwt lncomr (1990 Dollars)



       Birth Yur

       1     1 ISSRA Annuity
                   Adjurrted Benefit
                   Current Law Benefit

                                               Note: Benefit adjustment assumes inflation-adjusted age group rate of return of 2 percent

Objectives, Scope,and                          Our objective was to analyze how hypothetical Social Security benefi-
Methodology                                    ciaries (retired workers only) might fare under the ISSRA plan. As dis-
                                               cussed in appendix I, under the ISSRA program a person’s total
                                               retirement income would be comprised of the current law Social Security
                                               benefit minus the benefit adjustment plus the annuity on the ISSRA. Our
                                               analysis illustrates the change in retirees’ income under the ISSRA pro-
                                               gram by calculating these amounts using various assumptions about the
                                               market rate of interest and other factors.

                                               For our analysis, we set the amount of the annual ISSRAdeposits at 2
                                               percent of taxable payroll for the 26-year period from 1990 through
                                                2016. This is about the average amount of excess revenues in the por-
                                               tion of Social Security that represents retired worker benefits up until
                                                2016. At about that time, according to the current Alternative II-B pro-
                                               jections, the annual excess payroll tax will begin to decline rapidly. Both
                                               the size of the excess revenues and the exact time period over which

                                               Page 34                                     GAO/HRD91-22      Privatizing   Social Security Trust Pund
1   II

         Appendix II
         Propods   Advantage to Individual       Depends
         on Market Interest Rate

         they occur could change considerably with future economic and demo-
         graphic conditions.

         Next, for our hypothetical workers we used “steady earner” cases in
         which workers start their career at age 20 and have the same relative
         inflation-adjusted earnings level throughout their lives8 To illustrate the
         ISSRAplan’s impact by income level, we used the three hypothetical
         steady earners born in 1960 (see app. I). Workers born in 1960 fall in
         the m iddle of the age groups affected by the ISSRAproposal. Because
         they will work approximately from 1980 to 2027, they will be able to
         participate in all 26 years of the ISSRA diversions.

         For each of our hypothetical workers we (1) generated the current law
         benefit (PIA) based on the Alternative II-B assumptions, (2) constructed
         ISSRAS that reflect the diverted contributions and the accumulated
         interest at various specified rates, and (3) arrived at the benefit adjust-
         ment (see app. I) and ISSRA annuities by converting final balances into
         annuities that have features identical to the Social Security benefit.O

         As noted earlier, for interest on the ISSRAS, we used moderate, optim istic,
         and pessimistic rate-of-return assumptions. These roughly reflect the
         historical performance of different types of private market investments
         while also taking into account the alternative II-B assumptions about
         future economic performance. For the implicit Social Security age group
         returns, we used estimates that ranged from 2 to more than 3 percent,
         depending on the age group.

         *Analysts often use these steady earner cases,which reflect the range of Social Security benefit
         levels. They are not necessarily “typical,” however, because many workers experience employment
         interruptions and greater variation in earnings during their careers.
         ‘To simplify our comparison of the ISSRA with the current system, we assumed that the entire ISSRA
         balance is converted to an annuity that has features identical to the Social Security benefit. However,
         the Social Security benefit has features that may not be available in the private market. As a result,
         ISSRA annuities may be more or less generous than our analysis illustrates.
         Specifically, the private market does not in general offer unisex benefits or benefits with cost-of-
         living adjustments. Further, private annuities cost over 30 percent more, on average, than their actu-
         arially fair value, reflecting profits, administrative costs, and insurance risk, which would make the
         potential ISSRA annuities less generous. On the other hand, our analysis prices annuities using the
         interest rate paid on the Social Security Trust Fund while the private market could offer more gen-
         erous annuities using higher rates of interest on other investments.

         Thus, comparing the values of annuities in the private and public sector presents numerous problems
         of interpretation and valuation. For consistency and comparability, we have priced all annuities ln
         our analysis using features identical to those a Social Security benefit offers, including private-sector
         annuities on the ISSRA balances. In essence,we have circumvented all the above issues by sssumlng,
         for the purposes of this illustration alone, that everyone ultimately purchases annuities on their
         ISSRAs from Social Security.

         Page 36                                     GAO/HRD-91-22      Privatkdng   SocM Security Trust Fund
Appendix III

Chwiderationsin hnplementingan
rssm Program

                       The ISSRA proposal raises many issues concerning program design, struc-
                       ture, and implementation. Many operational and administrative chal-
                       lenges would have to be overcome before the ISSRA program could be
                       considered a feasible alternative use of the Trust Fund reserves. In this
                       section we highlight key aspects regarding Treasury transfer of funds
                       and the management and investment of ISSRAS. Also, because of the
                       potential for lower retirement income, the proposal raises questions that
                       may need to be resolved through the political process.

