Executive Summary

The President’s Commission to StrengthenSocial Security was appointed in May 2001

to formulate proposals that would protect bene-fits for today’s retirees; enhance Social Security’sfiscal sustainability for the long term; and giveyounger workers the opportunity to invest part oftheir payroll taxes in personal retirementaccounts that they would own, control, and beable to pass on to their children. The commis-sion’s three reform proposals, delivered to thepresident in December, fulfill those obligations.

The commission’s interim report, issued inAugust 2001, cast doubt on the current system’strust fund financing and questioned the pro-gram’s progressivity. Those findings generatedconsiderable public controversy. The commis-sion’s final report, which put forward three dis-tinct plans to strengthen Social Security throughpersonal accounts, generated even more debate.

Although the commission’s three plans covera spectrum of approaches, the proposals haveimportant characteristics in common. All threeplans would provide higher benefits than the cur-rent system can pay, and lower-income work-ers—who opponents of private accounts claim tobe most concerned about—would receive higher

benefits than are promised under the current sys-tem. Moreover, all three plans would producethose benefits at a cost lower than that of main-taining the current program.

The commission attracted significant criti-cism from opponents of personal accounts.What the commission’s work did not attractwas substantive counterproposals on how tokeep Social Security solvent and sustainableover the long term in the absence of personalaccounts. The next stage of the Social Securitydebate is for account opponents to put their ownproposals on the table. Inaction, the “policy”most often put forward by opponents, is not aviable option.

A review of the arguments and evidence findsthat the personal account-based proposals fromthe President’s Commission provide a better wayto pre-fund future Social Security benefits thanthe current program’s trust fund mechanism; thatprotections against poverty in old age would beincreased and progressivity enhanced; that work-ers would have the right to own, control, andpass on their Social Security savings; and thatpersonal account-based proposals have thecapacity to pay higher benefits at lower long-runcosts than the traditional pay-as-you-go methodof financing Social Security.

August 22, 2002 SSP No. 27

Perspectives on the President’s Commissionto Strengthen Social Security

by Andrew G. Biggs

Andrew G. Biggs is a Social Security analyst and assistant director of the Cato Institute’s Project on SocialSecurity Choice. Biggs served as a staff member on the President’s Commission to Strengthen Social Securityfrom May 2001 to January 2002.

Introduction

In May 2001 President Bush appointed thePresident’s Commission to Strengthen SocialSecurity to formulate proposals that wouldmaintain Social Security’s promise for today’sretirees while improving that promise foryounger workers through personal accounts.That was their task, and in the end they accom-plished it well.

The commission began its work with an inter-im report, issued in August 2001, outlining thestate of the current program. The interim reportgenerated significant controversy—particularlyits criticism of the Social Security trust fund andthe overall progressivity of the program.

The commission’s final report and recom-mendations, delivered to the president inDecember 2001, contains three separate reformproposals based on personal retirementaccounts. Although the plans encompass abroad range of ideas on how to maintain SocialSecurity, each would pay benefits at least ashigh as the current program at a lower long-term cost, while giving workers the opportuni-ty to build assets and wealth in personalaccounts that they would own and control.

The commission’s Plan 1 would do nothingmore than give workers the option to voluntarilyinvest a portion of their Social Security payrolltaxes in a personal retirement account. Because itmakes no other changes to the system, it is politi-cally attractive in the short term, but it does notaddress long-term concerns. Nevertheless, eventhis “accounts only” approach would pay higherbenefits to all retirees while reducing long-termgeneral revenue costs by 8 percent compared withthe current program.

Plan 2 would go further, by allowing work-ers to voluntarily invest 4 percent of theirwages up to $1,000 while indexing traditionalbenefits for new retirees to increases in pricesrather than wages. This step would make SocialSecurity sustainable indefinitely, reducing thelong-term general revenue costs of supportingthe program by 68 percent while paying retireeshigher benefits than under current law.

Plan 3 incorporates a combination add-onand carve-out account, wherein a worker whovoluntarily invests an additional 1 percent ofhis wages may redirect 2.5 percentage points ofhis payroll taxes, up to $1,000 annually. Plan 3would pay benefits higher even than those

promised by Social Security while putting 52percent less pressure on general revenues thanthe current program. More problematic is thefact that Plan 3 incorporates new, ongoing gen-eral revenue transfers to the traditional pay-as-you-go program. Although these fundingincreases are consistent with the desire of theplan’s sponsors for workers to continue toreceive a combination of defined benefit anddefined contribution benefits that exceed levelspromised by the current program, it is believedthat revenues are better applied to establishingthe funded portion of Social Security reformrather than bolstering the existing pay-as-you-go element.

The latter two plans incorporated significantnew protections for the most vulnerableAmericans. Both plans would guarantee 30-year minimum-wage workers a retirementabove the poverty line, a promise the currentprogram cannot make. This guarantee wouldlift up to one million retirees out of poverty by2018. Survivors’ benefits for lower-wage indi-viduals would be increased to 75 percent of thecouple’s prior benefit, increasing benefits fortwo to three million retired women.

The commission’s plans would also assistdivorced persons, who for the first time wouldgain a right to benefits on the basis of theirspouses’ earnings even if they divorced before10 years of marriage. Coupled with the pro-gressive funding of the personal accounts inPlans 2 and 3, these steps make reform basedon personal accounts unequivocally beneficialto lower-income Americans.

The commission attracted considerable criti-cism from opponents of personal accounts.What the commission’s work did not attractwas substantive counterproposals on how tokeep Social Security solvent and sustainableover the long term in the absence of personalaccounts. The next stage of the Social Securitydebate is for account opponents to make theircase and for the public and policymakers todecide what they want. Inaction, the “policy”most often put forward by account opponents,is not a viable option.

Background

The President’s Commission to StrengthenSocial Security was appointed with a mandate

The commissionattracted consid-

erable criticismfrom opponents

of personalaccounts. What

the commis-sion’s work didnot attract was

substantivecounter-

proposals.

2

The next stageof the SocialSecurity debateis for accountopponents tomake their case.

3

to “provide bipartisan recommendations to thePresident for modernizing and restoring fiscalsoundness to the Social Security System.”1

The 16-member commission, split evenlybetween Democrats and Republicans, was co-chaired by former senator Daniel PatrickMoynihan (D-N.Y.) and Richard Parsons, soonto be chief executive officer of AOL TimeWarner. The other members of the commissionincluded

• Leanne Abdnor (R), former vice presidentof the Cato Institute and executive directorof the Alliance for Worker RetirementSecurity;

• Sam Beard (D), founder and president ofEconomic Security 2000;

• John Cogan (R), former deputy director ofthe Office of Management and Budget,now a resident scholar at the HooverInstitution at Stanford University;

• Bill Frenzel (R), former U.S. representa-tive from Minnesota, now a resident schol-ar at the Brookings Institution;

• Estelle James (D), consultant with theWorld Bank, and lead author of the Bank’sinfluential 1994 book, Averting the OldAge Crisis;2

• Robert Johnson (D), chief executive officerof Black Entertainment Television;

• Gwendolyn King (R), former commission-er of Social Security (1989–92);

• Olivia Mitchell (D), professor of insuranceand risk management at the University ofPennsylvania’s Wharton School and execu-tive director of the Pension ResearchCouncil;

• Gerry Parsky (R), former assistant secre-tary of the Treasury (1974–77), now chair-man of Aurora Capital Partners;

• Tim Penny (D), former U.S. representativefrom Minnesota, now senior fellow at theUniversity of Minnesota’s Hubert H.Humphrey Institute of Public Affairs;

• Robert Pozen (D), former vice chairman ofFidelity Investments, now lecturer in pub-lic policy, Harvard University;

• Mario Rodriguez (R), president, HispanicBusiness Roundtable;

• Thomas Saving (R), professor of econom-ics at Texas A&M University and publictrustee of the Social Security program; and

• Fidel Vargas (D), vice president of Reliant

Equity Investors and member of the1994–96 Advisory Council on SocialSecurity.

The commission, which was instructed to sub-mit its recommendations to President Bush byDecember 21, 2001, worked according to the fol-lowing principles outlined by the president.

1. Modernization must not change SocialSecurity benefits for retirees or near-retirees.

2. The entire Social Security surplus must bededicated to Social Security only.

3. Social Security payroll taxes must not beincreased.

4. Government must not invest SocialSecurity funds in the stock market.

5. Modernization must preserve SocialSecurity’s disability and survivors’ com-ponents.

6. Modernization must include individuallycontrolled, voluntary personal retirementaccounts, which will augment the SocialSecurity safety net.

These principles were the starting point for thecommission’s deliberations. However, they didnot dictate the commission’s conclusions. Thepresident’s principles are flexible enough, infact, not to rule out the approach advocated byformer vice president Gore—retention of the tra-ditional Social Security defined benefit, aug-mented by supplementary personal accounts.Nor do they dictate that accounts must befinanced by redirecting existing payroll taxes.Indeed one of the commission’s plans includesnew contributions by workers. In short, althoughcritics claimed that the commission was“stacked” in a certain direction, the principles itworked under—not to mention the proposals itarrived at—were anything but preordained.

Interim Report

The commission’s first task was to completean interim report3 outlining the challenges fac-ing the current Social Security system—that is,to define the problem the commission and thecountry have to address. Given the intense reac-tion to the report from reform opponents, itseemed at times that the commission, rather

than the program, was viewed as the problem.Nevertheless, despite allegations that the reportcontained numerous factual errors,4 it in factholds up quite well under scrutiny.

In many ways, the commission’s interimreport simply echoed the annual reports of SocialSecurity’s trustees, who noted that the programfaces substantial and ongoing deficits over thelong term. Much like the trustees, the commis-sion urged that reform be undertaken soonerrather than later.

In its interim report, the commission reachedthe following conclusions regarding the currentSocial Security program, which complement theprinciples for reform outlined by the president.

• If we are to support tomorrow’s retireeswithout overburdening tomorrow’s work-ers, this generation of Americans must saveand invest more.

• The existing Social Security program doesnot save or invest for the future. It was notdesigned to facilitate saving, and the politi-cal process cannot be relied upon to saveon behalf of American families.

• Under the existing system, Americans willsoon face inescapable choices: cut SocialSecurity benefits, raise taxes, cut othergovernment spending, or borrow on anunprecedented scale.

•Arguments for doing nothing amount to directadvocacy of one or more of these options.5

In addition, the commission established eightcriteria by which to evaluate proposals tostrengthen the Social Security system:

1. Encouragement of workers’ and families’efforts to build personal retirement wealthby giving citizens a legal right to a portionof their benefits.

2. Equity of lifetime Social Security taxesand benefits, both between and withingenerations.

3. Adequacy of protection against incomeloss due to retirement, disability, death ofan earner, or unexpected longevity.

4. Encouragement of increased personal andnational saving.

5. Rewarding individuals for actively partici-pating in the workforce.

6. Movement of the Social Security systemtoward a fiscally sustainable course that

reduces pressure on the remainder of thefederal budget and can withstand econom-ic and demographic changes.

7. Practicality and suitability to successfulimplementation at reasonable cost.

8. Transparency: Analysis of reform plansshould measure all necessary sources oftax revenue, and all benefits provided,including those from the traditional systemas well as from personal accounts.6

Taken together, these criteria provide a basisfor formulating and assessing proposals toreform Social Security.

The policy reasons for reform are fundamen-tally demographic. Because Social Security is apay-as-you-go system, in which taxes paid bytoday’s workers are used to pay benefits fortoday’s beneficiaries, the relative sizes of theworking and retired populations are crucial todetermining the tax rates or benefit levels thesystem must apply.

To illustrate these basic but important rela-tionships, the commission’s interim report con-tained a one-page section entitled “Basic SocialSecurity Math.” Although the trustees’ long-term projections encompass myriad economicfactors such as wage growth, interest rates,changes in hours of work and general work-force participation, the basic math of pay-as-you-go Social Security financing is drivenalmost entirely by demographics.

Consider the equation presented by the com-missioners:

Average benefits as percentage of average taxable wage

Worker-to-beneficiary ratio

= Program cost as percentage of average wage

Because today’s Social Security benefit aver-ages 36 percent of the average worker’s wage, andbecause there are currently 3.4 workers per bene-ficiary, the payroll tax required to support today’sbeneficiary is around 10.5 percent of his earnings(36/3.4 = 10.5). Because today’s payroll tax rate isset by law at 12.4 percent of wages, SocialSecurity currently runs a surplus.

Notice how the above equation is structured,with one variable that we assume to be exogenous(i.e., whose value is not determined by the othervariables): the worker-beneficiary ratio.7 If thenumber of workers per retiree falls, then either theaverage benefit must fall (to maintain a constant

Inaction, the“policy” mostoften put for-

ward by accountopponents, is

not a viableoption.

4

In many ways,the commis-sion’s interimreport simplyechoed theannual reportsof SocialSecurity’strustees, whonoted that theprogram facessubstantial andongoing deficitsover the longterm.

5

tax rate) or the tax rate must increase (to maintainconstant benefit levels). Under the current pay-as-you-go financing, these are the constraints policy-makers and the public face (see Figure 1).

For instance, if the worker-beneficiary ratio fallsfrom 3.4 to 2, one of two things must happen: Theaverage benefits must fall from 36 percent of theaverage worker’s wage to just 25 percent, or thepayroll tax rate must rise to around 18 percent ofwages to maintain the 36 percent benefit rate. Noamount of kind words about “sacred trusts” canchange that unfortunate fact. It is not a matter ofpolitical commitment or concern for the elderly, asreform opponents charge. It is a matter of simplemathematics: Under the current system we can paypromised benefits or we can maintain current taxrates, but we cannot do both. No amount of kindwords justifies promising benefits without specify-ing how those benefits are to be paid.

Tax and Benefit Baselines:Not Mix and Match

The above discussion shows that the current12.4 percent payroll tax rate is mathematically

incompatible with currently legislated benefitlevels under today’s pay-as-you-go financing.One or the other—or both—must change.

In comparing benefits paid under the com-mission’s (or any other) reform plans with thecurrent program’s, it is very important to recog-nize the distinction between “promised” and“payable” benefits. Promised benefits are exact-ly that: what Social Security’s benefit formulapromises a worker with a given wage history.Social Security’s “payable” benefit, by contrast,is what the underfunded system will in fact payunder current law. As Table 1 shows, the prom-ised and payable benefit baselines must bematched with the appropriate tax baselines,which have been termed pay-as-you-go and cur-rent law. Pay-as-you-go is the rate required topay promised benefits, while current law is the12.4 percent rate currently in effect.

These tax and benefit baselines are not like amenu from which you may chose one baselinefrom the tax column and another from the ben-efit column. Each benefit baseline has only oneappropriate tax baseline.

Opponents of personal accounts often violatethis rule. Critics cite Social Security’s promised

8

10

12

14

16

18

20

1970

1976

1982

1988

1994

2000

2006

2012

2018

2024

2030

2036

2042

2048

2054

2060

2066

2072

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

Figure 1Social Security’s Cost Rises as the Population Ages

Source: The 2001 Annual Report of the Board of Trustees of the Federal Old-Age and Survirvors Insurance and DisabilityInsurance Trust Fund.

benefits when making comparisons with reformproposals, which is perfectly acceptable as longas they produce a plan that can actually pay thatpromised level of benefits. Unfortunately, mostreform opponents don’t have such a plan and,when challenged by the president’s commissionto produce one, failed. Again, account opponentsare not required to have a reform plan, as long asthey’re satisfied to cite only Social Security’spayable level of benefits.

If the reform debate is simply going to be anexercise in promising benefits without payingfor them, why not promise everyone 10 timesthe benefits without any question of where themoney will come from? Consistency of base-lines is probably the single most importantthing to remember to avoid being deceived inthe Social Security debate. Without a reformplan, promised benefits are just a pipe dream.

A related issue involves clarifying what “cur-rent law” benefit levels truly are. Many com-mentators use the phrase current law to denotepromised benefits. Under the Social SecurityAct, taxes trump benefits. As the ConcordCoalition puts it:

All of the long-term spending projectionsfor Social Security—by the CongressionalBudget Office, the GAO, and the SocialSecurity Administration itself—assumethat current-law benefits will be paid in

full, even after the trust funds are empty.This cannot happen. If Social Security issimply left on autopilot, the law leaves nodoubt that the contradiction will beresolved the other way around—with mas-sive benefit cuts. Making this fact morewidely known would have two salutaryconsequences. In general, it would causethe public to take a more active interest inreform. And in particular, it would allow afairer comparison of reform proposals withcurrent law. As things stand, reformers facethe hopeless task of trying to out-promise asystem that is unsustainable.8

The Social Security Administration is author-ized to issue benefit checks only when sufficientfunds exist in the trust fund. Because these fundsare ultimately derived from legislated payroll taxrates, under current law, when the fund becomesinsolvent in 2038, taxes will remain constantwhile benefits for all Social Security beneficiaries(not simply new retirees) will be reduced to thelevel payable by payroll tax receipts.9

The current Social Security system’sfinances are driven almost entirely by tax rates,benefit levels, and the ratio of workers toretirees. Other factors, including economicgrowth and interest rates, play relatively minorroles. The problem is that currently legislatedtax rates, when combined with the projected

The basic mathof pay-as-you-go

Social Securityfinancing is

driven almostentirely by

demographics.

6

Table 1Tax and Benefit Baselines Must Match

Benefits Taxes

Promised 100 percent of Pay-as-you-goa 18.3 percent currently scheduled

benefits

Payable 72 percent of Current law 12.4 percentcurrently scheduled

benefitsb

Source: 2001 Trustees Report, Table IV.B1, “Estimated Income Rates and Cost Rates, Calendar Years 2001–2075.”a. Average of years 2038–2075, ranging from 17.8 to 19.4 percent of taxable payroll.b. Average of years 2038–2075, ranging from 74 to 69 percent of schedule benefits.

In comparingbenefits paidunder the com-mission’s (orany other)reform planswith the currentprogram’s, it isvery importantto recognize thedistinctionbetween “promised” and“payable” benefits.

7

worker-retiree ratios, simply cannot equal thebenefit levels promised under current law.

The Interim Report and theTrust Fund

Although much of the interim report’s dis-cussion was unremarkable, two issues in partic-ular attracted attention—most of it unfavorable.The first was the commissioners’ claim that theSocial Security trust fund has not been effectivein prefunding the program for the future. Thefund, the commission argued, is an asset toSocial Security but an equal and opposite lia-bility to the rest of the federal government.From the point of view of the taxpayer, itchanges very little.

Commissioners and staff knew that the inter-im report would ruffle a few feathers.Nevertheless, there was genuine surprise at thereaction to the report’s argument that the SocialSecurity trust fund has not effectively prefund-ed the system for the future.

Upon release of the report, the commission’sview of the trust fund came under immediateattack in a paper written by Henry Aaron, AlanBlinder, Alicia Munnell, and Peter Orszag andissued by the Century Foundation and theCenter on Budget and Policy Priorities:

The commission asserts that the SocialSecurity Trust Fund does not hold realassets. The Social Security Trust Fund,however, currently holds more than $1 tril-lion in Treasury securities. These assets arebacked by the full faith and credit of theU.S. Government, the benchmark of secu-rity in global financial markets.10

Similar statements came from Rep. BobMatsui (D-Calif.), who said, “This report false-ly asserts that the Social Security system will beunable to meet its obligations because theassets held by the Trust Fund are ‘not real.’ Tothe contrary . . . because U.S. Treasury Bondsare the benchmark for security in global finan-cial markets and to assert otherwise is to sug-gest that the U.S. Government will no longerhonor its debt obligations.”11 For these andother assertions, Matsui said, the commissionshould have been disbanded.

Congressional Democrats such as House

Minority Leader Richard Gephardt echoedthese remarks. Gephardt claimed:

The president’s commission has pub-lished a misleading, misguided report thatis one of the most skewed documents thatI have seen in many, many years. . . . Theassertion that Social Security is goingbust in 2016 flies in the face of all reality.The facts are Social Security has enoughreserves in the trust fund to last until atleast 2038.12

As discussed below, the foregoing state-ments do not accurately reflect the argumentsmade by the commission. Moreover, the com-mission’s arguments were in the past supportedby the very critics who attacked them.

Given this controversy, it is worth outliningwhat the commission’s interim report did anddid not say about the trust fund.

• The commission did not say the trustfund’s trillion dollars in government bondsare not “real,” nor that they are not assets ofthe Social Security system.

• Moreover, the commission did not say thatthe government would default on the trustfund’s bonds, despite what some commen-tators have repeatedly charged.13 On thecontrary, on p. 18 of the interim report thecommissioners specifically declared that“the bonds in the Social Security TrustFund will be honored.” Nothing could beclearer than that, and the commission’sown reform proposals bore out that pledge.

What the commission said about the TrustFund is subtler than the caricature presented byreform opponents, and in retrospect the interimreport could have been clearer and more com-prehensive in its presentation. Whether thiswould have quieted the opposition is debatable,but it would at least have reduced misunder-standings among the public.

