Andrew G. Biggs is a Social Security analyst at the Cato Institute.

Executive Summary

Aconsensus has developed across the politi-cal spectrum that the Social Security pro-

gram faces significant problems and is in needof far-reaching modifications. Would-bereformers debate vigorously on the bestchanges for Social Security. Some argue fortransforming the nation’s pension program to adefined-contribution system of personal retire-ment accounts while others support retainingthe current defined-benefit structure through aseries of tax increases and benefits cuts orthrough investing a portion of the program’sassets in equities.

But some people in politics, the press, andthe policy community are questioning that con-sensus, calling Social Security’s projected fund-ing shortfalls merely the result of pessimisticeconomic and demographic projections by theprogram’s Board of Trustees. If the economygrows faster than projected, as they believe itsurely will, then wages and payroll tax revenueswill rise and Social Security will become, in thewords of Rep. Jerrold Nadler (D-N.Y.), “a crisisthat doesn’t exist.”

However, independent assessments of theTrustees’ projections for productivity, laborforce growth, and longevity show the projec-tions to be reasonable and perhaps even opti-

mistic. For Social Security to remain solvent,even in a bookkeeping sense, would demandunprecedented levels of economic growth.More important, even if the economy doesgrow more quickly, Social Security’s benefitliabilities and its funding shortfalls will eventu-ally rise along with the economy. Even underassumptions vastly more optimistic than thosethe crisis deniers put forward, Social Securitystill faces trillions of dollars in tax increases orbenefit cuts if the system is to stay in balance.

A possible corollary exists to the argumentmade by skeptics of the Social Security crisis. Ifthe economy grows as slowly as the trusteesproject, can market investments like stocks andbonds continue to produce returns superior tothose from Social Security? Although futurereturns from market investments cannot beguaranteed, the differences in returns betweenSocial Security and market investments are sogreat that even under a worst-case scenario per-sonal retirement accounts invested in stocks andbonds would produce far higher returns thanSocial Security.

In short, Social Security’s crisis is real andmay be even larger than commonly thought.While debate may continue over the propercourse of action, doing nothing in hopes that theeconomy will come to the rescue is wishfulthinking, at best.

October 5, 2000 SSP No. 21

Social Security Is It “A Crisis That Doesn’t Exist”?

by Andrew G. Biggs

1

Introduction

Supporters of the current pay-as-you-goSocial Security system have long been on thedefensive. According to the latest report ofSocial Security’s Board of Trustees, by 2015payroll tax revenue will be less than benefit lia-bilities, and by 2037 the nation’s public pensionsystem will be able to pay less than three-quar-ters of promised benefits, pushing millions oflow-income retirees into poverty. For the pastseveral years, both sides of the political spec-trum have explored the issue. President Clintonspent a year at town-hall style meetings stress-ing the need for reform. There has been dis-agreement on the proper mode of reform; somecall for buttressing the current system with gen-eral tax revenues, while others advocate trans-forming Social Security into a system of per-sonal retirement accounts invested in stocksand bonds. Nevertheless, until recently all sideshave agreed with the president that this momentof economic prosperity is the proper time toaddress Social Security reform, to “fix the roofwhile the sun is shining.”1

But some now deny that the roof even needsfixing and question whether the rain will evercome. To these “crisis deniers,” Social Security’sproblems are simply the product of pessimisticeconomic and demographic assumptions by theprogram’s trustees, with politicians and activistsof both left and right eager to exploit these errors.If the economy’s growth exceeds the trustees’assumptions, as they believe it surely will, SocialSecurity becomes, in the words of Rep. JerroldNadler (D-N.Y.), “a crisis that doesn’t exist.”2

Many members of the press have adopted thisargument. For instance, financial columnist JaneBryant Quinn said in 1998 that “We can’t dragour feet any longer on Social Security reform,”3

but now doubts whether the crisis will material-ize at all. Today, she calls herself “the only kid inthe village who’s not crying wolf” on SocialSecurity.4 Likewise, the editors of Business Weekcall the trustees’ economic projections “ridicu-lously low,” making the reform debate betweenprivatizers and those seeking marginal change a“phony conflict over a phony problem.”5