                       The ISSRA program would represent a large transfer of government funds
Treasury Transfer of   to private-sector financial institutions. Fully implemented, approxi-
Funds to ISSRAs        mately 24 percent of annual payroll taxes for the retirement portion of
                       Old-Age, Survivors and Disability Insurance (OASDI) would flow into pri-
                       vate ISSRAS held by various financial institutions. In 1990 this would be
                       approximately $48 billion.

                       Payroll taxes could be paid and flow into the Treasury as they now do,
                       where they would be credited to the Trust Fund. However, unlike the
                       current situation where the Trust Fund accumulates large amounts of
                       Treasury securities, the Trust Fund accounts would be reduced as pay-
                       ments are made to financial institutions under the ISSRA program.

                       The ISSRA program would involve the creation of approximately 130 mil-
                       lion separate accounts in the name of each worker covered under Social
                       Security. These accounts would be held at a large number of financial
                       institutions throughout the country. Presumably, individuals would
                       have the opportunity to select the financial institution at which their
                       account would be held. This could be done by completing a form and
                       designating a direct deposit type of transfer. At least two concerns that
                       arise are determining, at regular intervals, how much should be sent for
                       each individual and the manner and frequency with which Treasury
                       would forward these funds.

                       Payroll taxes are transmitted by employers directly to the Treasury
                       throughout the year. Employers file quarterly tax reports with the
                       Internal Revenue Service (IRS) where they report total employees’earn-
                       ings and pay taxes due. Self-employed workers also report their earn-
                       ings to the IRS.

                       Separately, on an annual basis, employers report summaries of workers
                       earnings directly to SSA, and these data are used to credit the proper

                       Page 36                        GAO/HRD-91-22   Privathing   Social Security Trust Fund
Appendix l.U
&won&Ie;~rain       Implementing   an

earnings amount to workers’ Social Security earnings records. IRS pro-
vides data from the tax returns of self-employed persons, and SSA uses
these data to credit their earnings records. These earnings, however, are
not credited immediately to workers’ earnings records. Currently, com-
pleting the posting of earnings requires about 7 months. This task of
posting earnings would have to be completed before SSAcould determine
the amount of ISSRA contribution for each worker and forward this infor-
mation to Treasury. This could mean a delay of 8 months or more before
ISSFLA payments are received and credited to individual accounts by
private-sector financial institutions.’

Even if the determination of the amount of ISSRA contribution could be
done in a manner that reduces the lag between payment of taxes and
crediting to ISSRA accounts, there may be some difficulty in the Treasury
forwarding these funds. Treasury would have to sum up the individual
contributions to each financial institution and then forward these funds
accordingly. Although this appears to be a straightforward task, experts
tell us that it is not and Treasury would encounter difficulty particu-
larly if a large number of institutions receive ISSRAfunds. Also, there
could be cash management considerations because it is likely that
transfer of funds to ISSRAS might not occur regularly throughout the
year, but in one large annual payment.

One way to avoid problems associated with Treasury payments to ISSRAS
is to have employers transmit part of the workers’ payroll taxes to des-
ignated financial institutions directly. However, this moves away from
the idea of allocating the Trust Fund reserves to private accounts
because the ISSRA program would then become a kind of mandatory indi-
vidual retirement account (IRA) system. This would also impose an addi-
tional administrative burden on private firms.

‘Also there have been difficulties in reconciling reports of employers to the Treasury via tax-return
data with data received by SSA via annual earnings reports. These difficulties would likely carry over
to allocation of the ISSRAs. While this is a problem that is solvable administratively, some effort and
resources would have to be expended to deal with it. See,Social Security: More Must He Done to
Credit Earnings to Individuals’ Accounts (GAO/HRD-87-62, Sept. 1987); Social Security: Resolving
                           (GA--l             1, Oct. 1989); Social Security: Alternative Wage Reporting
                          -36, Nov. 1989); and Social Security: IRS Data Can Help SSA Credit More
mAO/HRD-90-112,              Aug. 1990).
Another possible difficulty arises for those who, for whatever reason, do not designate an institution
to receive their funds. A special account would have to be set up to hold these funds and provision
would have to be made to determine how to allocate them.

Further complications might arise because many workers have multiple employers requiring reconcil-
iation of data to determine the ISSRA contribution, and many ISSRA diversion amounts might be very
small for part-time/part-year workers.