First, the commission pointed out thatalthough the trust fund’s bonds are an asset toSocial Security, they are an equal and oppositedebt to the rest of the government. This is not amatter of economics, but merely of accounting:an asset to Social Security must be a debt tosomeone, and that someone is the federalTreasury and, by extension, the taxpayer. The

interim report illustrated this point with a num-ber of quotations from the Clinton administra-tion and such nonpartisan sources as theCongressional Budget Office, the Congression-al Research Service and, here, the GeneralAccounting Office:

[Social Security] Trust Funds are not likeprivate Trust Funds. They are simplybudget accounts used to record receiptsand expenditures earmarked for specificpurposes. A private Trust Fund can setaside money for the future by increasingits assets. However, under current law,when the Trust Funds’ receipts exceedcosts, they are invested in Treasury secu-rities and used to meet current cash needsof the government. These securities are anasset to the Trust Fund, but they are aclaim on the Treasury. Any increase inassets to the Trust Funds is an equalincrease in claims on the Treasury. 14

Hence, the trust fund is indisputably not a netfinancial asset with which the government canpay Social Security benefits. That is, the fund’sbonds do not put off the need for tax increasesor spending cuts by a day or a dollar, becauseprecisely the same steps must be taken to repaytrust fund bonds as would be needed to pay fullbenefits directly via tax increases or spendingcuts. After all, without a trust fund, beginning in2016 we would have to raise taxes, borrow, orcut other spending to pay full Social Securitybenefits. With a trust fund, beginning in 2016we must raise taxes, borrow, or cut other spend-ing to repay the trust fund bonds that will payfull Social Security benefits. In terms of timingand cost, from the perspective of the overallbudget—and the taxpayer—there is no substan-tive difference between having a trust fund andnot having a trust fund.

A second point, however, is more important.The accounting identity pointed out by thecommission—assets to Social Security equaldebts to the government—does not mean thatthe trust fund did not or could not effectivelyprefund future benefits at the economic level.There was an economic argument in the inter-im report that is more important than theaccounting point of view, and it addresses notwhat happens when trust fund bonds areredeemed in 2016 but what actually happened

in the period from 1985 to today when thosebonds were amassed.

In essence, the commission argued that theSocial Security payroll tax surpluses seen sincethe mid-1980s were never “saved” in a trueeconomic sense. Although the trust fund was“credited” with government bonds equal in sizeto these cash surpluses, thereby saving them ina financial or accounting sense, the cash itselfwas not used to reduce government borrowingor repay existing government debt. Rather thanincreasing government saving, these payroll taxsurpluses tempted Congress to either spendmore or tax less than it otherwise would have.In this case, though the trust fund was still cred-ited with bonds, paying future Social Securitybenefits would still require raising future taxes.Future workers would suffer a net loss as taxesrose to repay the trust fund’s bonds.

The commission’s critics rejected this viewas well. The authors of the Century Foundationpaper, for instance, stated:

The commission asserts that in the future,“the nation will face the same difficultchoices as if there had been no Trust Fundat all.” This assertion ignores the real eco-nomic contribution of the Trust Fund. Theaccumulation of Trust Fund reserves rais-es national saving, reduces the public debtand thereby reduces the annual cost ofpaying interest on that debt, and promoteseconomic growth.15

Along the same lines, Munnell and Weaverargue that trust fund surpluses have madeunequivocal additions to national saving: “Theexcess payroll taxes that have been used tobuild up reserves in the Trust Funds haveincreased national saving and in recent yearshave helped pay down the debt. In this way, thevery real sacrifice of today’s workers has boost-ed investment and enhanced our capacity to payfuture benefits.”16 How this process is thoughtto have worked merits some explanation.

According to this view, past payroll tax sur-pluses reduced government borrowing and, inseveral years, allowed existing governmentdebt to be repaid. In doing so, they increasednational saving, which in turn increased theamount of capital per worker, thereby increas-ing productivity. Because productivity lies atthe root of wages, these too would increase. By

If the reformdebate is simply

going to be anexercise in

promising bene-fits without pay-

ing for them,why not promise

everyone 10times the bene-

fits without anyquestion of

where themoney will come

from?

8

Although thetrust fund’sbonds are anasset to SocialSecurity, theyare an equal andopposite debt tothe rest of thegovernment.This is not amatter of eco-nomics, butmerely ofaccounting.

9

the time Social Security needed to tap the trustfund, wages and national income in generalwould have risen by an amount greater than thetax increases needed to repay the trust fund’sbonds. Hence, the trust fund could be repaidwithout making future workers worse off thanthey would have been had the entire enterprisenever taken place.

In a 1989 study, Aaron, Bosworth, andBurtless conclude that trust fund financingcould, in theory, succeed through precisely thismechanism:

If national saving and domestic investmentare increased by the additions to SocialSecurity reserves, wages will rise about 7percent more than trend growth. Thatincrease would pay for the added pensioncosts generated by the rising proportion ofbeneficiaries in the total population.Workers active during the twenty-first cen-tury would actually enjoy a higher standardof living than in a world where the propor-tion of pensioners did not increase. The cen-tral question is whether Social Security sur-pluses will be used to add to national savingor to finance current consumption.17

Indeed that is the central question, as thecommission stated directly in its interim report.

In the Century Foundation paper, however,the authors—Aaron included—overlook thiscentral question and treat Social Security’simpact on national saving as a fairly cut-and-dried affair in which, as a matter of accounting,Social Security surpluses raise saving on a dol-lar-for-dollar basis:

If the non–Social Security portion of thebudget had a deficit of $300 billion in agiven year, and Social Security ran a $100billion surplus, the net deficit would be$100 billion smaller—and national saving$100 billion higher—than otherwise. Theonly way in which Social Security surplus-es would fail to increase government sav-ing is if Congress decided to increasespending or reduce taxes in the non–SocialSecurity part of the budget because of thesurplus in Social Security.18

Of course, it is this last option that the commis-sion considered to be most consistent with the his-

torical evidence.19 As shown below, in their moreacademic work the commission’s critics acknowl-edge the very same issues the commission noted.And, by and large, these critics’ own work lendssupport to the commission’s position.

The Century Foundation authors are, ofcourse, correct that, all other things being equal,a dollar of Social Security surpluses equals adollar of extra saving. But, as many arguedfrom early in the post-1983 period, all thingsaren’t equal: rather than save surplus SocialSecurity funds, the government could use themto hide the size of the deficit in non–SocialSecurity federal spending. For instance, in 1995President Clinton acknowledged,

We clearly have been using payroll taxesfor 12 years now, long before I ever camehere, to minimize the size of the deficitexclusive of the payroll tax, so that from1983 forward, previous Democratic con-gresses and Republican presidents madejudgments that it was better and political-ly more palatable to tax payroll thanincome, even though it’s a burden onworking people and small businesses.20

Similarly, North Dakota senators KentConrad and Byron Dorgan wrote in 1995 thatthe payroll tax “is dedicated solely for workingAmericans’ future retirement, it shouldn’t beused either for balancing the operating budgetor masking the size of the budget deficit.” If itis used for those purposes, they warned, whenneeded “the retirement fund would have noth-ing but IOUs in it.”21 That is to say, if payrolltax surpluses are devoted to consumption ratherthan saving, then the overall burden on futureworkers is not reduced and the overall capacityof the economy to support Social Security pay-ments is not enhanced.

Unfortunately, that is exactly what tookplace. In a 1989 report to Congress, the GeneralAccounting Office stated:

The changes to Social Security enacted in1983 are not producing the result of less-ening the burden of paying for the retire-ment benefits of the baby boom genera-tion. The budgetary reality is that the pay-roll taxes are being used to finance the cur-rent operations of government and aremasking the size of the on-budget deficit.

The economic reality is that the Trust Fundreserves consisting of Treasury securitiesthat are financing current consumptionrather than productive investment are illu-sory. They will remain so until the rest ofthe government achieves approximate bal-ance between revenues and outlays.22

For the record, it was only in 1999–2000 thaton-budget government finances reached surplus.In 1990 testimony to Congress, the GAO’s head,Comptroller General Charles A. Bowsher, said“the luxury of these reserves has provided a con-venient excuse for avoiding the tough choicesneeded to cut the general fund deficit.”23

In other words, the GAO found that the on-budget balance was not independent of the SocialSecurity balance; larger surpluses in SocialSecurity facilitated larger deficits elsewhere. Inthat case, Bowsher said, “The growing SocialSecurity surpluses are serving more as a substitutefor other deficit reduction action than as a netaddition to national savings. . . . If we do not usethe accumulating Social Security reserves toincrease our national savings rate, we will be in nobetter position to meet our obligations to futureretirees than we would be if we had remainedunder pay-as-you-go financing.”24

Lawrence H. Thompson, the GAO’s assis-tant comptroller general, put it most plainly:“We shouldn’t kid the American people intothinking extra savings is going on.”25

The Commission’sOpponents Share Its View

of the Trust FundAlthough the commission’s view of the trust

fund clearly has support among politicians ofboth parties and from the nonpartisan GeneralAccounting Office, it is ironic that perhaps thestrongest support for its view comes from thevery analysts who so severely criticized thecommission’s interim report.

As pointed out above, it is possible—indeedplausible—that payroll tax surpluses wouldtempt Congress to spend more or tax less thanit otherwise would. Doing so would not changeSocial Security’s finances on paper, but wouldgreatly alter the economic realities of payingfuture benefits. In the early 1980s, soon afterthe decision to “prefund” Social Security via

trust fund surpluses was made, Munnell fearedthis could be the outcome:

If the payroll taxes earmarked to pay futureretirement and disability benefits are used tocover current outlays from the general fund,then the government debt held by the TrustFunds will be no more than paper claims.When the baby boom generation retires afterthe turn of the century, the Trust Funds willredeem their claims on the Treasury in orderto pay promised benefits. The Treasury,however, will not have accumulatedresources to meet its obligations but ratherwill be forced to raise taxes at that time topay off its debts. Thus, the full burden ofsupporting the beneficiaries will come fromthe future taxpayers—just as if the systemhad been financed on a pay-as-you-go basisall along.26

This is precisely the fear expressed by thecommission.

It is, of course, difficult to know what thelevel of non–Social Security federal spendingand taxation would have been had there beenno payroll tax surpluses.27 Nevertheless, thesurprising fact was that most of those arguingagainst the commission’s portrayal of the trustfund had in the past actually agreed with it—atleast until the commission’s argument was usedto buttress the case for personal accounts.

For instance, while Congressman Matsuiattacked the commission’s depiction of the trustfund, in the past he had clearly embraced boththe commission’s reasoning and its conclu-sions. In a 1990 op-ed coauthored with Sen.Bob Graham (D-Fla.), Matsui’s rhetoricalattack against the fund went beyond anythingstated in the interim report:

Trust Fund reserves are growing at thepace of a billion dollars a week. But thesebillions won’t be available to the next gen-erations of America’s retirees. As quicklyas the surpluses amass, they are beingsiphoned off to help finance the deficit.Bluntly put, the federal government isspending more than $1 billion a week ofthe Social Security surplus as though itwere general revenues. All that the TrustFund gets for these expenditures are chitsfrom the U.S. Treasury. . . . If those monies

The commissionargued that

Social Securitypayroll tax sur-

pluses werenever “saved” ina true economic

sense.

10

Rather than savesurplus SocialSecurity funds,the governmentcould use themto hide the sizeof the deficit innon–SocialSecurity federalspending.

11

are really to be there, if when we retire weare going to be left with anything morethan a vault full of Treasury DepartmentIOUs, integrity must be restored to the useof the Trust Fund surpluses.28

To prevent further government spending ofSocial Security surpluses, Matsui and Graham pro-posed investing those funds in nonfederal municipalbonds, though the lower rates of return on thosebonds relative to the fund’s current holdings wouldrequire ongoing general revenue subsidies to makeup the losses. In other words, Matsui and Grahamapparently believed so strongly that Social Securitysurpluses were being spent rather than saved thatthey proposed a plan that was, in financial terms, amoney-loser simply to get those surpluses out of thehands of the federal government.

This aspect of Matsui’s proposal is particular-ly interesting given Matsui’s charge that a faultof reform is “plans to allow people to direct partof their payroll taxes into individual accountsmake Social Security’s financing problemworse, not better”29 and “privatization proposalswould only make Social Security’s challengesharder to solve.”30 Of course, Social Security’sactuaries state, “If the personal accounts are con-sidered as part of ‘Social Security,’ it is reason-able to combine the amounts of Trust Fundassets and personal accounts for a representationof total system assets.”31 By this standard, asdetailed below, the commission’s three reformproposals make an immeasurable improvementover current law.

Moreover, as recently as the summer of2000, Matsui warned his constituents, “Whenthe Baby Boomers begin to retire in the next 10years, Social Security will begin to pay outmore in benefits than it receives in revenue,”32

apparently giving credence to the 2016 datestressed in the interim report.

A similar warning was sounded by the coau-thors of the Century Foundation paper. In 1988,for instance, Henry Aaron took a very skepticalview of the trust fund:

The economic justification for additionsto Social Security reserves is that suchsurpluses increase national saving, add tothe U.S. capital stock, and boost produc-tive capacity in anticipation of the extracosts a growing population of retirees willgenerate. Current budget policy over-

whelms this sound policy. Instead ofadding to U.S. national saving, currentfiscal policy simply diverts a part of pay-roll taxes to pay for ordinary operationsof government.33

It’s worth pointing out that the budget policyAaron was decrying in 1988 is the same budg-et policy that, with brief exceptions, has heldevery year since 1985.

Moreover, in 1988, Aaron, Bosworth, andBurtless clearly rejected the idea that it wasbeing effectively carried out:

If OASDHI [Old Age, Survivors,Disability and Hospital Insurance] revenuesexceed annual expenditures, the resultingsurpluses may be used to pay for currentpublic or private consumption (eitherthrough increases in non-OASDHI govern-ment expenditures or through reductions innon-OASDHI taxes). As the OASDHI sur-pluses increase, so would deficits elsewherein the federal budget. Although this policymay seem peculiar, it closely resembles thecourse on which the United States isembarked today. Under this policy, thereserve does not add to national saving(because it does not reduce the overall gov-ernment budget deficit) and, hence, it doesnot add to future productive capacity. Ineffect, the OASDHI surpluses are borrowedto pay for current government services,replacing income tax revenues or cuts inother government programs sufficient tobalance the non-OASDHI budget.

Although such a policy might holddown future payroll tax rates, it cannotprotect future taxpayers from shoulderingthe expense of rising benefit costs. Whenand if the Trust Funds are drawn down topay for future benefits, other federal taxeswill have to be increased to finance therepurchase of government debt previous-ly bought by the OASDHI Trust Funds. Inaddition, the incomes against which thosetaxes are imposed will be no larger than ifthe reserves never existed. Since futurebenefits must be paid out of future pro-duction, the burden on future taxpayerswould not be reduced.

The authors concluded:

The growing surpluses in the SocialSecurity system camouflage a major dete-rioration in the budget balance for non-OASDHI operations. . . . In effect, the cur-rent policy is to borrow the OASDHI sur-plus to finance a deficit in the rest of thebudget. As a result the payroll tax, ostensi-bly earmarked for retirement, survivors,disability, and hospital insurance, is beingused increasingly to pay for other govern-ment expenditures, such as defense andinterest on the public debt.34

Aaron has not, to the author’s knowledge,explained whether or why he changed his view.It is arguable that trust fund surpluses were atleast partially saved during the years of on-budget surpluses in 1999–2000, yet Aaronappears to have changed his mind about whattook place before those years as well, withoutproviding evidence as to why.

Alicia Munnell expressed similar skepticismtoward trust fund financing in the 1980s on thebasis of evidence from other countries thatattempted trust fund financing and the factorsthat influenced their success or failure. Threefactors that reduce the prospects for successfultrust fund financing:

• Whether the pension fund’s surpluses areconsidered part of a unified budget.

• Whether the fund can invest in privatesecurities or is a “captive market” of theTreasury.

• Whether the government fluctuated politi-cal control from party to party.

As Munnell noted, all of these criteria applyto the United States:

One factor in this regard is probablywhether the Social Security programs areincluded in some type of unified budgetor are accounted for separately. If TrustFund activity is integrated with other fed-eral functions and the total reported as asingle figure, as has been true in theUnited States since 1969, Congress andthe public would be encouraged to thinkthat the Trust Fund reserves are availableto cover general government outlays.

Another closely related factor is theease with which the Treasury can borrow

from the Trust Funds. This depends on theextent to which the administration and thefinances of the Social Security TrustFunds and the rest of the government areintertwined.

In addition, Munnell argued, the use of a“unified budget” concept in the United Statestends to blend Social Security and non–SocialSecurity funds in the eyes of lawmakers:

In the United States the finances of theSocial Security Trust Funds and the restof the budget are closely intermingled.The Treasury Department, rather than theSocial Security Administration, collectsthe earmarked payroll taxes and depositsthem in a general account with other rev-enues it receives. The Trust Funds arethen issued with special federal securitiesin a compensating amount. While the bal-ances of the securities reflect theresources available to the Social Securityprograms, they more closely resemblespending limitations than control overresources. One would expect less use ofTrust Fund revenues for general govern-ment expenditures in situations where theTrust Funds are more than a bookkeepingactivity on the part of the TreasuryDepartment.

Munnell goes on to point out:

The likelihood of the members ofCongress responding to the Social Securitysurpluses in this manner probably dependslargely on their ability to count the surplus-es toward overall budget deficit reduction.All three countries studied keep their SocialSecurity accounts very separate from therest of the budget, and this appears to havediscouraged the legislatures from incorpo-rating Social Security surpluses in their gen-eral budget decisions or their deficit reduc-tion efforts. As long as the United Statesretains a unified budget and frames itsdeficit targets in these terms, Congress willbe tempted to keep one eye on the surplus-es when voting on tax and expenditure pro-posals.

A somewhat discouraging result, forthose committed to increasing national

Largersurpluses in

Social Securityfacilitated

larger deficitselsewhere.

12

Although SocialSecurity sur-pluses haveclearly fundedthe system in anarrow sense, inthe broadersense of raisingnational saving,they have fallenshort.

13

saving through accumulating reserves inthe Social Security Trust Funds, is that thegreatest success has occurred in countrieswith stable and disciplined political envi-ronments, where one party has been inpower almost continuously since theexperiment began.35

For these reasons, soon after the 1983reforms Munnell argued against running SocialSecurity surpluses at all.

With her co-author Lynn Blais, now of theUniversity of Texas School of Law, Munnellelaborated:

The assumption that an increase in TrustFund reserves will represent a net incre-ment to national saving may not be valid.Instead, surpluses in the Social SecurityTrust Funds may very likely be offset bydeficits in the rest of the budget, so thatthey are, in effect, used to finance generalgovernment expenditures. Indeed, the pat-tern shown in Table 2 [showing SocialSecurity and general budget balance from1946–1983] indicates that this may havebeen the case in the past. Of course, thesefigures are far from conclusive because itis difficult to determine what the balancewould have been in the federal budgetwithout the surpluses in the Social SecurityTrust Funds. Nevertheless, our best guessis that the scheduled buildup of assets inthe Social Security Trust Funds over thenext 35 years would be used to offsetdeficits elsewhere in the federal budgetand thus would contribute little to overall

saving and capital accumulation.In this country it appears it would be

difficult as a practical matter to stockpilereal resources in anticipation of futurebenefit payments. It is more likely thatCongress would divert surpluses in theTrust Fund to offset deficits in other partsof the federal budget.