One reason for the crisis deniers’ line ofargument may be that public opinion on reformhas seemingly settled on plans based on per-sonal retirement accounts, which would giveworkers the option to invest part of their payroll

taxes in stocks or bonds. Public opinion pollsshow enthusiasm for personal accounts amongAmericans of all political, ethnic, and gendergroups,6 and bipartisan leaders in both housesof Congress are promoting personal accountplans on Capitol Hill. But to a few old-guardsupporters of big government, individualinvestment is ideological heresy. Yet, the moreconventional alternatives to personal accountsas a means of shoring up the system—payrolltax hikes, benefit cuts, increasing the retirementage, and even investing the Social Security trustfund in the stock market—are flatly rejected bythe public.7 The crisis deniers’ solution? Simplydeny the crisis exists at all. If the Social Securitycrisis no longer exists, then radical reform suchas personal retirement accounts are unnecessary.

But the crisis deniers’ claim that SocialSecurity reform is a solution in search of a prob-lem is not aimed simply at reformers who favorpersonal accounts, but at all who see seriouslong-term problems with Social Security andseek equally serious changes to address them. Toassess those claims, we must first determinewhether the trustees’ projections for the mostimportant economic and demographic variablesaffecting Social Security’s solvency are reason-able; and, second, we must determine whethermore optimistic projections would “save SocialSecurity.” A related matter to be examined iswhether stock market returns in a slowly grow-ing economy would pay a higher rate of return toworkers than the current program.

An examination of these issues shows that,while no one can predict the future with certain-ty, the trustees’ assessments of key economic anddemographic variables are generally reasonableand, in some cases, perhaps even optimistic. Andeven if economic growth greatly exceeds thetrustees’ projections, when workers pay moretaxes into Social Security they are entitled togreater benefits when they retire. Hence, muchof the benefit of economic growth is simplywashed away. Even under assumptions vastlymore optimistic than those the crisis deniers putforward—where economic growth increases,unemployment falls, life expectancies barelyincrease, and immigration brings millions ofnew workers into the system—Social Securitystill faces trillions of dollars in tax increases orbenefit cuts if the system is to stay in balance.

Baker and Weisbrot declare that, “As anyonewho has looked at the numbers knows, Social

2

3

Security is financially sound for as far into thefuture as we would ever want to worry about.”8

Closer analysis will show that anyone expect-ing to be in retirement any time past the year2015 has ample reason for worry. Policymakersand the public should not be distracted by argu-ments over whether Social Security’s crisisexists and by appeals to wait and see if prob-lems arise before taking action. Instead, theyshould focus now on how big the crisis is andhow it will be addressed.

The Argument

Social Security’s Board of Trustees, made upof government officials and outside appointees,produces annual reports on the financial condi-tion of the program. The latest Trustees Report,issued in March 2000, projects payroll taxinsolvency for the program in a little more than15 years and a 75-year payroll tax shortfall ofover $20 trillion (in 2000 dollars).9 UnlessSocial Security is reformed, either payroll taxrates will have to increase by up to 50 percentor the system’s already meager promised bene-fits will have to be cut by almost a third. Thecentral thesis of those who deny SocialSecurity’s crisis is that Social Security’strustees use unusually pessimistic assumptionsin projecting these grim scenarios. For instance,Dean Baker and Mark Weisbrot, authors ofSocial Security: The Phony Crisis,10 claim that“any shortfall that Social Security may have inthe future can result only from a dismal eco-nomic performance.”1 1 Former Labor secretaryRobert Reich agrees:

This crisis mongering is simply wrong.As a former trustee of the Social Securitytrust fund, I can tell you that the actuary’sprojections are based on the pessimisticassumption that the economy will growonly 1.8 percent annually over the nextthree decades. Crank the economy up justa bit, to a more realistic 2.2 percent a year,and the fund is nearly flush for the nextseventy-five years.1 2

It is unclear why, during his time as a trustee,Reich did not point out these seemingly obvi-ous failings to his colleagues, but it is not justpoliticians who take this line. Many in the pol-

icy community and the press echo these views.Subsequent to Reich’s statement, SocialSecurity’s trustees revised their economic pro-jections slightly upward. But to the EconomicPolicy Institute’s Christian Weller and EdieRasell, they didn’t go nearly far enough.