Page 37                                     GAO/HRD91-22      Privatizing   Social Security   Trust Fund
                      While administrative considerations such as these are not insurmount-
                      able, indications are that an ISSRAprogram would require developing
                      new administrative systems, in both the public and private sectors or at
                      least augmenting existing ones and dealing with numerous details and

                      The ISSRAproposal raises questions concerning the rules under which
Management and        funds would be invested, the investment performance of the accounts,
Investment of IsSWs   and the security of these investments if ISSRAS are to achieve their pur-
                      pose of providing a~net addition to individuals’ retirement income. Some
                      of these issues raise questions that may need to be resolved through the
                      political process.

                      While the development of a system of ISSRAS   in the private sector does
                      not appear to present any serious difficulty, there are potential compli-
                      cations that would need to be addressed. Banks and other financial insti-
                      tutions open up investment accounts for individuals on a regular basis.
                      Individuals would designate the institution to hold their accounts and
                      administrative provisions could be made to allow them to change insti-
                      tutions as appropriate. However, under the ISSRAplan the nature of
                      these accounts would likely be more complex. Furthermore, assuring
                      that these funds are managed and invested properly may require some
                      degree of restrictions, government regulation, and oversight affecting
                      the accounts.

                      One such provision mentioned in framing the ISSRAproposal is that
                      financial institutions holding ISSRAS would be required to act as fiducia-
                      ries in managing the accounts.2 This would not be unusual for many
                      financial institutions that already function in this capacity. However, it
                      could affect the desirability of handling ISSRAS for some financial institu-
                      tions and possibly the cost of managing these accounts.

                      2A fiduciary is defined as “an individual, corporation, or association, such as a bank, to whom certain
                      property is given to hold in trust, according to an applicable trust agreement. The property is to be
                      utilized or lnvested for the benefit of the property owner to the best ability of the fiduciary. Adminis-
                      trators and executors of estates and trustees of organizations are common examples of fiduciaries.
                      Investment of trust funds are usually rest&ted by law.” Bee Jerry M. Rosenberg,Dictionary of
                      Banking and Finance, John Wiley and Sons, New York, 1982, p. 224.
                      Also, under the Employee Retirement Income Security Act various rules am applied to fiduciaries of
                      pension plans. Civil penalties may be applied to individuals who violate the fiduciary requirements of
                      the act. See Dan M. McGill, Fundamentals of Private Pensions, 6th ed., Richard D. Irwin, Inc., Home-
                      wood, Illinois, pp. 47-66.

                      Page 38                                     GAO/H&D91-22       Prlvatizlng   Social Security Trust Fund
cOnrlder8tiomi   in Implementing   an

Other provisions aimed at reducing market risk could be introduced
through legislation to limit the types of financial assets in which ISSRAS
could be invested. For example, investment in certain types of equities
or bonds might be precluded. This could improve the security of ISSRA
benefits but also reduce the potential of ISSFZAS to improve retirement
income through higher market returns.

Even with restrictions on @WAS,some individuals could be worse off
than they are under the existing system. Higher earners might be in a
better position financially to assume the risk of loss from poor invest-
ment performance. Those at lower earning levels, however, are likely to
experience substantial reductions in their living standard if investments
turn out poorly. This would erode retirement benefits and could be
viewed as weakening the current provision of a minimum level of retire-
ment income for all retirees through Social Security.

Poor investment performance could create constituent pressures to
establish a federal guarantee that no one would be less well off with
ISSRAS. But a guarantee would create incentives for investors to take
risks they would not assume without the guarantee, since they would be
compensated for any losses. This would drive up the government’s costs
and could result in a transfer of income from the working population to

Page 39                                 GAO/HR.DVl-22   Privatizing   Sodal Security Trust Fund
                                                                                            .       .’
Appendix IV

Major Contributorsto This Report

Human Resources   Kenneth J. Bombara, Senior Economist
Division,         Kenneth C. Stockbridge, Evaluator
Washin&on, DC.

(103632)          Page 40                      GAO/HlWVl-22   Privatizing   Social Security Trust Fund
Ordering   Information

TIIP first five copies of each GAO report are free. Additional copies
are $2 each. Orders should be sent to the following address, accom-
panied by a check or money order made out to the Superintendent
of Documents, when necessary. Orders for 100 or more copies t,o be
mailevi to :L single address are discounted 25 percelrt.

IIS General Accounting Office
PA). Hox 6016
(;aithersburg, MI1 20877

Ordthrs may also be placed by calling (202) 2756241.
              -__-         ----wl-                               -

                                          First-Class M ail
                                        Podage (It Fws Paid
                                     L-L  Perm  it No. GlOO          I
Of’fic+il Ihwirhtw
Pc*milly f‘or Privz1t.thIIsth !HOO