Munnell and Blais concluded at the time thatit would be preferable to abandon efforts to pre-fund Social Security through the trust fundfinancing mechanism:

In view of the improbability that SocialSecurity surpluses will increase nationalsaving and the possibility that, if increasedsaving did materialize, it would haveadverse fiscal implications or disrupt finan-cial markets, the authors conclude that itwould be preferable to return the SocialSecurity system to pay-as-you-go financingwith a substantial contingency reserve.36

Incidentally, Senator Moynihan took the sameview when, in 1989–90, he argued that if SocialSecurity surpluses were not truly being saved,the system’s finances should return to a pay-as-you-go basis. “We cannot continue to use regres-sive payroll taxes to finance general governmentexpenses. This is not acceptable,” he said.37

Alan Blinder of Princeton University has writ-ten less on Social Security trust fund surplusesand national saving than Aaron or Munnell.Nevertheless, in a 1990 commentary on a paperby Nobel laureate James Buchanan in whichBuchanan argues that “a small dose of public

Table 2Personal Account Plans Would Maintain or Enhance System Progressivity

Current Law Plan 1 Plan 2 Plan 3

Benefit to low-wage retiree $767 $770 $868 $823Benefit to high-wage retiree $1,673 $1,684 $1,556 $1,646Low as percentage of high 46% 46% 56% 50%

Source: Derived from Stephen C. Goss and Alice H. Wade “Estimates of Financial Effects for Three Models Developed bythe President’s Commission to Strengthen Social Security,” Memorandum dated January 31, 2002, pp.74–76. Based onworkers retiring in 2022 and holding the default investment portfolio.

choice theory might have dampened the enthusi-asm of those who sought to ensure the integrityof the system” by running Social Security sur-pluses within the context of the overall budget,38

Blinder stated his agreement with Buchanan onthis point: “Buchanan asserts that the SocialSecurity surpluses of the next thirty years or sowill make the government prone to do more non-Social Security spending than it otherwise wouldhave done. As I have already made clear, I sus-pect he is right.”39

Peter Orszag of the Brookings Institution,another author of the Century Foundationpaper, has also written less in the past on theimpact of trust fund accumulations on nationalsaving. Nevertheless, in recent Congressionaltestimony Orszag implicitly acknowledges thatpast surpluses may not have contributed tonational saving in the dollar-for-dollar manneroutlined in the Center on Budget and PolicyPriorities paper. Orszag argues, in much thesame way as the commission, that it is impor-tant to

distinguish between “narrow” and “broad”prefunding. In its narrow sense, prefundingmeans that the pension system is accumu-lating assets against future projected pay-ments. In a broader sense, however, pre-funding means increasing national saving.The broader definition of prefunding—higher national saving—should be our ulti-mate objective. But we can sometimes beled astray, because prefunding in the narrowsense need not imply prefunding in thebroader sense (i.e., higher national saving).40

Orszag points out, correctly, that both trust fundand personal account financing could create nar-row, but not broad, prefunding if extra savingthrough the pension system were offset by lackof saving elsewhere. Individuals could reducecontributions to non-Social Security investmentaccounts or the government could increasespending or lower taxes elsewhere in the budget.This is precisely the distinction made in the com-mission’s interim report: although SocialSecurity surpluses have clearly funded the sys-tem in a narrow sense, in the broader sense ofraising national saving, they have fallen short.

Orszag follows the above passage with thestatement that “the emerging political consen-sus not to spend the Social Security and

Medicare surpluses is precisely what ensuresthat the narrow prefunding in the Trust Fundsalso corresponds to higher national saving(i.e., broad prefunding).”41

But if a lock-box is needed to ensure thatSocial Security surpluses translate intoincreased national saving, then we can assumethat when the lock-box budget mechanism wasabsent—from 1983 through 1999—increasedsaving is less likely to have occurred. Viewedanother way, if Social Security surpluses weresaved as a matter of course, as Orszag and theother authors of the CBPP paper maintain, anylock-box would be redundant. Orszag’s com-ments appear to acknowledge this. Moreover,although lock-boxing future Social Securitysurpluses may raise national saving, it cannotchange the fact that past surpluses were nottreated in this way.

Other Congressional leaders held similarviews of the efficacy of the Trust Fund. In 1989Rep. Richard Gephardt (D-Mo.) argued that thegovernment “should stop borrowing on work-ers’ retirement benefits. People should besecure in the knowledge that the system will beable to send out their monthly Social Securitychecks. If the practice of borrowing againstSocial Security monies continues, that securityis threatened.”42 In 1990, Gephardt said, “WhatDemocrats want to do is we want to stop thestealing of [trust] funds to mask the deficit.”According to the article, Gephardt said, “Thegovernment would have to pay interest on theborrowed money. To pay back Social Securityby 2015 or 2018, taxes would have toincrease.”43 That is precisely the point the com-mission’s interim report made. It should now beclear that most of the commission’s mostprominent critics agreed with it.

As an aside, many commentators who have crit-icized the tax cuts passed by Congress in 2001 asweakening Social Security simultaneously take theview that Social Security’s trust fund is “real”regardless of whether accompanying payroll taxsurpluses are spent or saved. But the bookkeepingbalance of Social Security’s trust fund is unaffectedby the balance of the non-Social Security budget;whether the tax cut was passed or not, SocialSecurity would be credited with the same amount ofbonds. Critics are free to argue that funds sent backto the public in last year’s tax cut could have beenbetter used by saving them for Social Security.44 Ifthat is the case, though, those critics must also accept

If Social Securitysurpluses were

saved as a matterof course, any

lock-box wouldbe redundant.

14

Although lock-boxing futureSocial Securitysurpluses mayraise nationalsaving, it cannotchange the factthat past sur-pluses were nottreated in thisway.

15

the commission’s logic that the “raids” on the trustfund that took place practically every year from1985 to the present must have weakened SocialSecurity far more because these prior “raids” wereused to finance additional government spending.There is at least a theoretical argument that therecent tax cuts will stimulate long-term economicgrowth. In their responses to the commission’s inter-im report, account opponents failed to reconciletheir short-term arguments against tax cuts with theirlonger-term defense of the current trust fund financ-ing structure.

Now, one might ask, “Why do these argu-ments about the trust fund even matter?” In onesense, they don’t: the assets we have are theassets we have, and nothing we do today canchange the past. This was the point of the com-mission’s accounting argument: as benefit costsbegin to rise, there is no separate store of wealthwith which to pay them. Looking forward,though, the trust fund issue is extremely impor-tant. If we conclude that trust fund financingdoesn’t truly reduce burdens on future taxpay-ers, we have only three other options:

• Return to pay-as-you-go financing.• Have the government invest the trust fund

in the private assets.• Invest individually through personal

accounts.

Few wish to return to pay-as-you-go financing,which entails a 50 percent increase in payrolltax rates to meet current benefit promises. And,politically speaking, a battle between govern-ment investment and personal accounts wouldbe a rout. The economics may be the same, butthe politics are worlds apart: the public simplydoesn’t trust the government to invest in privatecorporations, and even after two years of weakmarket performances public support for per-sonal accounts remains strong.45

What made the trust fund controversy sur-prising was that the very people who mostprominently disagreed with the commission’sview last summer had prominently agreed withit before—before, that is, it was used to supportthe case for personal accounts.

Progressivity

The second point of controversy in the inter-

im report was the commission’s contention thatthe current Social Security system isn’t nearlyas progressive as many of its proponents claim.If Social Security’s progressivity is called intodoubt, the transition to personal accounts—which are often designed with less explicitregard to progressivity—would not significant-ly alter the distribution of costs and benefits tothe system as a whole.

This idea, like the commission’s views onthe trust fund, came under attack. For instance,Peter Coy of Business Week wrote:

Today’s Social Security is progressive—that is, it transfers money from rich to poorin large part because of the benefits formu-la. The higher your average lifetime income,the less of it is replaced by your benefitcheck in retirement. That Robin Hood for-mula largely offsets the fact that low-income people tend to collect fewer checksbecause they die younger. “PrivatizingSocial Security may have some merits,”Coy said, “but the argument that it wouldbenefit the poor is deeply flawed.”46

In fact, the academic research that Coy reliedon in his critique of private accounts showsclearly how little Social Security’s so-calledRobin Hood benefit formula actually takes fromthe rich and gives to the poor. Social Security’scomplex benefit formula does a great deal ofredistribution on the basis of longevity, maritalstatus, and the relative wages of spouses, butvery little on the basis of income. Personalaccount reform proposals—even proposals withno redistributive intent—would leave systemprogressivity largely unchanged. In fact, thecommission’s personal account plans are con-sciously progressive, and enhance benefits forlow-wage workers, as discussed below.

Research on redistribution in Social Securityis complex and ongoing, and ranges from theconstruction of models of representative work-ers to the statistical analysis of large numbers ofactual earnings records. The following summa-ry relies on a recent study by Alan Gustman ofDartmouth College and Thomas Steinmeier ofTexas Tech, who used Social Security earningsrecords contained in the National Institute onAging’s Health and Retirement Study to disag-gregate the various factors influencing SocialSecurity’s overall progressivity. 47 Although

details differ among researchers, the Gustmanand Steinmeier article is representative of thegeneral direction in which academic researchon Social Security’s progressivity is flowing.Some other research finds even less progressiv-ity in the current program.48 Nevertheless, theGustman and Steinmeier article is reasonablyrepresentative of current research.

On its face, Social Security is highly redis-tributive because its progressive benefit formu-la replaces a substantially higher portion of alow-income worker’s preretirement earningsthan of that a high-income worker. On the basisof the benefit formula alone, Gustman andSteinmeier find that Social Security shouldredistribute 12.6 percent of total benefits fromhigher to lower-income workers (see Figure 2).This would be sufficient to substantiallyincrease the benefits for low-wage retirees.

But Gustman and Steinmeier find four majorfactors that substantially reduce SocialSecurity’s true progressivity.

First, the correlation between income andlife expectancy means that, while low-incomeretirees may receive relatively higher monthly

benefits, total lifetime benefits are not nearly soprogressive. Once differential mortality is fac-tored into the equation, net progressivity isreduced by 16 percent, leaving 10.6 percent oftotal benefits redistributed.

The influence of differential mortality onSocial Security’s progressivity could increasein the future. Although life expectancies for allAmericans are increasing, there is evidence thatthey are increasing more slowly for individualswith lower wages and educations. If that is thecase, Social Security could become even lessprogressive in this regard.49

A second factor is that Social Security paysspousal benefits. The spouses of high earners onaverage live longer and receive higher benefitsthan the spouses of lower earners, reducing pro-gressivity by another 30 percent. This leaves 6.8percent of total benefits redistributed.

Although Gustman and Steinmeier make noprojections regarding the future, this issuecould be complicated in coming years. On theone hand, as more women have entered theworkforce and gained benefit eligibility on thebasis of their own earnings, the spousal benefit

Account oppo-nents failed toreconcile their

short-term argu-ments against

tax cuts withtheir longer-

term defense ofthe current trust

fund financingstructure.

16

12.6

10.6

6.8

5

2.5

0

2

4

6

8

10

12

14

Basic benefitformula

Income/lifeexpectancy

Spousal benefits Individual vs.household

Actual vs.potential earnings

Figure 2Actual Social Security Redistribution Is One-Fifth of What Basic Benefit Formula Implies

Source: Alan Gustman and Thomas Steinmeier, “How Effective Is Redistribution under the Social Security BenefitFormula?” Journal of Public Economics 82, no. 1 (October 2001): 1–28.

Factors That Reduce True Progressivity

may play a smaller role in the future than itdoes today. On the other hand, the spousal ben-efit could increase disparities in benefitsbetween individuals of different races, as the“marriage gap” between blacks and whites haswidened in recent decades. In 1960, some 67percent of white females aged 15 and over weremarried, versus 57 percent of black females. In1998, however, although the percentage ofwhite married females had dropped to 58 per-cent, for black females it had fallen to 36 per-cent, making most black females ineligible forspousal benefits at retirement.50

A third factor that reduces Social Security’sprogressivity is that much of the program’sapparent redistribution is from the richer to thepoorer spouse living in the same household,rather than from an upper-income household toa lower-income household. In other words,when benefits are redistributed from husband towife, there is no effect on household progres-sivity. Because spouses share income andexpenses, the household is the more relevantstandard for public policy purposes. From thatperspective, Social Security’s progressivity isreduced another 14 percent, leaving 5 percentof total benefits redistributed.

Finally, although many individuals with lowlifetime incomes have worked full careers atlow wages, many others are high-incomeworkers who took time out of the workforce.51

In fact, Gustman and Steinmeier found thatmost of the individuals with the lowest lifetimeearnings were women with total householdincomes far above what could be expected onthe basis of their own earnings. That is to say,having a high-income spouse enabled some ofthese women to have lower incomes them-selves. Adjusted for potential lifetime earn-ings—earnings if individuals worked a fullcareer—Social Security’s total progressivity isreduced by another 20 percent, so that just 2.5percent of total benefits are redistributed.

To sum up, Social Security’s true progressiv-ity is just one-fifth of what it seems to be on thesurface. For that reason, Gustman andSteinmeier state:

It is clear from these results that the generalperception that a great deal of redistributionfrom the rich to the poor is accomplished bythe progressive Social Security benefit for-mula is greatly exaggerated. As a result,

adoption of a Social Security scheme withindividual accounts designed to be neutralwith regard to redistribution would makemuch less difference to the distribution ofSocial Security benefits and taxes amongfamilies with different earnings capacitiesthan is commonly believed.52

In fact, the plans adopted by the commissionincluded personal accounts, but these accountswere not neutral with regard to progressivity andmade Social Security a better deal for the poorand for other vulnerable members of society.Although Peter Coy argued that “attempts toinsulate the poor from the regressive features ofa privatized Social Security system wouldrequire so many awkward compromises that itmight leave no one happy,” the commission’splans enhanced progressivity in ways that wouldbe seamless from the point of view of the indi-vidual and the administrators of the program.

Moreover, though Coy acknowledged thataspects of the commission’s proposals would atthe least maintain system progressivity—suchas the progressive funding of the accountsthemselves, in Plans 2 and 3—he argued thatsuch provisions might not “survive the bud-geters’ axe” to be enacted.53 But this objectioncan be applied to any reform proposal, as wellas to current law, because Congress changeslaws as it wills. Coy’s is simply not an intellec-tually respectable objection to the commis-sion’s progressive reform plans, which weredesigned from the bottom up with progressivi-ty in mind. The commission’s plans withouttheir provisions to enhance retirement securityand wealth building for the least advantagedAmericans are simply not the commission’splans; to pretend otherwise is to portray a cari-cature of the commission’s proposals.

In addition, Coy argued, “It’s not clear, how-ever, that those offsets alone, even if they sur-vived the budgeters’ ax, would leave SocialSecurity as progressive as it is today.” Althoughcomprehensive measures of progressivity arecomplex, an easy shorthand measure is simplyto compare benefits received by low-wageretirees with those received by high-wageretirees.

As Table 2 shows (see p. 13), under currentlaw, a low-wage retiree in 2022 would receivebenefits equal to 46 percent of those received bya high-wage retiree.54 Under Plan 1, the current

Few wish toreturn to pay-as-you-go financ-ing, whichentails a 50 percent increasein payroll taxrates to meetcurrent benefit promises.

17

46 percent progressivity ratio would be main-tained, while under Plan 2 it would increase sub-stantially to 56 percent, and under Plan 3 to 50percent. It is worth noting that although Plan 1would not enhance progressivity per se, benefitswould be higher for all recipients.

Some critics argue, of course, that althoughSocial Security’s retirement and survivors pro-grams may not be progressive, its disabilityprogram disproportionately benefits the poor,thereby making up the difference. The commis-sion noted in its interim report that the poor andminorities are more likely to receive SocialSecurity’s disability payments.55 But even whendisability benefits are counted, a low-incomesingle male still receives a lower return fromSocial Security than does a high-income single-earner couple, according to Social Security’sactuaries.56 Moreover, changes to SocialSecurity’s retirement program, such as person-al accounts, need not reduce the progressivityof the disability program. Although the com-mission’s treatment of disability provisions isdiscussed at greater length below, under thecommission’s plans the special provisions forlow-wage workers apply to their disability aswell as their retirement benefits, so disabledworkers would not be disadvantaged relative tohigher wage earners.

The Commission’s ReformPlans—CommonCharacteristics

In its report to the president in December, thecommission outlined three reform proposals.Although they all contain personal accounts,beyond that they cover a spectrum of reformoptions. Nonetheless, the plans have a numberof features in common.

Optional Personal Accounts

Under all three models, workers would havethe option to invest part of their payroll taxes ina personal retirement account. Note that thesepersonal accounts would be voluntary. Noworker would be forced to take an account, andno worker with an account would be forced toinvest in the stock market. Workers seekingextra security could invest solely in govern-ment or corporate bonds.

For simplicity’s sake, Social Security’s actu-aries estimate personal account benefits underthree stylized portfolios:

1. The standard portfolio is assumed to consistof 50 percent stocks (with an annual rate ofreturn of 6.5 percent after inflation); 30 per-cent corporate bonds (3.5 percent annualreturn); and 20 percent government bonds(3.0 percent annual return). Assumingadministrative costs of 0.3 percent of assetsmanaged, the net annual return is assumedto be 4.6 percent after inflation.

2. The low-yield portfolio is assumed to holdonly government bonds, with a yield, netof administrative costs, of 2.7 percentannually.

3. The high-yield portfolio assumes that 60percent of the portfolio is invested in equi-ties or, alternately, that the equities held inthe default portfolio return the historicalaverage of 7.1 percent after inflation ratherthan the assumed return of 6.5 percent.

Analysis of the three reform plans produced bySocial Security’s actuaries assumes that aworker holds a particular portfolio throughouthis working lifetime. More realistically, work-ers would likely adjust their portfolios as theyage, beginning with greater stock allocationsand moving toward fixed income investmentsas they near retirement. Most workers aged 60to 65 hold approximately 40 percent of their401(k) account portfolios in equities.57 Assetallocations could be expected to differ some-what with personal accounts for SocialSecurity, but the trend from equities to fixedincome investments over the course of the lifecycle should be expected to continue.

Under all three commission plans the per-sonal accounts would be the property of theworker: The government could not “raid” it topay for other programs, and it could be passedon to the worker’s heirs in case of prematuredeath. The ownership aspect of personalaccounts is of particular benefit to AfricanAmerican males, one-third of whom do notsurvive to age 65.

Common Administrative Structure

All three plans would use a centralized andsimplified administrative structure similar to

Even after twoyears of weak

market perform-ances public

support for per-sonal accounts

remains strong.

18

Much of theprogram’sapparent redis-tribution isfrom the richerto the poorerspouse living inthe same house-hold, rather than from an upper-income household to alower-incomehousehold.

19

that of the federal Thrift Savings Plan, whichwould keep administrative costs to just 0.3 per-cent of assets managed while simplifyingaccount management for first-time investors.Account holders would choose from a range ofsimplified index funds; no individual stocks orsector funds (such as the NASDAQ) could bechosen, which would ensure adequate diversifi-cation for long-term investment purposes.

Cost of Living Adjustments (COLAs)Maintained

All three proposals would maintain annualadjustments to traditional benefits to maintainpurchasing power in the face of inflation. Inaddition, the actuaries’ analysis assumes that, atretirement, workers would convert their entireaccounts into fixed or variable annuities,though the commission made no formal recom-mendation that workers be required to do so.58

Benefit numbers cited herein assume conver-sion to a fixed annuity that pays constant bene-fits for life, adjusted annually for inflation.Workers who chose variable annuities wouldreceive benefits 4 to 9 percent higher than thosewith fixed annuities, assuming that the annu-ities are invested in the default portfolio of 50percent stocks, 50 percent bonds.59

Offset Interest Rate

A worker choosing to invest part of his pay-roll taxes in a personal account accepts an off-set to his traditional Social Security benefitsequal to the amount of his account contribu-tions compounded at the designated offsetinterest rate (3.5 percent, 2 percent, and 2.5 per-cent for Plans 1, 2, and 3, respectively). As longas a worker’s account interest rate exceeds theoffset interest rate, the worker will be assuredof receiving higher total benefits by taking theaccount. In two of the three plans, a workerwould receive higher total benefits simply byinvesting in government bonds, which areassumed to return 3 percent after inflation. Thisoffset is applied up-front, at the time contribu-tions are made to the account, and simply rep-resents the decision to invest those contribu-tions in the account (instead of investing themin the traditional system and earning anassumed interest rate of 3.5, 2, or 2.5 percent).There is no diminution of personal account bal-ances at the point of retirement as a conse-quence of the offset.60

Minimum Benefit

Plans 2 and 3 incorporate new minimum ben-efit guarantees, such that by 2018 a minimum

$1,000

$500

$666$750

$0

$200

$400

$600

$800

$1,000

$1,200

Couple Benefit Widow Benefit,Current Law (low)

Widow Benefit,Current Law (high)

Widow Benefit,Plans 2 & 3

Figure 3Lower-Income Survivors Benefit Increased from 32 to 50 Percent Compared with CurrentLaw

Source: President’s Commission to Strengthen Social Security, final report, p. 107.

wage worker would be assured of a benefit equalto 120 percent of the poverty line in Plan 2, or100 percent of the poverty line in Plan 3. UnderPlan 2, the minimum benefit would rise annual-ly with inflation. Under Plan 3, the minimumbenefit, while initially lower than under Plan 2,would rise at the faster rate of wage growth.One-half million to one million seniors could belifted out of poverty by virtue of these new pro-tections, according to Social Security’s actuaries.

Increased Survivors Benefit

Plans 2 and 3 also increase survivors benefitsto 75 percent of the couple’s prior benefit, forbelow average-income widow(er)s (see Figure3). For a couple with equal incomes, this wouldmean a 50 percent increase in benefits to thewidow. An estimated 2 to 3 million widowswould receive increased benefits as a result ofthis new provision.

Assets Split in Divorce

All three plans dictate that account assets besplit in the event of divorce (see Figure 4). Undercurrent law, before 10 years of marriage a spouseis not eligible to receive any benefits on the basisof the husband’s or wife’s earnings. Under thereform plans, that spouse would receive half the

husband’s or wife’s account assets. These assets,if simply left to accumulate until retirement,could result in lump sums ranging from $10,000to $40,000 for the spouse of an average-wageworker. At retirement, the account could increasethe spouse’s monthly income by $55 to $215.Given that women divorced at an early age areamong those most vulnerable to poverty inretirement, this provision could assist a particu-larly vulnerable group.