Even with these positive changes, though,the report continues to be based on pes-simistic assumptions about the futureeconomy. Recent developments suggesthigher real GDP and productivity growththan the trustees assume. Hence, realwage and payroll-tax revenue growthshould be greater than predicted by thetrustees’ report, increasing the size of thetrust fund. Given the report’s improvedforecast in spite of these pessimisticassumptions, there is even less need to cutbenefits or to privatize the system.1 3

The 2030 Center concurred, terming “the pro-jections for the long-term shortfall . . . very pes-simistic.”1 4

Press commentators have adopted this themeas well. Financial columnist Quinn declares,“We don’t even know, for sure, that the trustfund will dry up in 2037. That’s just a projec-tion. To be on the safe side, Social Security’strustees have assumed slower economic growththan we’ve averaged over the past 75 years. If itturns out that future growth equals that of thepast, the Social Security problem all but goesaway. . . .”15

Even Vice President Gore has been temptedby these ideas. While President Clinton warnsthat “a demographic crisis is looming” thatcould bankrupt the system,16 Vice PresidentGore has backed off the most far-reaching ofthe Clinton administration’s reform propos-als—investing the Social Security trust fund inthe stock market—declaring “If it ain’t broke,don’t fix it.”1 7

But when we get beyond hopeful generalitiesto examine how Social Security works andwhere the system currently stands, economicgrowth by itself is a false promise. Under rea-sonable economic assumptions, the currentSocial Security system is unsustainable over thelong term. And unsustainable conditions, aseconomist Herbert Stein famously noted, can-not go on forever. Sooner or later, deficits musteither be cleared through large tax hikes or ben-

efit cuts, or a funded alternative based on mar-ket investment and higher rates of return mustbe implemented.

Are the Trustees’ ProjectionsPessimistic?

Social Security’s Board of Trustees annuallyreports on the program’s current financing situ-ation and makes estimates of its financinghealth over the following 75-year period. Thetrustees, with the aid of the Social SecurityAdministration’s actuaries and in consultationwith outside experts, constructs three scenarios:low cost, high cost, and intermediate cost. Theintermediate-cost projections are those com-monly used by commentators and analysts onSocial Security, and it is these projections thathave come under fire.

The core of the crisis deniers’ argument isthat the trustees severely underestimate futureeconomic growth. Should these estimatesindeed turn out to be pessimistic and the econ-omy were to grow faster, the argument goes,wages would rise, payroll tax receipts wouldincrease, and the added revenue would keepSocial Security solvent indefinitely. In a word,the crisis would be phony.

Outside AnalysisSome critics of the trustees’ analysis even go

so far as to accuse them of actuarial malprac-tice, of violating basic actuarial standards. Forinstance, David Langer, a consulting actuary,accuses the trustees of breaching several rulesof the Actuarial Standards Board, particularlythose requiring actuaries to use both recent andlong-term dates in their projections.18 And in atwist, Langer and others question whether thesemethodological errors are in fact errors at all;the trustees’ projections, they assert, are in factpart of a deliberate conspiracy designed toundermine the Social Security system. Langer,for instance, declares:

The trustees tell the actuaries the deficitlevel they desire. The actuaries will thenput together the appropriate assumptionsand computations for the trustees’ annualreport. . . . The political trustees clearlyhad the motivation, opportunity, andmeans to advance the spurious concept of

Social Security bankruptcy, and the evi-dence suggests they used their strategicposition to further their goals.19

Mark Weisbrot also subscribes to this view.2 0

These accusations come despite the follow-ing pledge by Chief Actuary Harry Ballantynecontained in the 2000 Trustees Report:

The techniques and methodology usedherein . . . are generally accepted within theactuarial profession; and the assumptionsused and the resulting actuarial estimatesare, individually and in the aggregate, rea-sonable for the purpose of evaluating thefinancial and actuarial status of the trustfunds, taking into consideration the experi-ence and expectations of the program.21

But even if one accepts the unlikely notion that theSocial Security Administration’s professionalactuaries would be silently complicit with a cam-paign to discredit and destroy the program theywork for, it is difficult to discern the trustees’ moti-vation to dissemble once one considers who theyare: the secretaries of Labor, the Treasury, andHealth and Human Services; the Commissionerof Social Security; plus two outside trusteesappointed by the president. For Langer’s argu-ment to hold, we must accept the implausiblepremise that such people as Health and HumanServices secretary Donna Shalala and Labor sec-retary Alexis Herman are conspiring to privatizethe New Deal’s crown jewel program.