Commission Plan 1

Plan 1 was designed as a flexible frameworkto demonstrate the power of personal accounts,absent any other considerations. In other words,it is an “accounts only” plan that does nothing tothe current Social Security program other than togive workers the opportunity to invest throughpersonal retirement accounts. This is both itsstrength and its weakness. On the one hand, Plan1 shows the power of personal retirementaccounts to increase Social Security’s financialrate of return and to give workers the rewards ofpersonal asset accumulation. Moreover, a num-ber of different account options could be inte-grated into Plan 1’s basic framework.

On the other hand, Plan 1 makes no attemptto address larger system-solvency issues. And

No workerwould be forced

to take anaccount, and noworker with anaccount would

be forced toinvest in the

stock market.

20

$10,733

$19,932

$15,203

$20,944

$25,112

$34,895

$41,800

$23,876

$18,791

$-

$5,000

$10,000

$15,000

$20,000

$25,000

$30,000

$35,000

$40,000

$45,000

Low Medium High Low Medium High Low Medium High

Figure 4Lump Sums Available at Age 67 to Spouse of Average Earner Divorced after 10 Years

Source: Author’s calculations. Assumes marriage at age 25 to average earner, no account contributions during mar-riage, and divorce just before 10 years of marriage. Account assets split at divorce and compounded until age 67.

Investment Returns

Plan 1 Plan 2 Plan 3

Under all threecommissionplans the per-sonal accountswould be theproperty of theworker: Thegovernmentcould not “raid”it to pay forother programs,and it could bepassed on to theworker’s heirsin case of pre-mature death.

21

although solvency is by no means the only cri-terion of a successful reform proposal, it is avery important one. Plan 1, while bringingSocial Security marginally closer to solvencyand long-term sustainability, does not go nearlyfar enough to satisfy that important criterion.

Plan 1 includes an optional account intowhich workers could invest 2 percentage pointsof their taxable wages. As the commissionersnoted, Plan 1 presents a flexible framework:

The accounts could be made larger, orsmaller. They could be funded in a pro-gressive fashion (with a higher contribu-tion rate based on the first dollars of earn-ings than on higher earnings amounts).Some have proposed that such accounts besupplemented with extra contributions foryounger workers or that such accounts befunded from general revenues. . . . Othershave suggested that the accounts be madelarger, with the requirement that a certainamount be invested in federal securities asa means of limiting the total size of thetransition investment. Though the planscored here envisions a 2 percent accountfor all wage earners, any of the above vari-ations could be fit within this framework.61

As a flexible structure, Plan 1 leaves open

the possibility of the account being funded asan “add-on” with general revenues rather than a“carve-out” with payroll taxes.

In exchange for the personal account, indi-viduals would accept an offset to their tradi-tional benefits at an interest rate of 3.5 percent.This offset rate, being higher than the bond rate,means that account-holders must accept someminimal risk to receive higher expected bene-fits than under the traditional system.

Under Plan 1, a low-wage married workerretiring in 2052 could expect total benefitssome 5 percent higher than the current systempromises (see Figure 5). I use such a worker toillustrate benefit levels under all three commis-sion reform models because low-wage workersillustrate the special protections built into Plans2 and 3. Because opponents of personalaccounts claim that the poorest would be leftbehind, it is useful to show that this is not nec-essarily so. As a note, single workers wouldreceive benefits approximately 10 percent high-er from their accounts, as they would not berequired to purchase joint-and-survivors annu-ities providing spousal coverage.

In financing terms, Plan 1 is clearly a mixedbag. If funded as a “carve-out,” Plan 1 increas-es Social Security’s 75-year actuarial deficit by0.32 percent of payroll, so in theory it makesthe system “worse off” over the measurement

$637$713

$940$1,032 $1,052

$986

$-

$200

$400

$600

$800

$1,000

$1,200

Today's Benefit CurrentProgram"Payable"

Low-Yield Medium Yield High Yield CurrentProgram

"Promised"

Figure 5Monthly Benefits for Low-Wage Retiree under Plan 1

Source: Stephen C. Goss and Alice H. Wade, “Estimates of Financial Effects for Three Models Developed by thePresident’s Commission to Strengthen Social Security,” Memorandum dated January 31, 2002, p.74

period. However, this actuarial decline is pure-ly a function of timing. In the early years,account contributions are counted as lostincome and it is only when workers withaccounts actually retire that the account offsetreduces liabilities to the traditional system andthereby results in cost savings. If funded as an“add-on,” the general revenue requirementswould be identical.

Moreover, by the year 2042, Plan 1 would becheaper than the current system and wouldremain cheaper thereafter, while paying higherexpected benefits to all retirees (see Figure 6).Social Security would still be running annualpayroll tax deficits as of 2075, which merelyshows that a 2 percent account is not enough tofix the current program, but these deficitswould be 24 percent smaller than under the cur-rent system.62 More comprehensive measuresof the three plans’ financing are available in theappendices.

For those who argued that personal accountsby themselves hurt Social Security rather thanhelp it, Plan 1 comes as a surprise.63 It reducesthe size of general revenue infusions needed topay full benefits by 8 percent, versus the cur-

rent program, while paying higher benefits toall retirees, and by 2042 its costs are perma-nently lower than those of the current system.

Since Plan 1 is by design a flexible frame-work, Figure 7 shows the impact of a largeraccount funded with 6 percent of payroll, com-pared with the 2 percent account of Plan 2 andthe current law. The results are exactly as wouldbe expected: the up-front costs are larger, as thegovernment must fund the larger personalaccounts while simultaneously maintainingbenefits to current retirees, but once the pro-gram “turns the corner” the savings are greateras well. Moreover, total expected benefits forworkers would also be proportionately higher.

In short, Plan 1 has the following characteristics:

• Workers may voluntarily invest 2 percentof their taxable wages in personalaccounts.

• Traditional Social Security benefits are off-set by the worker’s personal account con-tributions compounded at an interest rate of3.5 percent above inflation.

• No other changes are made to traditionalSocial Security.

All three plansdictate that

account assetsbe split in the

event of divorce.

22

10

11

12

13

14

15

16

17

18

19

20

2001

2004

2007

2010

2013

2016

2019

2022

2025

2028

2031

2034

2037

2040

2043

2046

2049

2052

2055

2058

2061

2064

2067

2070

2073

Total Cost Rate with 2% Accounts

Current Law Cost Rate

Figure 6Cost of Plan 1 Compared with Cost of Current System

Source: President’s Commission on Strengthening Social Security, final report, p. 115.

In financingterms, Plan 1 isclearly a mixedbag.

23

• All retirees could expect higher benefitsthan under the current program.

• Beginning in 2042, Plan 1 would cost lessthan the current program.

• Additional revenues would be needed tokeep the trust fund solvent beginning in the2030s.

• Although Social Security’s finances wouldbe improved, the program would still notbe sustainable over the long term.

Commission Plan 2

Broadly speaking, the commission’s Plan 2reflects the outlook that Social Security shouldmaintain a real inflation-adjusted foundation onwhich other retirement savings could build, butthat this foundation should not grow at a rateunsustainable under current payroll tax rates. Inother words, it considers the adequacy of thereal, inflation-adjusted resources provided toretirees more important than other considera-tions, particularly income replacement rates.This point of view generated controversy, but ithas attracted much support in the past, evenfrom account opponents. Plan 2 was preferredby the most of the commissioners and would

be, from my point of view, the preferred direc-tion for Social Security reform.

In Plan 2 the commission analyzed the rateof benefit growth the current program can sus-tain without raising taxes, regardless of whetherpersonal accounts are introduced. This afford-able rate of benefit growth is slightly faster thanthe rate of inflation. Plan 2, then, ensures thatbenefits for all workers at least keep pace withinflation and, for lower wage workers, rise at afaster rate. It is important to understand thatPlan 2 incorporates Social Security’s affordablerate of benefit growth, whether or not personalaccounts are integrated into the program.Hence, steps taken in Plan 2 to restrain benefitgrowth to affordable levels are not due to theintroduction of personal accounts but to theinherent fiscal constraints faced by the currentprogram. Personal accounts in Plan 2 are not asource of the program’s fiscal limitations butare introduced to overcome the current pro-gram’s inherent fiscal limitations and allow thepayment of total retirement benefits substan-tially higher than the current pay-as-you-gosystem is capable of paying.

The commission’s Plan 2 allows each work-er to invest 4 percentage points of his payrolltaxes in a personal account, up to an annual

10

11

12

13

14

15

16

17

18

19

20

2001

2004

2007

2010

2013

2016

2019

2022

2025

2028

2031

2034

2037

2040

2043

2046

2049

2052

2055

2058

2061

2064

2067

2070

2073

Total Cost Rate with 2% Accounts

Total Cost Rate with 6% Accounts

Current Law Cost Rate

Figure 7Plan 1 Using 6% Accounts versus 2% Accounts Compared with Current Program

Source: Author’s calculations.

maximum of $1,000. Plan 2 therefore creates aprogressive personal account, with relativelylarger contributions for lower income workers.

In exchange, the worker would forgo tradition-al benefits at an offset interest rate of 2 percent.Because the offset rate is below the governmentbond rate of 3 percent, workers can increase theirtotal retirement benefits merely by investing inrisk-free government bonds, which are fullybacked by the government and are the legal prop-erty of the holder. Since Plan 2’s low offset ratemeans that workers can increase their benefitswithout risk, its progressive account structure par-ticularly benefits low-wage workers.

Under Plan 2, a low-wage worker retiring in2052 and holding the standard investment port-folio would receive $1,050 per month versusthe $986 per month promised by the currentsystem and the $713 that could actually be paid(see Figure 8). Even if that worker held onlygovernment bonds in his account, he wouldreceive a total monthly benefit of $867 (in 2001dollars), 22 percent higher than would be paidunder current law. 64

This point is relevant to the charge that thecommission reform plans would “cut guaran-teed benefits.” What critics reference as SocialSecurity’s “guaranteed benefit” is not guaran-teed in law, as the Supreme Court has ruled.65

Nor is it guaranteed in an economic or financialsense because the resources will not be avail-able to provide promised benefits. Quite thecontrary: We can be reasonably sure legislatedresources—current payroll tax rates—won’tprovide nearly what has been promised. Hence,the proper baseline for “guarantees” is what thelaw provides and what payroll taxes can affordto pay: that is, the so-called “payable benefit.”On this basis, there is simply no truth to claimsthat Plan 2 would “cut guaranteed benefits.”

Moreover, a worker who opted for a person-al account would have a stronger “guarantee”to his benefits in that, unlike the current pro-gram’s benefits, the account assets are his legalproperty and, in the case of government bonds,are backed by the full faith and credit of thegovernment.

Plan 2’s benefit increase for low-wage work-

Plan 1 reducesthe size of gener-

al revenue infu-sions needed to

pay full benefitsby 8 percent,while paying

higher benefitsto all retirees,

and by 2042 itscosts are perma-

nently lowerthan those of the

current system.

24

$637

$713

$867

$1,050$1,090

$986

$-

$200

$400

$600

$800

$1,000

$1,200

Today's Benefit Current Program"Payable"

Low-Yield Medium Yield High Yield Current Program"Promised"

Figure 8Monthly Benefits for a Low-Wage Retiree under Plan 2 (2052)

Source: Stephen C. Goss and Alice H. Wade, “Estimates of Financial Effects for Three Models Developed by thePresident’s Commission to Strengthen Social Security,” Memorandum dated January 31, 2002, p. 75.

Plan 2 reflectsthe outlook thatSocial Securityshould maintaina real inflation-adjusted foun-dation on whichother retirementsavings couldbuild, but thatthis foundationshould not growat a rate unsus-tainable undercurrent payrolltax rates.

25

ers is not simply a function of throwing moneyat the problem. Although Plan 2 pays low-wageworkers higher retirement benefits than thosepromised by the current program, it does so at acost 68 percent lower than that needed to main-tain the current system.

One reason for this is that under Plan 2, high-er wage earners do not do quite as well as low-wage earners. A high-wage worker with thedefault portfolio, for instance, would receive just88 percent of his full promised benefit, and anaverage-wage worker 94 percent. Nevertheless,even high-wage earners receive 25 percent morethan the current system can afford to pay whileavoiding the massive income-tax increasesmany account opponents favor to keep the cur-rent program solvent.66

Some may charge that the general revenueinfusions required to finance the transition topersonal accounts are unaffordable. However,concern over fiscal pressures should be a rea-son to favor the commission’s plans, not rejectthem. Account opponents who decry the rev-enue transfers under the commission’s plansshould detail how they would afford the muchlarger general revenue transfers necessary tomaintain the current program. It is only ifaccount opponents choose the true “do noth-

ing” option, in which Social Security becomesinsolvent and large benefit cuts are enacted, thatpressures on general revenue will be less thanunder the commission’s proposals.

To balance the current program, beginning in2009 Plan 2 indexes the initial wages eachcohort receives to the growth of prices, insteadof the generally higher rate of wages indexedunder current law (see Figure 9). Price indexingbrings Social Security back to solvency andmakes it sustainable over the long term.

This shift from wage to price indexing of ini-tial benefits has generated controversy andmerits a close look.

Social Security’s current benefit formulareplaces a higher percentage of low-wageworkers’ pre-retirement earnings than that ofhigher earners. For workers retiring in 2002,Social Security replaces 90 percent of the first$592 in average monthly pre-retirement earn-ings, 32 percent of monthly earnings between$592 and $3,567, and 15 percent of any amountbetween $3,567 and the taxable maximum. Thebenefit formula’s pairings of monthly incomelevels and income replacement rates are knownas “bend points” (see Figure 10).

For instance, a new retiree with $20,000average indexed earnings would receive $877

10

11

12

13

14

15

16

17

18

19

20

2001

2004

2007

2010

2013

2016

2019

2022

2025

2028

2031

2034

2037

2040

2043

2046

2049

2052

2055

2058

2061

2064

2067

2070

2073

Current System Cost Rate Reform Model 2 Total Cost Rate

Figure 9Cost of Commission Plan 2 Compared to Current System

Source: President’s Commission to Strengthen Social Security, final report, p. 126.

per month, or 53 percent of his pre-retirementmonthly earnings. A $50,000 worker, by con-trast, would receive $1,567 per month.Although this is almost twice what the $20,000worker receives, it is only 38 percent of the$50,000 worker’s pre-retirement earnings. Thisprogressivity is intentional: higher wage work-ers can save more outside of Social Security,and those outside assets can provide incomeduring retirement.

To maintain progressivity, Social Security’s“bend points” are increased annually accordingto the growth of wages. In 2001, for instance,Social Security’s 90 percent bend point wasplaced at the first $561 of a new retiree’s averageindexed monthly pre-retirement earnings, whilein 2002 the 90 percent bend point rose to $592.By 2015 the 90 percent replacement level willrise to the first $661 in earnings (in today’s dol-lars), and so forth ($767 by 2030 and $991 by2050). The upper bend points, marking replace-ment rates of 32 and 15 percent, are similarlyindexed to wage growth and increase each year.

Put another way, a $20,000 worker wouldretire today with monthly benefits of about$877, equal to 53 percent of his pre-retirementearnings. By 2050, the same worker earning thesame $20,000 (in today’s dollars) wouldreceive $1,091 monthly, or 64 percent of hispre-retirement earnings. In other words, the2050 retiree would receive 25 percent higherbenefits simply due to the passage of time, evenif he paid precisely the same taxes.

Moreover, under the current benefit formula,future increases in benefits will be greatest forthe highest earners (see Figure 11). For instance,a worker earning $10,000 today already receivesthe maximum 90 percent replacement rate onmuch of his wages. A similar worker making$10,000 (in 2002 dollars) in 2031 will receive a25 percent real benefit increase, and in 2071 a 58percent increase. By contrast, higher incomeworkers today have much of their wages coveredunder the upper bend points offering lowerreplacement rates. Over time, wage indexingpushes more of those wages into the bend points

There is simplyno truth to

claims that Plan2 would “cut

guaranteed benefits.”

26

0

200

400

600

800

1000

1200

1400

1600

1800

2000

0 1000 2000 3000 4000 5000 6000

Figure 10Social Security’s “Bend Point” Formula Provides Higher Monthly Benefitsto Those with Lower Average Wages

Average Indexed Monthly Earnings

Source: Social Security Administration.

Account oppo-nents who decrythe revenuetransfers underthe commis-sion’s plansshould detailhow they wouldafford the muchlarger generalrevenue trans-fers necessary tomaintain thecurrentprogram.

27

offering higher replacement rates. For instance, a$100,000 worker retiring at 65 in 2001 canexpect to receive about $330,000 in lifetimeretirement benefits. By 2031, lifetime benefitsincrease by 51 percent to $499,000, and by 2071by 120 percent to $725,000, under today’s(unsustainable) benefit formula.67

It should be stressed that these are workersearning the same annual wages in real, infla-tion-adjusted dollars. Clearly, the current wage-indexed benefit formula provides vastlyunequal benefits to otherwise identical workersretiring at different points in time. Wage index-ing increases benefits over time, and the largestincreases go to the highest earners. In fact,these figures underestimate differences in life-time benefits, as they assume identical lifeexpectancies for upper and lower wage work-ers. Because life expectancies are correlated toincome, the true results are likely to be evenmore extreme.68

Moreover, as the commission noted in itsinterim report, wage indexing of initial benefitsprevents the system from gaining much fromincreases in long-term economic growth.

Initial Social Security benefit levels are cur-rently indexed to the growth in national wage

levels. Consequently, even if there were nodemographic problem, Social Security costswould grow almost as fast as the economy asa whole. Faster economic growth means moretax contributions in the short term, and higherbenefit obligations in the long term. Thoughthis faster growth helps, it does far less thanmany people believe. Mostly it creates theillusion of improvement because short-termrevenue gains postpone the projected date ofTrust Fund depletion, whereas increased costswould occur mostly after the projected deple-tion date.69

Even if the economy grew twice as fast asthe trustees project, Social Security would stillbecome insolvent within the 75-year scoringperiod.70 Social Security’s wage indexing for-mula is in large part responsible for that.

A solution to the problems of financing andequity is to switch from wage to price indexing ofSocial Security’s benefit bend points. In doing so,the bend points would increase annually alongwith increases in the Consumer Price Index. 71

Under Plan 2, price indexing is instituted begin-ning in 2009, so current retirees and those overage 55 are entirely unaffected, and those underage 55 are affected only to the degree that their

0

100000

200000

300000

400000

500000

600000

700000

800000

2001 2011 2021 2031 2041 2051 2061 2071

$100,000 annual wage$50,000 annual wage$10,000 annual wage

Figure 11Wage Indexing Gives High-Wage Workers Largest Increase

Year of Retirement

Source: Author’s calculations.

working careers extend past 2009.72 In addition,Plan 2 is not a “pure” price indexing approach, asit contains special provisions targeted toward themost vulnerable Americans—low-wage workersand widows—so that they receive higher benefitsthan the current benefit formula promises, muchless can actually afford to pay. Whatever the large-scale financing impact of the switch to priceindexing, it is difficult to portray this principle asunfair; it simply treats relevantly similar individu-als in a similar way.

These facts blunt criticisms, such as thosemade by Kilolo Kijakazi and Robert Greensteinof the CBPP, that “replacing ‘wage indexing’with ‘price indexing’ would result in deepreductions over time in Social Security bene-fits.”73 As we have seen, under a pure priceindexing approach, workers with similar wageswould receive similar benefits, at whateverpoint in time they retired.

Greenstein and Kijakazi are correct that priceindexing reduces the replacement rate for anaverage-wage worker in any given year,74

which they illustrate with an average-wageworker retiring in 2040: Under the currentwage indexing approach, that worker would bepromised a benefit equal to 37 percent of hispre-retirement earnings. Under pure priceindexing, they say, his benefit would equal only28 percent of his pre-retirement earnings.Greenstein and Kijakazi extend their exampleto a worker retiring in 2070, making the notion-al “cuts” even larger.

Several points are worth making. First, undercurrent law, after trust fund depletion in 2038benefits would equal whatever level payroll tax

receipts are capable of paying, regardless of thewage-indexed benefit formula. Thus, under cur-rent law that 2040 retiree will actually receive a27.4 percent replacement rate even if priceindexing weren’t introduced. Moreover, underPlan 2 the personal account provides retireeswith benefits higher than those that a pure priceindexing approach implies. The 2040 retireecould expect total retirement benefits equal toabout 34.8 percent of pre-retirement earnings,assuming investment in the default 50 percentstock, 50 percent bond portfolio.75

Beyond this point, there is a more importantdifference between average-wage workers inthe future and average-wage workers today:average workers in the future will earn substan-tially more money. The average-wage workerin 2040 will earn some 48 percent more thanthat of today; in 2070, almost double today’saverage (and more than 20 percent greater thanthe SSA’s hypothetical “high earner” in 2001).Compared with today’s workers, these futureAmericans will have substantially higher stan-dards of living while working and several yearsgreater life expectancy in retirement.

Now, workers retiring today with earnings50 to 100 percent higher than the averagewould of course receive a lower incomereplacement rate than would an average-wageworker, and Greenstein and Kijakazi wouldsurely be the first to defend that practice. Yetthey argue that treating a similar worker in thefuture in a similar way is a “deep reduction.”

To be fair, Greenstein and Kijakazi acknowl-edge that the “benefit cuts” inherent in priceindexing would not, in fact, actually cut bene-

Table 3Monthly Benefit Payable to Low-Wage Disabled Worker Scheduled to Retire in 2055 (in 2001 dollars)

Promised $986Plan 1 $986Plan 2 $807Plan 3 $857Payable $713Benefit payable to low-wage worker in 2001 $637

Source: Derived from data presented in Stephen C. Gossand and Alice H. Wade, “Estimates of Financial Effects for ThreeModels Developed by President’s Commission to Strengthen Social Security,” Memorandum dated January 31, 2002.

Price indexingbrings Social

Security back tosolvency and

makes it sustain-able over the

long term.

28

The currentwage-indexedbenefit formulaprovides vastlyunequal bene-fits to otherwiseidentical work-ers retiring atdifferent pointsin time.

29

fits: “To be sure,” they say, “benefit levelswould keep pace with changes in prices.”76 Inother words, benefits would not be cut. But,they say, “beneficiaries would be precludedfrom partaking in the general increase in thestandard of living from one generation to thenext. Upon retiring, workers would essentiallydrop back to a standard of living prevalent in anearlier generation.”77

Of course, price indexing also precludesworkers from “partaking” in the 50 percentincrease in payroll tax rates necessary to main-tain the current wage-indexed benefit formula.The question, which Greenstein and Kijakazidon’t address, is whether workers should beforced to pay those extra taxes to achieve prom-ised replacement rates, particularly when indi-viduals desiring a higher retirement incomecould invest their tax savings privately at ratesof return two to three times higher than underSocial Security. Why must the governmentforce people to do something they could easilydo voluntarily if they so wished?

In effect, Greenstein and Kijakazi argue that a$100,000 annual wage worker retiring in 2070should receive some $725,000 in lifetime retire-ment benefits, even if payroll taxes must top 19percent, despite the fact that a similar workerretiring today can expect just $330,000 in life-time benefits. But why? If it is so important togive a $100,000 worker $725,000 in lifetimeretirement benefits, why not raise taxes today?Greenstein and Kijakazi argue that tomorrow’staxpayers should shoulder a tax burden that theyare unwilling to ask today’s taxpayers to bear,but provide neither the economic nor philosoph-ical rationale for doing so.78

The question remains: if higher earnerstoday receive lower replacement rates becausethey are able to save more outside of SocialSecurity, shouldn’t the same reasoning apply toidentical earners in the future? Isn’t this partic-ularly true given the financing burden thewage-indexed benefit formula places on thesystem and on the taxpayer?

In testimony before the president’s commis-sion, Hans Riemer of the Campaign forAmerica’s Future and the 2030 Center declaredthat “the level of guaranteed protection thatSocial Security provides today is about right.”79

Plan 2, and all of the other plans put forward bythe commission, would maintain the level ofbenefits provided today and enhance benefits

substantially for those in greatest need.Despite charges made by reform opponents

against price indexing today, in the past it hasreceived support even among those whooppose personal accounts. Peter Diamond ofMIT, one of the most prominent academic crit-ics of personal accounts, was a member of agovernment panel in the 1970s that recom-mended price indexing, calling it “fair and nec-essary.”80 As the panel’s report pointed out:

The wage-indexing method provides a sharptilt in favor of workers retiring in the future.The increases in benefits for workers alreadyretired are limited to increases in the rise inthe Consumer Price Index. Yet workers whoretire five years later will receive incrementsdue to both price changes and increases inreal wages. This difference in retirementbenefits can be substantial.81

When President Carter appeared to be favor-ing wage indexing over a price-indexedapproach, Diamond and the other panel mem-bers chided him for fiscal and generational irre-sponsibility:

President Carter would be displeased withhis predecessors if he were currently facedwith the choice of cutting Social Securitybenefits for present recipients or raising thesame amount of revenue as would be raisedby an increase in the payroll tax rate of fivepercentage points. Yet that is precisely whatthe best current estimates say he is proposingto do to some future President. . . . It appearsto us that correction of overindexing bychoice of a price indexing method would begreatly superior [to wage indexing]. . . . Useof the price indexing method would elimi-nate the need for a tax rise when the per-centage of retirees increases sharply earlynext century. . . . While the price indexingmethod implies protection from inflationand a growth in benefits with the real growthof the economy, the wage indexing methodcalls for a much larger growth in benefits forfuture retirees at a time when the countrymay not be able to afford it. Use of the priceindexing method would permit moderate taxand benefit increases to aid those recipientswith greatest need as perceptions of thoseneeds arise.82

The same charges could be made todayagainst those who wish to saddle future taxpay-ers with economic burdens they themselves areunwilling to bear today.

Henry Aaron of the Brookings Institutionconcurred with Diamond, arguing that priceindexing “would leave more options open forspending the productivity dividend of econom-ic growth. Congress could still raise pensions inthe future, but it could also decide that otherprograms such as housing, health insurance, ordefense have greater claims on availablefunds.”83 As chairman of the 1978–79 AdvisoryCouncil on Social Security, Aaron again arguedfor price indexing, but the change was notadopted.

As per capita income rises, the case forincreasing the amount of mandatory “sav-ing” for retirement and disability throughSocial Security is far weaker than was therationale for establishing a basic floor ofretirement and disability protection atabout the levels that exist today.

At the levels of real income prevailingin the 1930s (or perhaps even in the1950s), it can well be argued that it wasappropriate, indeed, highly desirable—perhaps even necessary for the preserva-tion of our society—that governmentshould, by law, have guaranteed to theaged and disabled and their dependentsreplacement incomes sufficient to avoidsevere hardship, and to have requiredworkers (and their employers) to financethis system with a kind of “forced saving”through payroll tax contributions. But asreal incomes continue to rise, it is not easyto justify the requirement that workersand their employers “save” through pay-roll tax contributions to finance ever high-er replacement incomes, far above thoseneeded to avoid hardship. Perhaps not allworkers will want to save that much, or tosave in the particular time pattern andform detailed by present law.

Aaron and his coauthors go on to say:

Future Congresses will be better equippedthan today’s Congress to determine the appro-priate level of and composition of benefits forfuture generations. . . . Congress might elect

to give more to certain groups of beneficiar-ies than to others, or to provide protectionagainst new risks that now are uncovered. Butprecisely because we cannot now forecastwhat form those desirable adjustments mighttake, we feel the commitment to largeincrease in benefits and taxes implied undercurrent law will deprive subsequentCongresses, who will be better informedabout future needs and preferences, of neededflexibility to tailor Social Security to theneeds and tastes of the generations to come.84

The primary difference between the priceindexing advocated by Diamond and Aaron andthat applied in Plan 2, besides certain technicalfactors of implementation, is that Plan 2 pro-vides a personal account to make up for thereduction in the growth of traditional benefits.

Subsequent to the release of the commis-sion’s recommendations, Diamond disownedhis prior support for price indexing and strong-ly criticized the commission for having advo-cated it.85 Diamond argued that he had favoredprice indexing in the 1970s because the financ-ing problems facing Social Security at the timewere much greater than those at present. It istrue that Social Security faced a short-termfinancing crisis in the late 1970s due to an errorin the benefit formula introduced in the 1972amendments that implied a quantum leap infuture benefit levels. But price indexing, whichis extremely slow to take effect, would havedone nothing to avert insolvency in the shortterm. Moreover, the wage-indexing alternativethat Diamond and the rest of the panel arguedso strongly against entailed long-term systemcosts barely higher than those projected bySocial Security’s actuaries today. 86 In addition,the Hsiao panel rejected the view that incomereplacement rates should be the basis of SocialSecurity’s benefit formula, as is the case underwage indexing.

To summarize Plan 2:

• Workers can voluntarily redirect 4 percentof their payroll taxes up to $1,000 annual-ly to a personal account (the maximumcontribution is indexed annually to wagegrowth). Traditional Social Security bene-fits are offset by the worker’s personalaccount contributions compounded at aninterest rate of 2 percent above inflation.

Price indexingalso precludesworkers from

“partaking” inthe 50 percent

increase in pay-roll tax ratesnecessary to

maintain thecurrent wage-

indexed benefitformula.

30

“As realincomes contin-ue to rise, it isnot easy to justi-fy the require-ment that work-ers and theiremployers ‘save’through payrolltax contribu-tions to financeever higherreplacementincomes, farabove thoseneeded to avoidhardship.”

31

• Workers opting for personal accounts canreasonably expect total benefits greaterthan those paid to current retirees, to work-ers without accounts, and to future benefitspayable under current law.

• Plan 2 establishes a minimum benefitpayable to 30-year minimum-wage work-ers of 120 percent of the poverty line.Survivors’ benefits for below-average-wage workers would be increased by 33 to50 percent.

• Beginning in 2009, calculation of tradition-al benefits would switch from wage index-ing to price indexing. Current retirees andworkers aged 55 and over would be entire-ly unaffected.

Commission Plan 3

Plan 3 aimed to match or exceed benefitscurrently promised by Social Security for allworkers, but to do so at lower cost than the cur-rent program. In this, it succeeds.

Plan 3’s accounts are designed as a combina-tion add-on and carve-out. That is to say, if aworker agrees to contribute an additional 1 per-cent of his earnings to the account, he may“carve out” 2.5 percentage points of his payrolltaxes up to an annual maximum of $1,000.

Workers opting for personal accounts wouldaccept an offset at a 2.5 percent interest rate.Again, for a low-wage worker investing in gov-ernment bonds, the guaranteed benefit is bothhigher and more “guaranteed” than under thecurrent system.

Under Plan 3, a low-wage worker with thedefault investment portfolio would receive$1,103 per month in total expected retirementbenefits, as opposed to the $986 promised bythe current program and the $713 actuallypayable (see Figure 12). All other workers, atall other times, could expect benefits at least ashigh as those promised by the current program.

To bring the traditional program back to bal-ance, Plan 3 requires new ongoing sources ofgeneral revenue, equivalent in size to increas-ing the payroll tax cap to 90 percent of taxablepayroll and redirecting to Social Security theportion of benefits taxation that currently flowsto Medicare.

Two points are worth making in this regard.First, it should be emphasized that Plan 3describes only the size of the general revenuetransfers and not the means. The commissionworked under the president’s principle that pay-roll taxes would not be raised, and this wastaken to encompass increases in the taxablewage base as well as the payroll tax rate.Second, increased revenue commitments were

$637$713

$943

$1,103 $1,137$986

$-

$200

$400

$600

$800

$1,000

$1,200

Today'sBenefit

CurrentProgram

"Payable"

Low-Yield MediumYield

High Yield CurrentProgram

"Promised"

Figure 12Monthly Benefits for Low-Wage Retirees under Plan 3 (2052)

Source: Stephen C. Goss and Alice H. Wade, “Estimates of Financial Effects for Three Models Developed by thePresident’s Commission to Strengthen Social Security,” Memorandum dated January 31, 2002, p. 76.

included for ensuring that benefits for allretirees exceeded the replacements rates prom-ised in the current benefit schedule. However, itis not clear why current replacement ratesshould be the standard many decades in thefuture, as the prior section on price indexingpoints out. Third, even if meeting promisedreplacement rates were the goal, given the rela-tive efficiencies of pay-as-you-go and fundedsystems it would make sense to meet this goalwith a larger personal account rather thanincreased taxation. Doing so would entailgreater prefunding, but on the whole that is adesirable thing.

One problem with Plan 3 is that mandatingadditional worker contributions to take advantageof the personal account might discourage lowerwage workers from taking part, even if the addi-tional contributions are partially subsidized by thegovernment. Indeed, experimental individualdevelopment account trials aimed at lower-income workers often have far from universal par-ticipation, even with substantially more generousmatches than included in Plan 3. 87 And accordingto the 1998 Retirement Confidence Survey con-ducted by the Employee Benefit Research, some40 percent of workers said they could not saveeven an extra $20 per week for retirement.88

Among low-wage workers, this percentage wouldlikely be even higher.

Adverse selection in terms of account partici-pation could complicate overall system financ-ing, and would also prevent lower wage workersfrom reaping the benefits of asset ownershipunder Plan 3. In terms both of participation ratesand administrative simplicity, the poor might dobetter if the progressive carve-out approach usedin both Plans 2 and 3 were simply expanded.This would make achieving actuarial balancewithin the 75-year window more challenging,but from a public policy perspective it would domore to enhance retirement income and assetaccumulation among the poor.

To balance system finances, Plan 3 makes sev-eral changes to Social Security’s benefit struc-ture. First, it would adjust benefits for futureretirees to account for increases in longevity. Inaddition, the benefit penalties for retiring early—and the rewards for working later—are increasedto encourage people to stay in the workforcelonger. Some would argue that these changesconstitute an increase in the retirement age.89 Thisis incorrect: under Plan 3, the normal retirement

age would remain the same as under current law.Thus, workers could still retire at any age past 62and still receive more than current law would paythem. It is ironic that the Campaign for America’sFuture has issued press releases attacking Plan 3’spurported increase in the retirement age, while intestimony before the commission Roger Hickeyof the CAF praised a reform proposal by HenryAaron and Robert Reischauer that explicitly rais-es both the normal and the early retirement ages.90

Moreover, Plan 3’s changes to benefitswould not even begin until 2009, so currentretirees and those over age 55 are not subject toany changes. For individuals retiring soon after2009, the changes would be tiny—just a pennyon the dollar—and that offset would be morethan made up by gains from the personalaccount.

Over the long term, Plan 3’s changes wouldpay higher benefits for a lower cost than underthe current program (see Figure 13). If the newgeneral revenue transfers were included, itwould bring Social Security back to cash sur-pluses and long-term sustainability for half thecost of maintaining the current program.

The commission’s treatment of disabilitybenefits has been received with some level ofconfusion. For instance, a memo prepared bythe staff of Rep. Robert Matsui (D-Calif.)alleged, “The President’s Social Security com-mission recommended cutting disability bene-fits to help pay for the cost of privateaccounts.”91 In fact, the commission made nospecific recommendations regarding the long-term financing of Social Security’s DisabilityInsurance program, which provides benefits toworkers who through illness or injury areunable to continue to work.

Under the commission’s proposals, a workerwith a personal account who becomes disabledwould receive the traditional Social Securitybenefit without any offset for holding theaccount. Only at retirement would he access hisaccount balance, and only at retirement wouldhis traditional benefit be offset on the basis ofhis account contributions. A worker whobecame disabled early in life would have madefew contributions, and thus would have a verysmall offset to his traditional benefits. But aslong as his account earned more than the offsetinterest rate, he could still anticipate highertotal benefits than if he had remained in the cur-rent program.

For a low-wageworker investing

in governmentbonds, the guar-anteed benefit isboth higher and

more “guaran-teed” than under

the currentsystem.

32

Mandatingadditionalworker contri-butions to takeadvantage of thepersonalaccount mightdiscouragelower wageworkers fromtaking part,even if the addi-tional contribu-tions are partial-ly subsidized.

33

For instance, a low-wage disabled workerscheduled to retire in 2055 would receive underPlan 1 disability benefits before retirementequal to those promised by the current systemand 30 percent higher than current law; Plan 2would pay benefits 13 percent higher than cur-rent law; and Plan 3, 20 percent higher thancurrent law. As these disabled workers wouldreceive more than current law could pay them,it is clear that disability benefits have not beencut to fund personal accounts.

Charges of “cuts” come about because dis-abled workers under the three commission pro-posals would receive the benefits dictatedunder those proposals, which under Plans 2 and3 are somewhere between those promised andthose actually payable under current law (seeTable 3). But, as noted, the “promised” benefitis not a valid benchmark of comparisonbecause no means of funding that promise isavailable at present. The current program can-not keep its promises past the year 2041 with-out increasing the payroll tax by 50 percent.Moreover, any reductions in promised benefitsare due to the system’s insolvency, not to theintroduction of personal accounts. Accounts aredesigned to make up for Social Security’sinability to pay promised benefits.

It is worth noting the complications facingany reform’s treatment of disability insurance.

On one hand, the Disability Insurance programis distinct from the Old Age and SurvivorsDisability Insurance program, having its owndedicated tax and its own trust fund. At thesame time, the two sides of Social Securityshare a common formula for calculating bene-fits, and changes to the calculation of retire-ment benefits can also affect the DI program.Moreover, the DI program faces financing chal-lenges even steeper than those of OASDI—without change, the DI program will run pay-roll tax deficits by 2009 and its trust fund willbe exhausted by 2028.

Lacking comprehensive disability reform,several options are open to reformers of theretirement program, none of them ideal:

• Separate the OASI and DI programs sothey run independently. At present,although the programs have separate taxesand trust funds, they share a common ben-efit formula. Although this would isolateDI from any changes to OASI, DI is evenmore severely underfunded and wouldtherefore become insolvent sooner.

• Retain the integration of OASI and DI butallow DI to continue under pre-reform cri-teria. While this would protect DI recipi-ents from any changes, it would also createan incentive for workers nearing retirement

1011121314151617181920

2001

2006

2011

2016

2021

2026

2031

2036

2041

2046

2051

2056

2061

2066

2071

Current System Cost Rate

Reform Model 3 Total Cost Rate

Figure 13Cost of Plan 3 Compared with Cost of the Current System

Source: President’s Commission to Strengthen Social Security, final report, p. 13.

to seek to qualify for disability benefits. Anincrease in DI applications could speed theprogram’s insolvency.

• Apply changes to OASI and DI universally,acknowledging that further steps must betaken to protect disabled workers as well asto reform the DI program in general.

It was the third option that the commission-ers took.

In doing so, the commission emphasized that“in the absence of fully developed proposals, thecalculations carried out for the commission andincluded in this report assume that defined bene-fits will be changed in similar ways for the twoprograms.” However, “this should not be taken asa commission recommendation for policy imple-mentation. . . . The commission recognizes thatchanges in Social Security’s defined benefit struc-ture and the role of personal accounts may havedifferent implications for DI and OASI benefici-aries. The commission urges the Congress to con-sider the full range of options available foraddressing these implications.”92 In other words,the commissioners anticipated that additionalsteps would be taken to address the DI program,and that these steps could require additional fund-ing as well as broader structural reform.

The commissioners agreed with the SocialSecurity Advisory Board, which declared, “Aftertwo additional years of study of the disability pro-grams. . . . we are convinced that the issues facingthe disability programs cannot be resolved with-out making fundamental changes.”93 Unfortu-nately, the deliberate focus of the president’s

commission on the financing problems of theretirement portion of Social Security and theshort period available for the commissions’ workmade the development of comprehensive dis-ability reform proposals impossible. The com-mission did recommend, however, “that thePresident address the DI program through a sep-arate policy development process.”

As the independent Social Security AdvisoryBoard has argued, the disability programrequires reforms extending beyond merefinancing changes.94 Decisions on disability eli-gibility vary greatly between states, with somestates approving DI claims at over twice therate of others, and are many times resolved onlythrough adjudication. Moreover, awards basedon mental conditions have more than doubledas a percentage of total DI claims between 1980and 1990, and now make up the largest singlereason for DI claims. State agency administra-tors and examiners report that at least half ofthe claims processed now involve issues relat-ing to mental impairment.95 Clearly, the natureof disability claims is changing quickly andradically, and the program requires a compre-hensive assessment of how it is to function andwhat clams should be met. The complexity ofDI’s structure and function causes its adminis-trative costs to be five times higher than thoseof OASDI relative to its income.96 Without adoubt, more changes will be needed to bring DIback to long-term health

To summarize Plan 3:

• Workers who invest an additional 1 percent

Plan 3’s changesto benefitswould not evenbegin until 2009,so currentretirees andthose over age 55are not subjectto any changes.

34

Table 3Monthly Benefit Payable to Low-Wage Disabled Worker Scheduled to Retire in 2055 (in 2001 dollars)

Promised $986Plan 1 $986Plan 2 $807Plan 3 $857Payable $713Benefit payable to low-wage worker in 2001 $637

Source: Derived from data presented in Stephen C. Goss, Chief Actuary and Alice H. Wade, Deputy Chief Actuary,“Estimates of Financial Effects for Three Models Developed by President’s Commission to Strengthen SocialSecurity,” Memorandum presented on January 31, 2002.

The deliberatefocus of thepresident’s com-mission on thefinancing prob-lems of theretirement por-tion of SocialSecurity and theshort periodavailable for thecommissions’work made thedevelopment ofcomprehensivedisabilityreform propos-als impossible.