Given these types of charges, the public isfortunate in having access to two independentappraisals of the trustees’ methods and assump-tions, which should shed light on the reasonable-ness of their projections for the system. The firstwas commissioned by the government’s GeneralAccounting Office and conducted by theaccounting firm of PricewaterhouseCoopers(PwC) at the request of Rep. Jerrold Nadler, aprominent promoter of the theory that SocialSecurity’s “crisis” is merely the product offaulty projections.2 2

PwC compared the trustees’ actuarial meth-ods and techniques with those used in the pri-vate sector and those employed in making pro-jections for social insurance systems in Canadaand the United Kingdom, finding that “theintermediate long-range projections of theSocial Security trust funds were developed in a

4

5

manner consistent with generally acceptedactuarial methods and techniques and that theycomply with standards of actuarial practice.”23

Going beyond methods, PwC “found that theassumptions underlying the calculations of thelong-range actuarial projections included in thetrustees’ 1999 report contained no materialdefects because of errors or omissions and thatthey were individually reasonable.”24 In short, thePwC study concluded that, taken as a whole, thetrustees’ intermediate assumptions for SocialSecurity represent “state-of-the-art” techniquesapplied to reasonable underlying premises.

The second study was conducted by the 1999Technical Panel on Assumptions and Methods,appointed by the independent, government-chartered Social Security Advisory Board.25

The Technical Panel, chaired by the UrbanInstitute’s Eugene Steuerle, had a bipartisanmembership consisting of economists, actuar-ies, demographers and other experts on socialinsurance programs. The panel examined theprojections in the trustees’ 1999 report, whichin some ways differ from those in the mostrecent report. The panel recommended numer-ous methodological additions to the trustees’projections to make them both more accurateand more flexible, as well as recommendingreporting changes to make projections moreunderstandable to the public.

Some of the panel’s assessments regarding spe-cific variables will be discussed later, but oneoverriding point is worth making: the TechnicalPanel concluded that the trustees’ assumptions forSocial Security are, if anything, optimistic regard-ing the program’s future. In particular, the panelfelt that the trustees’ projections for declines inmortality rates, which affect life expectancies andthe size of the beneficiary population, werestrongly biased in favor of the program’s solven-cy. The trustees’ intermediate-cost estimates in the2000 Trustees Report project a 75-year actuarialdeficit of 1.89 percent of payroll. Actuarial bal-ance is the difference between the program’s ben-efit liabilities and its assets, including payroll taxrevenues and the Social Security trust fund,expressed as a percentage of payroll. (E.g.,because Social Security is funded with a 12.4 per-cent payroll tax, if the program’s actuarial balanceshowed a 2 percent deficit, we can assume it to beunderfunded by approximately one-sixth.) TheTechnical Panel’s recommended changes toassumptions regarding mortality rates, real wage

growth, and the return on government bondswould increase the program’s actuarial deficit toapproximately 2.5 percent of payroll.2 6

In accusations similar to those made againstthe trustees, Baker and Weisbrot accused theTechnical Panel of “political manipulation,” cit-ing in particular panel head Eugene Steuerle’sservice in the Reagan administration. Baker andWeisbrot failed to mention Steuerle’s priorservice to Democratic administrations, nor thepanel’s broad and bipartisan membership.27

By walking through the trustees’ assump-tions regarding several important factors affect-ing Social Security’s future, we should gain abetter idea whether and to what degree the cri-sis deniers are correct. Following that, we willconsider whether increased economic growth,however likely it may be, would truly saveSocial Security from crisis. Together, they showthat substantially increased economic growth isless likely than the crisis deniers suppose, buteven if the economy grows far more quicklythan the trustees predict, Social Security willstill face massive funding shortfalls in thefuture. In short, faster economic growth isunlikely, but even if it comes it will not saveSocial Security.