35

of wages in a personal account may alsoinvest 2.5 percentage points of payrolltaxes, up to $1,000 annually. The add-oncontribution is progressively subsidized bya refundable tax credit, enhancing the pro-gressivity of the account.

•Account-holders accept an offset to their tra-ditional Social Security benefits at a realannual interest rate of 2.5 percent. Hence,under Plans 2 and 3, workers could increasetheir total retirement benefits by simplyinvesting in government bonds.

• Plan 3 increases system progressivity byestablishing a minimum benefit payable to30-year minimum-wage workers of 100percent of the poverty line (111 percent for a40-year worker). This minimum benefitwould be indexed to wage growth. Benefitsfor below-average-wage earners would beincreased as well.

• The growth rate of traditional benefitswould be adjusted to reflect increases inlife expectancy; the offset for early retire-ment and bonus for later retirement wouldbe increased; and the third bend point fac-

tor affecting higher-wage retirees would bereduced from 15 to 10 percent.

• Benefits payable to workers who do notopt for personal accounts would be morethan 50 percent higher than those paid totoday’s retirees.

• New sources of dedicated revenue wouldbe added in the equivalent amount of 0.6percent of payroll over the 75-year period,and continuing thereafter.

• Additional temporary transfers from gener-al revenues would be needed to keep theTrust Fund solvent between 2034 and2063. Total cash requirements would be 52percent of those needed to maintain thecurrent program.

Summary of the ThreePlans

Taken together, the commission’s threereform plans show the power of personalaccounts to address the problems associatedwith Social Security.

$637

$713

$1,032 $1,050$1,103

$986

$-

$200

$400

$600

$800

$1,000

$1,200

Today's Benefit Current Program"Payable"

Plan 1 Plan 2 Plan 3 Current Program"Promised"

Figure 14All Three Plans Would Pay Higher Benefits Than under Current Law

Source: President’s Commission to Strengthen Social Security, final report. Low-wage worker retiring in 2052;assumes investment in 50–50 stock–bond portfolio.

• Each of the three commission proposalswould pay future beneficiaries higherexpected benefits than those received bytoday’s retirees.

• Each of the three proposals would increaseexpected benefits relative to what the cur-rent system can pay.

• Each of the three proposals would increaseexpected benefits for those who opt forpersonal accounts relative to those whostay in the traditional system.

• Each of the comprehensive reform propos-als (Plans 2 and 3) would increase expect-ed benefits for those who opt for personalaccounts, even if participants invested inthe most conservative portfolio available(government bonds).

• Each of the three proposals would increaseexpected benefits for low-income partici-pants even relative to currently promisedbenefits (which cannot be paid without sig-nificant tax increases). These low-wageparticipants would receive higher benefitsthan if Social Security were fully fundedand faced no financial crisis whatsoever.Two of the proposals institute, for the firsttime, a guarantee that minimum-wageworkers not retire in poverty.

• Two of the three proposals would institutespecial benefit increases for widows oflow-income participants, raising incomesfor millions of Americans most vulnerableto poverty in retirement.

• Two of the three proposals would increaseexpected benefits for all of the participantsin personal accounts even relative to thebenefits the current system promises, muchless can afford to pay.

• One proposal makes no changes whatsoev-er to Social Security’s benefit structureother than to offer workers the opportunityto increase their total benefits by owning apersonal retirement account.

• Under all of the proposals, participantsreceive more benefits for less money rela-tive to the current system.

All three plans would pay higher benefitsthan are paid under the current system (seeFigure 14), and lower-income workers—thosewho account opponents claim to be most con-cerned about—receive benefits higher thanthose the current program even promises.

Moreover, all three plans would producethose benefits at lower general revenue coststhan the current program (see Figure 15).

Under Plans 2and 3, workerscould increasetheir total retire-ment benefits bysimply investingin governmentbonds.

36

$21.7

$10.4

$6.9

$20.0

$0

$5

$10

$15

$20

$25

Current Law Plan 1 Plan 2 Plan 3

7.7% Reduction

68.1% Reduction

52.2% Reduction

Figure 15All Three Plans Would Require Less General Revenue Than the Current Program

Source: President’s Commission to Strengthen Social Security, final report, p. 22. Based on current programfinances as projected in 2001 Social Security Trustees Report.

These low-wageparticipantswould receivehigher benefitsthan if SocialSecurity werefully fundedand faced nofinancial crisiswhatsoever.

37

Assuming future payroll tax surpluses areunavailable for transition financing, the transi-tion cost of Plan 1 is $1.1 trillion; Plan 2, $0.9trillion; and Plan 3, $0.4 trillion. If future pay-roll tax surpluses were made available to fundreform, as they should be, costs would declineto $0.7 trillion for Model 1, $0.4 trillion forModel 2, and $0.1 trillion for Model 3. Theselatter values equal 0.29 percent, 0.33 percent,and 0.10 percent of GDP during the scoringperiod, substantially less than the cost of main-taining the current program.

Even Plan 1, the “accounts only” approach,cuts 75-year general revenue costs by 8 percent,while Plan 2 reduces costs by 68 percent and Plan3 by 52 percent. In other words, the governmentcould devote twice as much extra money to thetraditional system and low-wage workers would

still receive lower benefits than they would underplans 2 and 3. The commission’s success was thatit showed how to deliver higher long-term bene-fits at lower long-term costs.

Finally, all three plans would leave SocialSecurity with substantially greater assets overthe long term. As noted previously, SocialSecurity’s actuaries opine that although person-al accounts are the property of individuals, foraccounting purposes their assets should betreated as part of total system assets. A compre-hensive view of Social Security’s asset positioncombines account balances with the balance ofthe traditional trust.

Table 4 details the increases in total systemassets as of 2076, the end of the traditional 75-yearactuarial scoring period. Clearly, any argumentsthat the commission’s personal account plans “de-

Table 4 Total System Assets as of 2076 under Current Law and Commission Reform Models(Present Value in Billions of Dollars, Discounted to January 1, 2001)

Likely OASDI Trust fund Current Personal Account

Participation Rate Assetsa and Annuity Assets Total

Present law -3,230 NA -3,230

Model 1 67% account

participation -3,826 1,080 -2,746

100% account

participationb -4,124 1,619 -2,505

Model 2 67% account

participation 380 1,290 1,670

100% account

participation 423 1,935 2,358

Model 3 67% account

participation 185 1,602 1,620

100% account

participation 270 2,401 2,671

Source: Based on table from Stephen C. Goss and Alice H. Wade, “Estimates of Financial Effects for Three ModelsDeveloped by the President’s Commission to Strengthen Social Security,” Memorandum dated January 31, 2002, p. 24. Thenet current accrual for future benefit offset equals the net future savings to the traditional system as the result of benefits off-set before the end of the scoring period.

a. Negative values are the OASDI unfunded obligation for the period 2001 through 2075.b. For Model 2, 67 percent participation is considered more likely if the benefit offset yield rate is computed as 2 percentabove the realized or expected inflation rate, but 100 percent participation is considered more likely if computed as 1 per-cent below the market yield on Treasury bonds. Based on the intermediate assumptions of the 2001 Trustees Report andother assumptions described in the text.

fund” Social Security are rebutted by these fig-ures. By 2076 the current Social Security programwould be underfunded by $3.23 trillion, in presentvalue terms. By contrast, even under Plan 1,which makes no structural changes to SocialSecurity’s revenue or benefit formulas other thanto incorporate optional personal retirementaccounts, Social Security’s deficit is reduced to$2.75 trillion, assuming that two-thirds of eligibleworkers choose to participate in personalaccounts. Under Plan 2, Social Security wouldhave net assets worth $1.67 trillion, and underPlan 3, $1.79 trillion. All of these figures are pres-ent values, that is, the amount those future lumpsum assets—or debts—would be worth to indi-viduals today.

Where from Here?

With the commission’s work done, wheredoes the Social Security debate go from here?The events of September 11, 2001, clearlyreshaped many of the government’s priorities,pushing national security concerns to the frontburner, and rightly so.

That said, baby boomers will still begin toretire in 2008, the American population willstill continue to age, and low birth rates willstill reduce the ratio of workers to retirees. Likeit or not, Social Security’s problems remain asreal and as pressing since September 11 as theywere before that date. In fact, the latest reportfrom Social Security’s trustees, released inMarch 2002, shows the program’s long-termcash deficits increasing, from $21.7 trillion inthe 2001 report to $23.9 in the latest edition.

If anything, the war on terrorism reminds usthat the federal government has other importantfunctions to accomplish, and allowing SocialSecurity and other entitlement programs togrow out of control clearly threatens the gov-ernment’s ability to conduct those duties.

Some have treated Social Security reform asa luxury to be undertaken when the governmenthappens to be flush with cash. In fact, it is anecessity that becomes even more importantwhen the government finds itself squeezedbetween its various duties. Those who intro-duced Social Security reform plans before theera of budget surpluses understood this, and therest of the reform community should as well.The correct question isn’t whether we canafford to reform Social Security, but whether

we can afford not to. Nevertheless, all of the commission’s pro-

posals demand an up-front investment. That’swhat pre-funding is: putting aside extra moneynow to save money later. When surpluses havedisappeared and their reappearance is question-able, how can we muster the courage—not tomention the cash—to move reform forward?

The common objection to the commission’splans is that they don’t say where they’ll get themoney to fund the transition. The quick answer tothat is that reformers will get it from the sameplace reform opponents will get the money tokeep the current system afloat, except they’ll needa lot less of it—up to 68 percent less, to be precise.

The more thoughtful answer is that toughdecisions indeed need to be made, but if wemake them now in the context of reforms usingpersonal accounts they’ll be a lot less tough andbe accomplished a lot sooner than if we don’t.Yes, there will be a transition period of movingto personal accounts, but after that we’re off thehook. If we don’t move to personal accounts andsimply let the program stagnate, we’re on thehook forever. It is reform opponents who need tocome up with the most cash, and most of themrefuse to put forward any proposals for doing so.

At this point, the Social Security debate islike a sporting event at which only one team hasshown up. The president has confronted one ofthe most contentious issues in politics andtaken the heat for doing so. Congressionalreformers such as Reps. Jim Kolbe (R-Ariz.)and Charlie Stenholm (D-Tex.), Sens. JuddGregg (R-N.H.) and John Breaux (D-La.), andReps. Richard Armey (R-Tex.), Jim DeMint(R-S.C.), and Clay Shaw (R-Fla.) have intro-duced their own reform plans and continue tofight for change. Each proposal has its pros andcons, but at least these reform plans are in thepublic arena where the costs and benefits canbe discussed and assessed.

But where is Senator Daschle’s reform plan?What does Dick Gephardt think we should do?Rep. Robert Matsui in press releases denouncesthe commission’s proposals, “I could have donethis by myself in two hours,”97 yet he apparentlyhasn’t found any spare time in a 24-year con-gressional career to actually sponsor a legislativeproposal other than one that would have the trustfund invest in municipal bonds that pay a lowerreturn than the fund’s current bonds.

Account opponents sometimes refer to the

All three planswould produce

those benefits atlower generalrevenue costs

than the currentprogram.

38

It is reformopponents whoneed to come upwith the mostcash, and mostof them refuseto put forwardany proposalsfor doing so.

39

reformers’ “secret plan,” but the biggest secretis what they would do to address the solvencyand sustainability of a program that constitutesthe biggest tax most workers pay as well as thebiggest source of income to most retirees.

This fact was made embarrassingly clear intestimony before the commission by representa-tives of the Campaign for America’s Future, acoalition of personal account opponents whowere active in opposing the commission. Whilecalling for “bipartisan dialogue,” the CAF tookevery opportunity to poison the well of publicdebate. Before the commission even met, theCAF issued “biographies” of the commissionersthat, for instance, described commission co-chairman Richard Parsons as having a “proventrack record as a corporate executive willing toundermine the retirement security of his employ-ees,” Robert Johnson of Black EntertainmentTelevision as “refusing to pay fair wages” toentertainers, and highlighted any and all linkscommission members may have to the invest-ment sector. Throughout the process, the CAFissued press releases and published op-edscharging the commission with “ignoring keycritics of President Bush’s privatization propos-al.” (As it happens, key critics of personalaccounts such as Rep. Robert Matsui were invit-ed to meet with the commission, in public or pri-vate, but declined to do so.)

Given these charges, one would assume thatwhen two leading members of the Campaignfor America’s Future were invited to testifybefore the commission, they would have comeprepared with constructive alternatives to per-sonal accounts. Indeed, at the commission’spublic hearing in San Diego it at first seemedthat specific options would be debated, whenCAF founder Roger Hickey stated:

It is well known that there are proposals tostrengthen Social Security without privatiz-ing and without across-the-board benefitcuts or tax hikes. They have been put for-ward by many experts: Henry Aaron, for-mer commissioner Robert Ball, economistPeter Diamond at MIT, Dean Baker andothers. We should be debating the details ofthese pragmatic plans, we believe.98

What appeared less well known to Hickeywas that the plans he cited are far from lacking“across-the-board benefit cuts or tax hikes”:

• Henry Aaron’s proposal with RobertReischauer increases both the early and thenormal retirement ages as well as increasingthe benefit computation period, both of whichconstitute across-the-board cuts in promisedbenefits (and are termed as such by Aaronand Reischauer); the Aaron-Reischauer planalso includes increases in the maximum tax-able wage, government investment in thestock market, and other changes.

• Robert Ball’s proposal with the 1994–96Advisory Council on Social Security wouldincrease payroll tax rates in the future, reduceannual cost-of-living adjustments, andincrease the benefit computation period,along with other changes such as governmentinvestment in the stock market.

• Dean Baker of the Center for Economicand Policy Research would increase pay-roll tax rates on all workers as well as thebase wage on which taxes are levied.

• Peter Diamond of MIT would force stateand local workers to enter the system; raisethe maximum wage subject to payrolltaxes; increase taxes on benefits and elimi-nate the current tax exemption for low-income retirees; index benefits for lifeexpectancy to about half the degree done inPlan 3; and phase in payroll tax increases.99

This proposal, however, has never beenfully analyzed for solvency.

Hickey eventually acknowledged to commis-sioner Fidel Vargas that his preferred course ofaction was to repeal the recent tax cuts passed byCongress and spend the proceeds on non-SocialSecurity programs, in hopes that this would spureconomic growth and benefit Social Security. Theprobability that additional federal spending willspur economic growth is for individuals to judge,but it is worth pointing out that economic growthwould need to double for Social Security toremain technically solvent throughout the 75-yearactuarial scoring period. 100

Hans Riemer, Roger Hickey’s colleague atthe Campaign for America’s Future, exhibitedbetter knowledge of existing reform proposalswhen he testified before the commission inWashington, D.C., but he demonstrated thesame lack of seriousness when it came to dis-cussing the alternatives to personal accounts.Riemer offered general prescriptions, such asrepealing the recent tax cuts, increasing or

removing the “cap” on income to which thepayroll tax applies, and investing the trust fundin the stock market (this last option is particu-larly puzzling given that Riemer simultaneous-ly endorses the view that future stock returnscannot exceed 3.5 percent annually, whichwould make trust fund investment practicallyworthless in terms of achieving solvency).101

When pressed for a more specific, comprehen-sive plan that could be scored by SocialSecurity’s actuaries and compared with theplans put forward by the commission, Riemeragreed to do so, and reiterated this agreement atlater dates.102 In the end, though, Riemer neverdelivered the promised proposal.

The commissioners did not claim that theirplans constitute a free lunch. Rather, they mere-ly maintained that their proposals met the pres-ident’s principles for reform and are superior tothe alternatives. Account opponents’ determi-nation to keep those alternatives a secretimplicitly acknowledges that the commission-ers are right. The sad state of the political dis-cussion over Social Security is revealed not inreformers’ disagreement with their opponents’policy proposals, but in the difficulty of dis-cerning precisely what policy proposals reformopponents actually favor. Realistically, how-ever, an unwillingness to embrace any reformplan is an acceptance of the status quo. This isfine, as long as everyone recognizes that thestatus quo allows Social Security to go broke.Until both sides lay their cards on the table anddeclare specifically what they wish to do, thereis little taking place in the way of debate.

Conclusion

The President’s Commission to StrengthenSocial Security took seriously its task to formu-late proposals incorporating voluntary personalaccounts that would not merely ensure the sol-vency of Social Security, but contribute to itslong-term sustainability as well. Sustainabilitymeans more than making Social Security’sassets equal its liabilities in a bookkeepingsense over a 75-year period. Reform mustensure that, in an overall economic and budget-ary sense, Social Security truly saves for thefuture, preparing the program and the countryto support a growing population of retireesthrough 2075 and beyond.

The commission’s three reform proposalsrepresent a range of ways to use private, indi-vidually controlled investment to strengthenSocial Security and to build assets and wealthfor Americans who need them most. None ofthe plans is perfect, and each constitutes a com-promise between commission members. Thefirst appendix includes the author’s own criteriafor reform, against which the commission pro-posals can be assessed.

Nevertheless, all three commission plansmove Social Security toward a sustainablefuture and contribute to the overall reformeffort. The commission’s Plan 2, in particular,would allow workers to regain control overtheir retirement savings while giving the feder-al government the budgetary flexibility thatcomes from a pension program that can livewithin its means.

In his 2002 State of the Union address,President Bush said that the fight against terror-ism wasn’t just our responsibility, but our priv-ilege. That sentiment applies just as well toSocial Security reform. It would be very easy tosit back and do nothing—to keep spending theprogram’s surpluses and pretending the prob-lem will fix itself, then feigning surprise whenit doesn’t. But to do so would sell us all short.Addressing Social Security’s problems today,making the tough choices now instead of pass-ing them off to others, is not just this genera-tion’s obligation, it is its privilege.

Appendix: AnalyticalFramework

What follows is a basic analytical frameworkof the primary policy issues that must beaddressed regarding Social Security reform.Answers to these four questions—whether tofund, where to fund from, how much to fund,and what to fund do not depend on any particu-lar philosophical or ideological beliefs.

Whether to Fund

The first question, whether to fund, askswhether it is desirable to move away fromSocial Security’s current pay-as-you-go financ-ing, in which each working generation pays thebenefits of the current retired generation, to afunded status, in which each working genera-

The commis-sioners did notclaim that theirplans constitute

a free lunch.Rather, they

merely main-tained that there

proposals metthe president’s

principles forreform and aresuperior to the

alternatives.

40

Reform mustensure that, inan overall eco-nomic andbudgetary sense,Social Securitytruly saves forthe future,preparing theprogram andthe country tosupport a grow-ing populationof retireesthrough 2075and beyond.

41

tion accumulates and holds assets to provideincome in its own retirement. Note that, at thisstage, no distinction need be made betweenfunding through a central investment strategyor through personal accounts.

The advantage of pay-as-you-go funding isthat it can begin paying benefits quickly: theSocial Security Act was passed in 1935 and theprogram began paying out retirement benefitsjust five years later. By contrast, a funded sys-tem demands a full working lifetime before itcan begin paying full retirement benefits.103

The advantage of a funded system is that atany given time it pays a substantially higherrate-of-return than does a pay-as-you-go pro-gram. This rate of return difference is not sim-ply the opinion of right wing economists; liber-al economists Paul Samuelson of MIT andHenry Aaron of the Brookings Institution estab-lished during the 1950s and 1960s that a pay-as-you-go system like Social Security will paya rate of return roughly equal to the growth rateof the taxable wage base—that is, labor forcegrowth plus wage growth.104 From 1960 to2000, this pay-as-you-go rate of return equaled2.9 percent after inflation.

By contrast, the return to capital during thatsame period was 8.5 percent after inflation.105

The difference was not based on risk—both“returns” fluctuated over time—but on the eco-nomic fundamentals involved.106 PeterDiamond of MIT, a prominent opponent of per-sonal accounts, found that under typical cir-cumstances the return from a funded systemshould exceed that of a pay-as-you-go system,and historically this difference has been wide.107

The upshot is that a funded system, howeverstructured, can pay the same benefits at substan-tially lower cost than a pay-as-you-go system.This advantage goes for a funded defined-bene-fit government-run system just as much as for adecentralized personal account program. Over a45-year working lifetime, a funded system at thehistorical return to capital pays benefits at one-sixth the cost of a pay-as-you-go plan. Even at amore modest return of 5 percent, which is closerto what could be expected from an actual invest-ment portfolio, the long-term cost is just half thatof a pay-as-you-go system.108

The conclusion then, based on the work ofeconomists opposed to personal accounts, isthat over the long term a pay-as-you-go systemis simply less efficient than a funded program.

Yet, while vastly inefficient over the longterm, pay-as-you-go programs like SocialSecurity are highly efficient in the short term:not in the sense that the program’s administra-tive costs are low—though for the retirementand survivors’ programs, they are—but in thatunder a pure pay-as-you-go system all of themoney paid in today is used today to pay bene-fits.109 There is no “waste, fraud, and abuse” tocut, no way to use the money actually paid toretirees more efficiently. In short, to move froman unfunded system to a funded system youhave to come up with additional funds. Thisraises the obvious question, “From where?”