Productivity and Real Wage GrowthProductivity and real wage growth are

cousins as far as Social Security is concerned.Productivity growth measures changes in out-put per worker, while real wage growth meas-ures changes in earnings per worker. Not sur-prisingly, the two often move in tandem: asworkers produce more, they tend to be paidmore. And because Social Security is fundedout of a 12.4 percent payroll tax, workers whoearn more tend to pay more into SocialSecurity.28 Pessimistic productivity and realwage projections would underestimate SocialSecurity’s revenues, worsening its financingposition in the short term.

The trustees’ intermediate assumptions pro-ject that over the next 75 years labor productiv-ity will improve at an annual rate of 1.5 percent,which means that in any period of time theaverage worker could produce 1.5 percent moregoods and services than in the prior year. Arethe trustees’ productivity estimates for thefuture compatible with past experience?Productivity growth in the last several years hastruly been impressive, with nonfarm output per

hour growing at 2.9 percent annually in 1998and 1999, and a spectacular 5.3 percent in thesecond quarter of 2000.29 This growth leadssome to conclude that the trustees are beingunrealistically conservative in their projections.

But, as Table 1 shows, placed in the contextof the past four decades the trustees’ projected1.5 percent productivity growth rate appearsmore reasonable. Productivity from 1959 to1998 increased at an annualized rate of 1.9 per-cent, while from 1979 to 1998 productivityincreases averaged less than 1.4 percent.30 Highproductivity growth has indeed been a veryrecent trend, and some even attribute the report-ed rise to mismeasurement of computer-relatedservices during preparations for the “Year2000” problem.3 1

It is true that productivity in the 1940s and1950s was substantially higher than that pro-jected for the next 75 years. But just as a sportsgambler counts a team’s recent wins and lossesmuch more highly than those of seasons past,the further back in history you look the lessresemblance the economy of that day has totoday’s economy and, presumably, to that of thefuture.3 2 Using recent history as the judge, thetrustees do not appear unreasonably pessimisticin their base productivity assumption.

Nevertheless, the impressive productivityincreases of the past several years deriving fromcomputerization have caused some commenta-tors to predict that the low productivity periodthat began in 1973 has ended, leading to a “NewEconomy” of permanently higher productivityand economic growth.33 But often ignored is thatrecent productivity increases have derived large-ly from increased productivity in the production

of computers—i.e., faster computers at a lowerprice—not increased productivity in the econo-my as a whole based on the use of computers. AsNorthwestern University economist Robert J.Gordon explains:

There has been no productivity growthacceleration in the 99 percent of theeconomy located outside the sectorwhich manufactures computer hardware,beyond that which can be explained byprice remeasurement and by a normal(and modest) procyclical response.Indeed, far from exhibiting a productivi-ty acceleration, the productivity slow-down in manufacturing has gotten worse;when computers are stripped out of thedurable manufacturing sector, there hasbeen a further productivity slowdown indurable manufacturing in 1995–99 ascompared to 1972–95, and no accelera-tion at all in nondurable manufacturing.3 4

The Congressional Budget Office largely con-curred with this view, concluding that “estimat-ing and projecting labor productivity in themedium term is best accomplished by model-ing technological change in the computer sectorseparately from that in other sectors.”3 5 Thisargument is not to say that the benefits of the“New Economy” cannot or will not take hold,merely that they have not. Assuming a rosyfuture for the economy and for Social Securityon this basis would seem overoptimistic.

The 1999 Technical Panel on Assumptionsand Methods took a similar view to theCongressional Budget Office, stating thatrecent productivity bursts do not yet justifymajor revisions in projected productivity overthe long term.36 On the basis of recent experi-ence, the CBO projects productivity at 2.2 per-cent annually over the next 10 years, but warnsthat current productivity increases may be partof a larger cycle encompassing years of below-average productivity growth from 1992 to1995.3 7 If so, then lower productivity could beexpected to return in the future as the cycle iscompleted. The Brookings Institution’s HenryAaron, an opponent of Social Security privati-zation, agrees, saying that “given the history oftrend reversals, the Trustees’ practice of cau-tious and highly damped adjustments to newevents is the only prudent course.”38

6

Table 1Productivity Growth

Period Annual Increase

1959–98 1.9%1959–68 3%1969–78 1.8%1979–88 1.3%1989–98 1.4%Projected 1.5%

Source: 2000 Trustees Report, pp. 150–51.

7

However, a major shift could be close to tak-ing place. In a forthcoming study, DaleJorgenson of Harvard University and KevinStiroh of the Federal Reserve Bank of NewYork have reversed some of their earlier skepti-cism regarding the productivity benefits ofcomputerization to the economy as a whole.While they declare that “the ‘new economy’view that the impact of information technologyis like phlogiston, an invisible substance thatspills over into every kind of economic activity. . . is simply inconsistent with empirical evi-dence,” Jorgenson and Stiroh nevertheless con-clude that a substantial portion of productivityincreases from 1995 to 1998 originated outsideof the information technology sector.39 Such afinding, if sustained, could lead to substantialrevisions in projected productivity growth forthe future. Whether and to what degree such arevision would affect Social Security will bediscussed in following sections.

The PwC analysis criticized the trustees’methods for projecting productivity growth ontwo fronts. First, PwC faulted the trustees’ useof the past 30-year period to project future pro-ductivity trends as arbitrary, pointing out thatthe period from 1970 to 1999 includes thestructural break of 1973–74, which ended thepostwar period of high productivity and com-menced two decades of much lower productiv-ity growth. The years before 1974 could beconsidered exceptional and excluding themwould lower projections of productivity for thefuture. Second, PwC also criticized the trusteesfor estimating productivity on an economywidebasis rather than examining individual sectorsof the economy and estimating total productiv-ity based on how these sectors are likely togrow or shrink over time. Following this proce-dure led the CBO to higher productivity esti-mates, at least for the period 2000–2009,though PwC speculates that this result mightnot hold over the long term.4 0

Any estimate of productivity growth over thelong term is problematic, particularly consider-ing that other factors such as labor force growthwill also change substantially over time.41 But itis incumbent upon those who argue that a peri-od of higher productivity growth is upon us toshow that the low-growth phase beginning in1973 has definitively ended. While recent expe-rience is encouraging, most believe it is tooearly to conclude that a new era of sharply

higher productivity growth has begun.Assuming that workers’ wages rise along

with their output, increased productivity oflabor leads to an increase in what the trusteescall the “real wage differential,” the differencebetween nominal wage increases and the con-sumer price index. This direct relationshipbetween productivity and wages may notalways be the case, as following sections willexplain. Nevertheless, just as the trustees esti-mate productivity growth to be slightly higherthan in the recent past their estimates of realwage growth are also higher. In fact, while from1975 to 1995 wages grew by just 0.48 percentannually after inflation, the trustees project realannual wage growth over the next 75 years at1.0 percent. As low as wage growth was fromthe mid-1970s onward, most objectiveobservers would not term a doubling of the realwage differential as “pessimistic.” Again, wagegrowth prior to the mid-1970s was generally farhigher than in the past quarter-century. But ifrecent experience is to count more heavily, thenthe trustees’ estimates appear reasonable.

The Technical Panel concluded that thetrustees’ 1999 projection of a 0.9 percent realwage differential was too low, recommending thata 1.1 percent differential was more appropriate.42

Following the Technical Panel’s report, thetrustees increased their estimate to 1.0 percent inthe 2000 report. The PwC analysis concurred thatthe trustees’ 1999 estimate of 0.9 percent was low,but did not specify a preferred figure.43

In a succinct declaration of the crisis deniers’claims, Baker and Weisbrot assert that, “usingany remotely realistic projection for the growthof wages and the economy, the Social Securitysystem will be solvent into the stratosphere ofAmerica’s science-fiction future.”44 The trustees’sensitivity analysis of wage growth makes thisclaim easy to verify.4 5 For Social Security toremain technically solvent over the next 75 yearsdemands real wage growth of 2.89 percent annu-ally,46 a rate 41 percent faster than during the1960s, when a surging economy pushed grossdomestic product growth to 4.5 percent annual-ly.47 To keep Social Security actuarially solventindefinitely requires permanent real wagegrowth of 5.7 percent annually and GDP growthtopping 6 percent.4 8 Moreover, this assumes thetrust fund to be a real economic asset, which fol-lowing sections will show not to be the case. Inshort, solvency into the science-fiction future

demands science-fiction rates of economicgrowth. In the world of fact, such growth ratesappear so implausibly high that even the mostardent advocates of the “New Economy” darenot even hope for them.