Where to Fund From

To move from an unfunded pension systemto a funded system, several sources of fundingare available. The first source, and the leastlikely on a large scale, is current retirees. If wewere to reduce benefits to today’s retirees,workers could simply shift their payroll taxesfrom supporting today’s generation to savingfor their own. But this would constitute achanging of the rules after the game has beenplayed, which most people would considerunethical.110 The fact that the current systemdoes not and cannot guarantee benefits does notmean that policymakers should fund reforms atretirees’ expense. Moreover, many retirees havefew resources other than Social Security, solarge-scale reductions in payments to currentretirees would throw many into poverty. This ishardly the goal of reform.

Some reform plans have adopted tax increas-es as a funding source. In one sense, it seems likean obvious solution: why not simply obtain themoney from the same place government gets allits other money? Although revenue increases arefeasible on a small scale, if not desirable from aphilosophical viewpoint, any tax increase is like-ly to fall on higher wage workers who alreadysave large portions of their incomes. Becausethese workers may reduce personal saving inresponse to such a tax increase, this route couldgive the appearance of increasing saving withoutactually accomplishing it.

Similar objections apply to debt financingthe transition to reform. It is common sense thatyou can’t increase saving through borrowing.The economic benefits of a funded system flowfrom increases in saving, and increased saving

means reduced consumption, at least in theshort term.111 Under a debt-financed transitionto “funding,” either based on personal accountsor centralized government investment, theincreased benefits from the funded systemwould be offset by higher debt service costs,and the higher return from the funded systemwould be offset by the rate of return on the out-standing debt. Debt financing, again, gives theappearance of economic pre-funding withoutthe substance.112

The fourth possible source of funding isreductions in other government spending. It issaid that a government program is the closestthing to immortality on this earth. Programs cancontinue year after year because they benefitthe interests of those who sponsor them, even iftheir net benefit to society and the economy aresmall or even negative.

Recalling the real return to capital citedbefore—8.5 percent after inflation—it is diffi-cult to imagine many government programsyielding this return at the margin, or even wellin from the margin. There are exceptions, ofcourse; as certain government functions such asnational defense or police forces are prerequi-sites to most nongovernment economic activi-ties, basic government services presumablyproduce average returns above those availablein the market.

In general, however, existing estimates sug-gest below-market returns from most publicinvestment. Paul Evans and Gregorios Karrasexamined state government spending on educa-tion, highways, health and hospitals, police andfire protection, and sewers and sanitation, meas-uring whether increased government investmentin those functions raised state economic output.In general, they found the opposite: “We findfairly strong evidence that current governmenteducational services are productive but no evi-dence that the other government activities con-sidered are productive. Indeed, we typically findstatistically negative productivity for govern-ment capital.”113 Because the federal governmentspends relatively little on education, we mayinfer that the losses from reductions in federalinvestment spending would be outweighed bygains to Social Security were those funds invest-ed on its behalf.

From this perspective, the preferred route toa funded pension system is through reductionsin existing or projected federal government

spending rather than through increased taxationor public debt. That said, though, how muchfunding is desirable? And thus, how big shouldthose reductions in spending be?

How Much to Fund

How much funding to seek is as much avalue judgment as an economic question. Froman efficiency standpoint, it stands to reason thatover the long term a public pension programshould be fully converted from an unfunded toa funded status. Although some argue that amix of pay-as-you-go and funded financingdiversifies risks, it is my view that—all otherthings being equal—the greater efficiency offunded pensions outweighs the benefits ofdiversification with pay-as-you-go financing.

That said, however, preferring funding overpay-as-you-go is only half the question. Theother half is, even in the absence of pay-as-you-go financing, how much funding do we wish todo? Do we wish to devote the entire 12.4 percentcurrent payroll tax rate to a future funded pro-gram, or should we settle with a smaller amount?

At the rate of return to capital, a funded pro-gram can produce the same benefits as the cur-rent system at a payroll tax rate of less than 3percent. Assuming a 5 percent return, therequired payroll tax rate would be approxi-mately 6 percent, though a cushion wouldobviously be necessary to account for fluctua-tions in the market. If we were devote the entire12.4 percent payroll tax rate to a funded system,the level of retirement, survivors, and disabilitybenefits would be substantially higher thanthose produced by the current program. Manyworkers, particularly those with shorter lifeexpectancies, would prefer greater consump-tion opportunities during their working life-times to the higher level of retirement incomesuch a program would provide.

At the same time, more funding—that ismore saving today—requires more forgoneconsumption today. Even assuming the transi-tion is financed from current governmentexpenditures, the public has substantial senti-mental and personal interest attachments tomany of these programs. Moreover, the furtherone moves from government expenditures atthe margin toward the core functions of gov-ernment, the higher the presumed return onthose programs and the greater the cost to pres-

42

43

ent generations of giving them up. If the goal of a personal account is merely to

fill the gap between Social Security’s promisedand payable benefits, then an account can besmaller still. However, while a relatively smallaccount may fill this gap, its long-term benefitsare also proportionately smaller.

What to Fund

It is possible to go almost all the way throughthe analytical framework with no significantmention of personal retirement accounts. Thereason is that the economic case for a fundedSocial Security system has little to do withaccounts per se, and the public debate over thecosts and benefits of funding can take placeoutside of the personal account context. Evenamong account opponents there is a clear pref-erence for funding over pay-as-you-go financ-ing. For instance, the Clinton administration’sSocial Security proposals, first to invest thetrust fund in the stock market and later to pre-fund through debt reduction, both acknowledgethe case for a funded pension system.

On a permanent basis, however, there areonly two viable ways to fund Social Security.Debt reduction is not one of them. Although ithas the same economic effects as other meansof funding, there is only a limited amount ofpublicly held debt to reduce. Any large-scalemovement toward funding would soon exhaustcurrent supplies of debt to retire.114 (No, it is notpermissible to run up new debt for the purposeof repaying it later).

Hence, the realistic choices for a funded sys-tem are between centralized investment of thetrust fund and decentralized investment throughpersonal accounts. The advantages of central-ized investing are reduced administrative costsand the spreading of investment risks awayfrom retirees and onto taxpayers. That may beof value, since even under current demograph-ic trends, the working population will alwaysbe larger than the retired population. Shiftinginvestment risk onto the working public mayencourage the retired or near-retired populationto lobby for more aggressive investment poli-cies than taxpayers would be wise to bear.115

The main disadvantage of centralized invest-ing is that it risks political influence over capi-tal markets. Commentators ranging from AlGore to Alan Greenspan have argued that the

dangers of political investing are simply toogreat to be risked. Gore, a former supporter ofgovernment investment, said, “The magnitudeof the government’s stock ownership would besuch that it would at least raise the question ofwhether or not we had begun to change the fun-damental nature of our economy. Upon reflec-tion, it seemed to me that those problems werequite serious.”116

Defenders of centralized trust fund investinginsist that no firewall will be left unconstructedto safeguard the nation’s stock and bond mar-kets from politically influenced investing oftrust fund reserves. That may be so at the begin-ning, but others may soon find it in their inter-ests to leverage the equity power of trust fundinvesting to accomplish goals they see asworthwhile. Few firewalls cannot be breachedif those assigned to preserve them are uncom-mitted to the task.

Indeed, overseas experience confirms theserisks. A World Bank study of centralized invest-ment of government pension reserves foundthat in most cases investment returns did notexceed those available from an ordinary banksavings account. The principal reason, the Bankfound, is that investment decisions “are largelydetermined by the mandates and restrictionsimposed on public pension fund managers.Asset allocation decisions are largely politicaland have little to do with any application ofportfolio theory. In short, the problem is thatinvestment policy is driven by politicalmotives.” Both Ireland and Canada beganinvesting their Social Security funds passivelybut will soon begin active targeted investmentsin infrastructure and domestic industries, rais-ing concerns that political concerns will trumpthe financial needs of pensioners.117

Personal accounts would largely bypass therisks of political influence, as workers wouldhave the incentive to monitor the investmentchoices available to them and protest anymanipulation of investment choices towardnonfinancial goals. Moreover, personalaccounts, unlike central investing, give workersa true property right to their retirement savings.Individuals desiring low-risk investment choic-es could opt for government bonds, makingtheir benefits truly backed by the full faith andcredit of the United States. Younger workerscould invest in equities or corporate bonds,according to their needs and their willingness to

live with risk. Personal accounts could alsobenefit those who, having shorter life expectan-cies, do not fare as well in the current system inwhich benefits are effectively annuitized.Finally, as the commission pointed out, person-al accounts carry significant nonfinancial bene-fits, entirely separate from the benefits theydeliver. In testimony before the commission,Washington University professor MichaelSherraden summarized research on asset-hold-ing that has been conducted using experimentalIndividual Development Accounts. Among thefindings—

• Asset-holding has substantial positiveeffects on long-term health and marital sta-bility, even when studies control forincome, race, and education.118

• Among participants in trial programs ofIndividual Development Accounts, 84 per-cent feel more economically secure, 59percent report being more likely to makeeducational plans, and 57 percent reportbeing more likely to plan for retirementbecause they are involved in an asset-building program.119

• Individuals with investment assets, as well astheir children, perform better on educationaltests and reach higher educational attainment,even after accounting for income.120

• Single mothers and their children are lesslikely to live in poverty if the mother came

from a family with asset holdings, evenafter controlling for education and socio-economic status.121

• Saving patterns are passed on from parentsto children; parents who save are morelikely to have children who save, even afterother factors are counted. Hence, assetholding could be a means to establish long-term patterns of greater saving.122

• Ninety-three percent of individuals withIndividual Development Accounts say theyfeel more confident about the future and 85percent more in control of their livesbecause they are saving. Approximatelyhalf of account-holders report that havingaccounts makes them more likely to havegood relationships with family members,and 60 percent say that they are more like-ly to make educational plans for their chil-dren because they are greater saving.123

In this context, it is worth noting that thecommission’s Plans 2 and 3 deliberately estab-lished progressive personal accounts mostly tobuild savings and wealth most among thosewho currently have the least.

Opinions may differ, and policymakers andthe public must make up their own minds in thecourse of the political debate, but the criteriaoutlined above establish a strong prima faciecase for funded personal accounts as part of alarger Social Security reform package.

44

45

Appendix: Specifications of Commission Reform Models

Model 1 Model 2 Model 3

Personal Accounts

Personal account size 2% of wages 4% of wages up to $1,000 1% add-on contribution plus 2.5%annually (indexed annually of wages up to $1000 annuallyto wage growth) (indexed annually to wage growth)

Voluntary Yes Yes Yes

Additional contributions required? None; however, Plan 1 is None 1% of wages required to partici-a generic plan that can pate (subsidized throughaccommodate new income tax systemcontributions

Real return that makes individual 3.5% 2.0% 2.5% better off with accounts than without(SS defined benefit offset rate).

Accounts owned by participants? Yes Yes YesAccounts can be bequeathed to heirs? Yes Yes Yes

Participants can choose from a mix Yes Yes Yesof low-cost, diversified portfolios?

Contributions and account earnings Yes Yes Yessplitting in case of divorce?

Traditional Social Security Benefits

New minimum benefit None By 2018, a 30-year minimum- By 2018, a 30-year minimum-wagewage worker is guaranteed worker is guaranteed benefit equalbenefit equal to 120% of to 100% of poverty level (111%poverty level, indexed a 40-year worker), indexedannually to inflation. annually to wage growth.

Widow/widower benefits No changes Increased to 75% of couple Increased to 75% of couple bene-benefits (vs. 50% to 67% fits (vs. 50% to 67% today) fortoday) for lower wage couples lower wage couples

Changes to growth rate of traditional None specified Indexed to inflation instead of Indexed to gains in average lifewages starting for those turning expectancy (results in average an-62 in 2009 nual growth of 0.5% over inflation)

Additional changes to traditional None specified None specified • Reduce benefit for earlyretirement and increase benefit for late retirement.

• Gradually decrease the upperbend point replacement rate from15% to 10% starting in 2009.

Source: Excerpted from President’s Commission to Strengthen Social Security, final report.

Appendix: System Financing

Summary Results: Fiscal Sustainability Assuming2/3 Participation in Personal Accounts (PA) Model 11 Model 2 Model 3 Current Law

1. Expected personal account assets in 2075 ($PV2 trillions) $1.1 $1.3 $1.6 NA2. Gain in total Social Security system assets at end of 2075

(Increase in Trust Fund + Expected PA Assets; $PV trillions) $0.5 $4.8 $5.0 NA3. Reduction in cash flow requirements from general

revenue relative to present law 3,4 $1.7 $14.8 $11.3 $0.0Reduction in 75-year total (Sum of annual amounts in $2001 trillions) 7.7% 68.1% 52.2% 0.0%

Percent reduction versus current law (in $2001)Reduction in 75-year total ($PVbillion/$trillions) -$0.2 $2.3 $1.7 $0.0Percent reduction versus current law (in PV) -3.8% 45.0% 33.9% 0.0%

4. Social Security cash flowWith dedicated general revenue (GR) transfers

Cash flow positive by end of valuation window? No Yes Yes5 NoIncome Rate (including GR Transfer)-Cost Rate in 2075 (% of payroll) -4.56 1.41 0.125 -6.05Without dedicated GR transfers 4

Cash flow positive by end of valuation window? No Yes No NoIncome Rate (excluding GR Transfer) Cost Rate in 2075 (% of payroll) -4.56 1.41 -0.75 -6.05

5. Improvement in Actuarial Balance over 75-year periodImprovement with GR transfer (% of payroll) -0.32 1.99 1.88 0Percent improvement with GR transfer -17% 107% 101%Improvement without GR transfer (% of payroll) 4 -0.32 1.15 0.876 0Percent improvement without GR transfer -17% 62% 47%

6. Transition investmentAssuming current law surplus not used for financing 7

$PV trillions $1.1 $0.9 $0.4As % of GDP over years included in calculation 0.36% 0.49% 0.25%Includes current law surplus available for financing 8

$PV trillions $0.7 $0.4 $0.1 NAAs % of GDP over years included in calculation 0.29% 0.33% 0.10% NA

46

Source: President’s Commission to Strengthen Social Security, final report.1. Model 1 does not include additional transfers to maintain actuarial balance.2. PV = present value.3. Cash flow requirements include only general revenue required in any year to maintain solvency.4. Taxes on benefits and on PRA distributions are treated as Social Security revenues, not general revenue.5. Includes new dedicated sources of revenue.6. Improvement in actuarial balance would be +1.50 percent of payroll if new dedicated sources of revenue are included.7. Unified budget concept: Difference between income and cost of proposed model versus present law.8. Reflects extent to which negative balance in any year is more negative than under current law.

47

Notes1. Executive Order 13210, President’s Commission toStrengthen Social Security, May 2, 2001.

2. Estelle James, Averting the Old Age Crisis (Oxford:Oxford University Press, 1994).

3. President’s Commission to Strengthen SocialSec urity, interim report, August 2001. Hereinaftercited as interim report.

4. Henry J. Aaron, Alan S. Blinder, Alicia H. Munnell,and Peter R. Orszag, “Perspectives on the Draft InterimReport of the President’s Commission to StrengthenSocial Security,” The Century Foundation Center onBudget and Policy Priorities, July 23, 2001, p. 3.

5. Interim report, p. 9.

6. Ibid., p. 32.

7. It is not wholly true to say that the worker-retireeratio is fixed. Steps such as the scheduled increasein the normal retirement age should alter the ratioslightly in the program’s favor, though such stepsare also comparable to reductions in benefit pay-ments at any given age.

8. Neil Howe and Richard Jackson, “What Happensto Benefits When Social Security Goes Bankrupt?”The Concord Coalition, April 19, 2000.

9. In practice, rather than reducing each checksent to beneficiaries, checks would be withhelduntil sufficient funds existed to pay “full” benefits;over the course of a year, however, total benefitsreceived would be the same as if each monthlybenefit had been reduced.

10. Henry Aaron, Alan Blinder, Alicia Munnell, andPeter Orszag, “Perspectives on the Draft InterimReport of the President’s Commission toStrengthen Social Security,” The CenturyFoundation and the Center on Budget and PolicyPriorities, July 23, 2001, p. 3.

11. Robert T. Matsui, U.S. House of Representatives,Remarks from the Capital Hilton Hotel,Washington, July 24, 2001.

12. P. Mitchell Prothero, “Social Security Panel DrawsFire,” United Press International, July 24, 2001.

13. See Dean Baker, “Defaulting on the Social SecurityTrust Fund Bonds: Winners and Losers,” Center forEconomic and Policy Research, July 23, 2001.

14. David M. Walker, Comptroller General of theUnited States, “Social Security and Surpluses:GAO’s Perspective on the President’s Proposals,”February 23, 1999.

15. Henry Aaron, Alan Blinder, Alicia Munnell, andPeter Orszag, “Perspectives on the Draft InterimReport of the President’s Commission to StrengthenSocial Security,” The Century Foundation Center onBudget and Policy Priorities, www.tcf.org, July 23,2001, p. 3.

16. Alicia H. Munnell and R. Kent Weaver, “SocialSecurity’s False Alarm,” Christian Science Monitor,July 19, 2001.

17. Henry Aaron, Barry Bosworth, and Gary Burtless,Can America Afford to Grow Old? (Washington: TheBrookings Institution, 1988), p. 14. Emphasis added.

18. Aaron, Blinder, Munnell, and Orszag, p. 14.

19. Interim report, p. 17.

20. Eric Black, “Memo to My Editor; Re: SocialSecurity, Politics and the Balanced BudgetAmendment,” Minneapolis Star Tribune, March 10,1995, p. 14A.

21. Byron Dorgan and Kent Conrad, “UnfairLooting,” Washington Post, March 16, 1995, p. A21.

22. General Accounting Office, Social Security: TheTrust Fund Reserve Accumulation, the Economy, and theFederal Budget, Washington, January 1989, p. 6.Emphasis added.

23. Statement of Charles A. Bowsher, ComptrollerGeneral of the United States, before the SenateCommittee on Finance, “The Question of RollingBack the Payroll Tax: Unmasking the DeficitIllusion,” February 5, 1990.

24. Ibid., pp. 6, 8.

25. “Warning Issued by GAO about SocialSecurity,” St. Louis Post-Dispatch, January 29, 1989,p. 1A.

26. Alicia Munnell, “At Issue: Social Security andthe Budget,” Fiscal Policy Forum 4, no. 1 (Winter1986). Emphasis added.

27. However, regression analysis of Social Securitysurpluses and on-budget expenditures from 1979to 2001 indicate that improvements in SocialSecurity cash balances are more than offset byreductions in non-Social Security surpluses. SeeKent Smetters, “Has Mental Accounting BeenEffective in ‘Lock-Boxing’ Social Security’s Assets?Theory and Evidence,” University of Pennsylvaniaand National Bureau of Economic Research,August 2001.

28. Bob Graham and Robert Matsui, “SocialSecurity Safeguard,” Washington Times, September12, 1990, p. G3.

29. Robert Matsui, Testimony submitted to thePresident’s Commission to Strengthen SocialSecurity, August 15, 2001.

30. Robert Matsui, press release, “Trustees’ ReportShows Social Security Health Improving,” March26, 2002.

31. Stephen C. Goss, chief actuary, and Alice H.Wade, deputy chief actuary, “Estimates of FinancialEffects for Three Models Developed by thePresident’s Commission to Strengthen SocialSecurity,” Memorandum dated January 31, 2002,pp. 23–24.

32. Office of Rep. Robert Matsui, constituentnewsletter, Summer 2000.

33. Henry Aaron, “Costs of an Aging Population:Real and Imagined Burdens of an Aging America,”in ed. Henry Aaron, Social Security and the Budget(Lanham, Maryland: University Press of America,1988), p. 57. Emphasis added.

34. Henry Aaron, Barry Bosworth, and GaryBurtless, Can America Afford to Grow Old?(Washington: The Brookings Institution, 1988),pp. 7, 11. Emphasis added.

35. Alicia Munnell and C. Nicole Ernsberger,“Public Pension Surpluses and National Saving:Foreign Experience,” New England Economic Review,March/April 1989.

36. Alicia Munnell and Lynn Blais, “Do We Want LargeSocial Security Surpluses?” New England EconomicReview, September/October 1984. Emphasis added.

37. Pat Wechsler, “Will Social Security Be There forYou?” Newsday, January 14, 1990, p. 79.

38. James Buchanan, “The Budgetary Politics of SocialSecurity,” in ed. Carolyn L. Weaver, Social Security’sLooming Surpluses: Prospects and Implications, (Lanham,Maryland: University Press of America, 1990).

39. Alan Blinder, “The Budgetary Politics of SocialSecurity,” in Weaver, 1990. Disagreeing withBuchanan, Blinder argued that Trust Fund decu-mulations post-2016 would have an equal andopposite effect on government spending, reducingit to less than it would otherwise be. The commis-sion’s interim report notes this possibility andoutlines the degree of spending restraint thatwould be required under such a scenario.