Hence, Steuerle and John Bakija are correctin warning that, “although economic growth isalmost always advantageous, one should not bemisled regarding what it would achieve. Evenvery high rates of economic growth would notautomatically solve the problems of imbalancein the Social Security system.”49 Of course, anyprojection extending 75 years into the future isjust that: a projection. And the trustees’ 1.0 per-cent real wage estimate may well be seen assplitting the difference between the 1.4 percentannual wage growth from 1969 to 1974 and the0.5 percent growth in the following twodecades. But whatever the differences of opin-ion regarding the proper rate of assumed realwage growth, no objective observers predictwage growth high enough to keep the systemsolvent indefintely. To remain complacent re-garding Social Security’s problems in the hopethat unparalleled economic and wage growthwill come to the rescue seems rash, at best.

Fertility, Immigration, and Labor ForceGrowth

The principal source of the trustees’ pro-jected decline in economic growth over thenext 75 years is not falling output per workerbut simply a reduced number of workers. Inother words, the trustees project extremelyslow growth of the labor force. If a largernumber of workers equals a larger economythen, all other things being equal, a smallernumber must mean the converse. In fact, prac-tically all of the projected slowdown in eco-nomic growth can be traced to slow laborforce growth.

The baby boom following World War IIpushed labor force growth rates to over 2.5 per-cent annually in the 1970s. But the first babyboomers can begin taking early retirement in2008 and fewer new workers will be availableto take their place. The trustees project thatlabor force growth in 2020 will be just one-sev-enth those from 1960–2000, and by 2045 laborforce growth will fall to a mere 0.24 percentannually (Figure 1).

As the large baby boomer generation retiresand succeeding generations of retirees grow

larger because of increased life spans, thenation will have practically no increase in thenumber of workers to support them. FederalReserve Board Chairman Alan Greenspan testi-fied recently to the Senate Select Committee onAging:

The expected slowdown in the growth ofthe labor force, the direct result of thedecrease in the birth rate following thebaby boom, means that financing ourdebt—whether explicit debt or the implic-it debt represented by Social Security andMedicare’s contingent liabilities—willbecome increasingly difficult.5 0

For instance, the labor force today consists of153.5 million workers, compared to a benefici-ary population of 38.2 million. By the year2050 there will be 105 percent more beneficiar-ies, but just 21 percent more workers.5 1 Unlesssteps are taken now for the future, that relative-ly small labor force of the future will face a bur-den far more onerous than that borne by work-ers today.

Variations in labor force growth have twoprimary sources: changes in the fertility rateand changes in net immigration levels. Let usexamine them in turn.

The Fertility Rate.The principal determinant ofthe size of the labor force is the fertility rate, theaverage number of children each woman bearsduring her lifetime. A higher fertility rate shouldlead in time to a higher number of workers payinginto Social Security. For the intermediate projec-tions, the current fertility rate of 2.06 children perwoman is expected to decline by 2024 to the long-term rate of 1.95 children per woman (Figure 2).A fertility rate of 2.1 is considered necessary foran advanced country to maintain its population;hence, immigration will be needed simply to keepthe U.S. population stable.

The Technical Panel examined the trustees’ fer-tility rate projections for the 1999 Trustees Report.The panel acknowledged recent increases in thefertility rate, but speculated that they could be theresult of women choosing increasingly to havechildren at a later age. This trend would produce ashort-term increase in births but not necessarily along-term increase in the fertility rate. The panelnoted that, “The persistence of rates above 2.0during the past decade suggests that the assumedintermediate rate of 1.9 [in the 1999 Trustees

8

9

0

0.5

1

1.5

2

2.5

3

1960

1965

1970

1975

1980

1985

1990

1995

2000

2005

2010

2015

2020

2025

2030

2035

2040

2045

2050

2055

2060

2065

2070

Figure 1Labor Force Growth

Five-Year Average

Source: 2000 Trustees Report, Table II.D.1.