40. Peter Orszag, testimony to the Social SecuritySubcommittee of the House Ways and MeansCommittee, “Global Aging and Social SecurityCrises Abroad,” September 21, 2000. See also,Peter R. Orszag and Joseph E. Stiglitz, “RethinkingPension Reform: Ten Myths about Social Security

Systems,” presented at the World BankConference, New Ideas about Old Age Security,September 14–15, 1999.

41. Ibid. Emphasis added.

42. Steve Gerstel, “Senate Begins Debating $1.1Trillion Fiscal 1990 Budget,” United PressInternational, May 2, 1989.

44. Leo Fitzmaurice, “Republicans MaskingDeficit, Gephardt Says,” St. Louis Post-Dispatch,February 13, 1990, p. 10A.

44. But they must first devise a way for SocialSecurity to truly save these funds rather thanmerely crediting them to the trust fund and trans-ferring them to general revenue to be spent.

45. For instance, an August 2, 1999, ZogbyInternational poll asked, “If Social Security fundsare invested in stocks and bonds, who should dothe investing—the government through a centralfund, or individual workers through privateaccounts like an IRA or 401(k)?” Respondentsfavored private accounts by a more than four-to-one margin. A January 25–27, 2002, CNN/USA/Gallup Today poll showed that, by a margin of 63to 33 percent, respondents favor the option toinvest part of their payroll taxes in a personalretirement account, despite a lengthy period oflow market returns.

46. Peter Coy, “Who Loses under Social SecurityReform?” Business Week Online, December 19, 2001.

47. Alan L. Gustman and Thomas L. Steinmeier,“How Effective Is Redistribution under the SocialSecurity Benefit Formula?” Journal of PublicEconomics 82, no. 1 (October 2001): 1–28.

48. See Julia Lynn Coronado, Don Fullerton, andThomas Glass, “The Progressivity of Social Security,”National Bureau of Economic Research, March 2000.

49. See Gregory Pappas, Susan Queen, WilburHadden, and Gail Fisher, “The IncreasingDisparity in Mortality between SocioeconomicGroups in the United States, 1960 and 1986,” NewEngland Journal of Medicine 329 (July 8, 1993):103–109.

50. U. S. Bureau of the Census, Current PopulationReports, Series P20-514; Marital Status and LivingArrangements: March 1998 (Update); and earlierreports.

51. This could happen in two ways: by working parttime or by withdrawing entirely from the workforcefor a time. In either case, Social Security’s benefitformula has difficulty distinguishing betweenthose who worked full time at low wages and those

48

49

who worked part time at high wages. For instance,Ms. Smith is a single mother who works full time athalf the average wage ($15,000) while simultane-ously raising her children. Ms. Jones is married andworked for nine years at twice the average wage($60,000) before leaving work to raise her children.At retirement, these two very different womenwould receive similar benefits—about $620 permonth. Even though Ms. Smith worked full time atlow wages and Ms. Jones part time at high wages,their average wage—upon which their benefits arebased—is the same. And because that average is low,both benefit from Social Security’s “progressivity”even if only one actually needs it.

52. Alan L. Gustman and Thomas L. Steinmeier,“How Effective Is Redistribution under the SocialSecurity Benefit Formula?” National Bureau ofEconomic Research Working Paper no. 7597,March 2000, p. 33.

53. Peter Coy, “Who Loses under Social SecurityReform?” Business Week Online, December 19, 2001.

54. As defined by the Social Security Administra-tion, a low-wage worker receives 45 percent of theaverage wage, while a high-wage worker receives 160percent of the average wage.

55. Interim report, p. 26.

56 Orlo R. Nichols, Michael D. Clingman, and MiltonP. Glanz, “Internal Real Rates of Return under theOASDI Program for Hypothetical Workers,” SocialSecurity Administration, Office of the Chief Actuary,Actuarial Note no. 144, June 2001.

57. Robert L. Clark, Gordon P. Goodfellow,Sylvester J. Schieber, and Drew A. Warwick,“Making the Most of 401(k) Plans: Who’s ChoosingWhat and Why,” in ed. Olivia S. Mitchell, P. BrettHammond, and Anna M. Rappaport, ForecastingRetirement Needs and Retirement Wealth (Philadelphia:University of Pennsylvania Press, 2000), pp. 95–138.

58. The commission recommended, “personalaccount distributions should be permitted to betaken as an annuity or as gradual withdrawals,and balances above a threshold can also be takenas a lump-sum distribution. The thresholdamount should be chosen so that the yearlyincome received from an individual’s defined ben-efit plus the joint (if married) annuity keeps bothspouses safely above the poverty line during retire-ment, taking into account expected lifetimes andinflation.” Final report, pp. 41–42. Some accountcritics charge that requirements to annuitize elim-inate the possibility of wealth building andbequests, popular features of personal accounts.(See, for example, Bernard Wasow, “Setting theRecord Straight: Two False Claims about AfricanAmericans and Social Security,” The Century

Foundation, March 2002, p. 2. The increasedpoverty protections incorporated into the tradi-tional program as part of Plans 2 and 3, however,mean that relatively small fractions of the accountbalance would be required to be annuitized tomaintain an income above the poverty line, thusenabling retirees to leave lump sums to their heirs.Moreover, account balances of workers who diebefore retirement could also be passed on, whichis of considerable value to groups with shorter lifeexpectancies.

59. Whereas the current Social Security programoffers a de facto inflation-adjusted annuity, whichis often touted as superior to anything the marketcould provide, private inflation indexed annuitiesare now on the market. Moreover, there is evidencethat individuals would prefer variable annuitiesthat invest in equities to real, inflation-indexedannuities. See Jeffrey R. Brown, Olivia S. Mitchell,James M. Poterba, “The Role of Real Annuities andIndexed Bonds in an Individual AccountsRetirement Program,” in ed. Zvi Bodie, P. BrettHammond, and Olivia S. Mitchell, Innovations inFinancing Retirement (Philadelphia: University ofPennsylvania Press, 2002), pp. 175–97.

60. For instance, assume that a worker investedthe maximum of $1,000 annually into his or herpersonal account. Compounding these contribu-tions at 2 percent would total $60,644 after 40years, and computing this notional lump sumthrough an annuity formula would reduce tradi-tional benefits by approximately $360 per month.However, the true account balance, growing at anassumed interest rate of 4.6 percent, would equal$112,265—an amount sufficient to increasemonthly benefits by $672. Again, as long as theaccount’s interest rate exceeds the offset interestrate, total retirement benefits will increase.

61. Final report, p. 98

62. Ibid., p. 113.

63. See Henry J. Aaron, Alan S. Blinder, Alicia H.Munnell, and Peter R. Orszag, “Governor Bush’sIndividual Account Proposal: Implications forRetirement Benefits,” The Century FoundationIssue Brief no.11, June 2000.

64. This amount would be 88 percent of whatSocial Security promises but cannot pay. With anet return of around 4.2 percent, the 2042 retireecould receive more than his promised benefit.

65. Flemming v. Nestor, 363 U.S. 603 (1960).

66. If payroll tax shortfalls were financed throughincome tax revenues, rates would have to rise byapproximately 17 percent across the board by 2052. Ifincome tax increases were restricted to upper rates, the

increases would be even larger. See Andrew G. Biggs,“The Cost of Not Reforming Social Security,” TheDismal Scientist, April 17, 2001, www.Economy.com.

67. Author’s calculations.

68. For discussion of differential mortality accord-ing to income, see Eugene Steuerle and Jon M.Bakija, Retooling Social Security for the 21st Century:Right and Wrong Approaches to Reform, (Washington:Urban Institute Press, 1994), pp. 115–119.

69. Interim report, p. 13.

70. The latest report from Social Security’s trusteesillustrates this point, albeit on a smaller scale. Thetrustees increased their projections for future pro-ductivity growth, but this had little effect on systemsolvency and, overall, the program’s actuarial balancedeclined and its long-term deficits increased.

71. In practice, there are several means to accom-plish price indexing. Under the commission’s Plan2, price indexing would actually be implemented bymultiplying the PIA bend point factors (i.e., 90, 32,and 15 percent replacements) by the ratio of theConsumer Price Index to the Average Wage Index insuccessive years. The bend point dollar amountswould remain indexed to wages. By indexing thefactors rather than the dollar amount, the progres-sivity of the benefit formula is retained. If the bendpoint dollar amount were indexed to prices, overtime retirees would see practically all of their pre-retirement wages covered under the 15 percentbend point, creating a de facto flat replacement ratethroughout the income distribution.

72. In other words, a worker retiring just after2009 would have his working years before 2009calculated under the old system, with only yearspast 2009 calculated under the new formula.

73. Kilolo Kijakazi and Robert Greenstein,“Replacing ‘Wage Indexing’ with ‘Price Indexing’Would Result in Deep Reductions over Time inSocial Security Benefits,” Center on Budget andPolicy Priorities, December 14, 2001.

74. Replacement rates would also fall for workers withwage levels derived from the average in any given year.The Social Security Administration’s “low-wage” earn-er, for instance, is defined as earning 45 percent of theaverage wage, whatever that average wage may be.

75. Assuming an all-bond portfolio, the replace-ment rate would be approximately 29.4 percent,while with a 60-40 stock bond portfolio it wouldbe approximately 36 percent.

76. Greenstein and Kijazaki, p. 6.

77. This, of course, is untrue when benefits from

the personal accounts are counted in. All workerswould receive more than the current system canpay, and many low-wage workers would receivemore than is even promised.

78. Their position is valid only when workers’ eco-nomic well-being is judged not by their absoluteincomes but by their incomes relative to others. Inthis sense, the low-wage worker in 2070 will be dis-advantaged in a way that an average-wage workertoday is not, despite their having virtually identi-cal incomes.

79. Testimony of Hans Riemer before thePresident’s Commission to Strengthen SocialSecurity, August 22, 2001, www.csss.gov, p. 27.

80. Peter Diamond, James Hickman, William Hsiao,and Ernest Moorhead, “Report of the ConsultantPanel on Social Security to the CongressionalResearch Service,” August 1976, p. 23.

81. Ibid., p. 9.

82. Peter Diamond, James Hickman, WilliamHsiao, and Ernest Moorhead, letter to the NewYork Times, May 29, 1977, sec. 4, p. 14.

83. “Propping up Social Security,” Business Week,July 19, 1976, p. 34.

84. Statement of Henry Aaron, Gardner Ackely,Mary Falvey, John Porter, and J. W. Van Gorkom,Social Security Financing and Benefits, Report of the1979 Advisory Council (Washington: GovernmentPrinting Office, 1979), pp. 212–15.

85. “Will Voluntary Personal Accounts Save SocialSecurity?” American Enterprise Institute SeminarSeries in Tax Policy, April 5, 2002.

86. For instance, the wage indexing plan advocat-ed by the Ford administration implied systemcosts in 2030 of 18.9 percent of payroll, contrast-ed with 17.24 percent projected in the 2002 SocialSecurity trustees report. The Hsaio panel faultedthe Ford administration’s plan for leaving “a sig-nificant actuarial deficit in the financing of theOASDI system” and highlighted “the stability ofthe tax rates needed to finance promised benefitsunder this Panel’s recommendation—a stabilitynot enjoyed by other major recommendationsthat Congress is considering.” Report of theConsultant Panel on Social Security to the CongressionalResearch Service, p. 6.

87. See Michael Sherraden, “Saving in IDAPrograms: Supplement to Invited Testimony tothe President’s Commission on Social Security,”October 28, 2001, p. 1, which noted that even witha 2-to-1 match, participants contributed to theiraccounts an average of only 7 out of 12 months of

50

51

the year and saved only 67 percent of the maxi-mum eligible for the match.

88. Paul Yakoboski, Pamela Ostuw, and JenniferHicks, “What Is Your Savings Personality? The1998 Retirement Confidence Survey,” EmployeeBenefits Research Institute Issue Brief no. 200,August 1998, p. 11, www.ebri.org/rcs/T114.pdf.

89. Hans Riemer, “Bush Social Security CommissionProposes Benefit Cuts to Pay for Privatization,”Institute for America’s Future, December 6, 2001.

90. Testimony of Roger Hickey before thePresident’s Commission to Strengthen SocialSecurity, September 6, 2001, p. 226, www.csss.gov.Hickey cited the Aaron-Reischauer proposal asone of several plans he approved of that reachedactuarial balance “without privatizing and with-out across-the-board benefit cuts or tax hikes” (p.204). On being reminded that the Aaron-Reischauer plan increased the retirement ageHickey declared his opposition to that provision,but fully 38 percent of the plan’s progress towardsolvency is achieved through changes in the retire-ment age. Aaron-Reischauer also contains anotheracross-the-board benefit cut in the form of anincrease in the benefit computation period from35 to 38 years, which constitutes another 13 per-cent of the plan’s progress toward solvency. SeeHenry J. Aaron and Robert D. Reischauer,Countdown to Reform: The Great Social SecurityDebate, the Century Foundation Press, New York,2001.

91. Staff of Robert Matsui, U.S. House ofRepresentatives, “Six Problems with PrivatizingSocial Security,” March 6, 2002.

92. Interim report, p. 138; emphasis in original.

93. Social Security Advisory Board, “Charting theFuture of Social Security’s Disability Programs:The Need for Fundamental Change,” January2001, p. 11.

94. Ibid., p. 11.

95. Ibid., p. 5.

96.“Administrative Expenses as a Percentage ofContribution Income and of Total Expenditures,Fiscal Years 1996–2000,” Table III.A8 of the 2001OASDI Trustees Report.

97. Robert A. Rosenblatt, “Options Given to SaveSocial Security; Privatization: Panel Will LetPresident Bush Choose among Varied Plans toRestore Retirement Program to Solvency,” LosAngeles Times, November 30, 2001.

98. Roger Hickey, Testimony before the President’s

Commission to Strengthen Social Security, p. 204.

99. Peter Diamond, Remarks at “Will VoluntaryPersonal Accounts Save Social Security?” AmericanEnterprise Institute Seminar Series in Tax Policy,Friday, April 5, 2002.

100. On the effect of economic growth on system sol-vency see Andrew G. Biggs, “Social Security: Is It ‘ACrisis That Doesn’t Exist’?” the Cato Institute, SocialSecurity Privatization Paper no. 21, October 5, 2000.

101. Hans Riemer, “Young Social Security Beneficiariesin the 50 States,” 2030 Center, October 2, 2000, p. 5.

102. Hans Riemer, testimony before the President’sCommission to Strengthen Social Security,October 18, 2001, p. 44ff.

103. Individuals who pay into a funded system for onlypart of their working lifetimes would, of course, stillreceive proportionate benefits at retirement.

104. See Paul A. Samuelson, “An Exact Consump-tion-Loan Model of Interest with or without theContrivance of Money,” The Journal of Political Economy,December 1958, pp. 467–482; Henry J. Aaron, 1966.“The Social Insurance Paradox,” Canadian Journal ofEconomics and Political Science, vol. 32, no. 3. In practice,Social Security’s pay-as-you-go return is slightly lowerthan the sum of its two components because the num-ber of hours worked generally declines over time.

105. James Poterba, “The Rate of Return toCorporate Capital and Factor Shares: New EstimatesUsing Revised National Income Accounts andCapital Stock Data,” National Bureau of EconomicResearch, April 1999, pp. 9–10. The standard devia-tion of returns on capital was 1.0 percent.

106. If the return to capital were lower than the pay-as-you-go return (a so-called “dynamically ineffi-cient” economy), all generations would bene-fit byfunding present consumption on a pay-as-you-gobasis until such time as the capital stock were deplet-ed sufficiently to raise the return from capital abovethe pay-as-you-go rate. This was the case in Canadaduring the 1960s when it established its pay-as-you-go pension program, though even then such a cir-cumstance could not be expected to long continue.For the United States to reach dynamic inefficiencywould require a national saving rate of approximate-ly 30 percent, significantly above the present level.Countries with extremely high saving rates couldtheoretically benefit from pay-as-you-go financing,though only the highest saving countries such asSingapore could possibly qualify and globalized cap-ital markets make such benefits even less likely. See,for example, Kenneth Kassa, “Does Singapore InvestToo Much?” Economic Letter, Federal Reserve Bank ofSan Francisco, May 15, 1997. Of course, this analysisleaves aside other aspects of reform, such as risk and

ownership, which would also play into any decision.

107. See Peter A. Diamond, “National Debt in aNeoclassical Growth Model.” American EconomicReview no. 55, 1965, pp. 1126–1150.

108. The distinction between the real, pre-taxreturn to capital and the returns to investmentaccounts is most commonly made by MartinFeldstein; see “The Missing Piece in Policy Analysis:Social Security Reform,” The Richard T. ElyLecture, American Economic Review 86, no. 2 (May1996): pp. 1-14. Feldstein points out that while theindividual receives the return to his investment onlyafter state, local, and federal corporate taxes havebeen paid, the real return to new capital investmentis in fact the pre-tax return to capital. Some SocialSecurity reform plans attempt to “capture” that fullreturn by estimating the amount of new saving theplan would create, deriving the increased federalcorporate tax revenues based on that new saving,and crediting those new revenues to Social Security.

109. This is complicated somewhat in that SocialSecurity currently runs payroll tax surpluses.Although these surpluses are credited to the trustfund, many analysts (as detailed above) believethey are not contributing to national saving.

110. However, if post-1983 payroll tax surpluseswere consumed, then taxpayers during that periodenjoyed more government services or lower taxesthan they would have had those surpluses beensaved. Moreover, future workers will earn lowerwages than if the surpluses had been saved. In thiscase, future workers could argue with past workersthat they were not provided with the economicmeans to finance full promised benefits.

111. Over the long term, however, an economywith a higher rate of saving will in general alsohave a higher rate of consumption, up to the so-called “golden rule” level of saving.

112. There are, of course, other reasons for movingto a personal account–based pension system, suchas personal ownership, control over investments,and the ability to pass investments on to an heir.These reasons could make the transition worth-while even if debt financing were involved.

113. Paul Evans and Gregorios Karras, “Are govern-ment activities productive? Evidence from a panel ofU.S. States,” The Review of Economics and Statistics76, no. 1, February 1994, pp. 1-11. See also, DogulasHoltz-Eakin, “Public Sector Capital and theProductivity Puzzle,” Review of Economics andStatistics 76, no. 1, February 1994, pp. 12–21.

114. See General Accounting Office, Federal Debt:Debt Management Actions and Future Challenges,

GAO-01-317 (Washington: Government PrintingOffice, February 28, 2001).

115. On the risk-related costs of centralized invest-ment, see George M. Constantinides, John B.Donaldson, and Rajnish Mehra, “Junior Must Pay:Pricing the Implicit Put in Privatizing SocialSecurity,” National Bureau of Economic ResearchWorking Paper no. 8906, April 2002.

116. Richard W. Stevenson and James Dao, “GoreDefends Stock Investment Switch,” The New YorkTimes, May 25, 2000, p. 27.

117. Canadian Chamber Of Commerce, “Canada’sRetirement Income System: Still in Need ofReform,” February 2002; and Philip Lane,“Gambling with the Future of Our Pensioners,”The Irish Times, February 12, 2002.

118. R. J. Galligan and S. J. Bahr, “Economic Well-Being and Marital Stability: Implications forIncome Maintenance Programs,” Journal ofMarriage and the Family, 1978, pp. 283–290;Hampton, R. L., “Family Life Cycle, EconomicWell-Being and Marital Disruption in BlackFamilies,” California Sociologist 5, 1982, pp. 16–32;and S. J. South and G. Spitze “Determinants ofDivorce over the Marital Life Course,” AmericanSociological Review 51, no. 4 (1986): 583–590.

119. A. Moore, S. Beverly, M. Schreiner, M.Sherraden, M. Lombe, E. Cho, L. Johnson, and R.Vonderlack, Saving, IDA Programs, and Effects ofIDAs: A survey of Participants. Downpayments on theAmerican Dream Policy Demonstration: A NationalDemonstration of Individual Development Accounts,Washington University in St. Louis, GeorgeWarren Brown School of Social Work, Center forSocial Development, 2001.

120. S. Mayer, What Money Can’t Buy: Family Incomeand Children’s Life Chances (Cambridge,Massachusetts: Harvard University Press, 1997);M. S. Hill and G. J. Duncan, “Parental FamilyIncome and the Socioeconomic Attainment ofChildren,” Social Science Research 6 (1987): 39–73.

121. L. Cheng, “Asset Holding and Intergeneration-alPoverty Vulnerability in Female-Headed Families,”Paper presented at the Seventh InternationalConference of the Society for the Advancement ofSocio-Economics, Washington, April 7–9, 1995.

122. M. E. Pritchard, B. K. Meyers, and D. Cassidy,“Factors Associated with Adolescent Saving andSpending Patterns,” Adolescence 24, no. 95 (1989):711–723.

123. Moore et al., Saving, IDA Programs, and Effects ofIDAs.

52