1

1.5

2

2.5

3

3.5

4

1940

1945

1950

1955

1960

1965

1970

1975

1980

1985

1990

1995

2000

2005

2010

2015

2020

2025

2030

2035

2040

2045

2050

2055

2060

2065

2070

2075

Figure 2Fertility Rate, 1940–2075

Source: 2000 Trustees Report, Table II.D.2, projected from 2000 to 2075.

Report] may be too low, but that rate appears to bereasonable over long periods of time.”5 2 Hence,the panel “recommends no change now in theintermediate assumption.”53 However, in the 2000Trustees Report the ultimate total fertility rate pro-jection increased to 1.95, from 1.90 in the 1999report.

The PwC analysis exposed a possible contra-diction in the trustees’ fertility estimates. Thetrustees expect that the differing fertility rates ofethnic groups in the United States will convergeover time, on the basis that fertility rates derivemore from income levels than from cultural atti-tudes.54 But the trustees simultaneously believethat cultural factors distinct to the United Stateswill keep overall American fertility rates wellabove those of other developed nations. In otherwords, the trustees appear to believe that incomedetermines differences in fertility rates within theUnited States, but differences in fertility ratesbetween the United States and other countries arebased upon culture. On one hand, if fertility ratesindeed derive from culture, then the trustees’ esti-mates of fertility levels should incorporate projec-tions of the future ethnic makeup of the UnitedStates, which they currently do not.

On the other hand, if fertility rates derive fromincomes, then U.S. rates may fall closer to those ofother developed countries. At present, U.S. fertilityrates are 40 percent higher than those in EuropeanUnion countries, according to United Nationsdata.5 5If U.S. fertility rates fell only to the levels ofthe United Kingdom, Social Security’s long-termdeficits would increase by 16 percent.56 Were U.S.fertility to come to more closely resemble that ofSpain, Italy, or Germany, Social Security’s deficits

would be 38 percent higher than currently predicted.The trustees’ optimistic low-cost fertility estimate,which would reduce the program’s long-term deficitby 14 percent, still would make U.S. fertility higherthan in any European Union country, and substan-tially higher than the average. International trendshint that the trustees’ intermediate projections forfertility rates have far more room to fall than to rise.

Immigration. A second factor determining laborforce increases is the level of immigration into theUnited States from other countries. Higher immi-gration rates increase the workforce, providing aboost to Social Security’s finances. As shown inTable 2, immigration rates today are relatively highby the standards of recent history. For 1998 and1999, net legal immigration (minus emigration) isestimated to total 495,000. Total immigrationincluding illegal immigrants is estimated at 795,000.The trustees project future total immigration to beapproximately one-eighth higher annually on anominal basis than at present, at 900,000 yearly.57

Changes in immigration rates are difficult topredict. Legal immigration levels are set by law,and the trustees’ intermediate-cost projectionsassume that levels of immigration compatiblewith present law will continue into the future. Onecan imagine that public sentiment againstincreased immigration might keep immigrantquotas at current levels, or alternately that low fer-tility rates among native-born Americans mightcreate economic pressure for higher immigrationquotas or increased illegal immigration to preventlabor shortages. For these reasons, the TechnicalPanel recommended no changes to the trustees’intermediate assumptions for immigration but didrecommend that the range of estimates contained

Table 2Immigration by Decade

Period Immigrants per 1,000 Population

1930s 0.41940s 0.71950s 1.51960s 1.71970s 2.11980s 3.11991–97 3.8

Source: Statistical Abstract of the United States, 1999, p. 10.

10

11

in the high- and low-cost projections be madebroader to reflect the issue’s uncertainty.58 Ineither case, the effects of immigration policyextend far beyond Social Security, touching emo-tional chords among some regarding the characterof the nation and presenting possible ancillarysocial and governmental costs and benefits thatmust be factored into the equation.

At any rate, increased legal immigration pre-sents paltry net benefits to Social Security financ-ing because each legal immigrant worker eventu-ally would become eligible to collect benefits.Each 100,000 immigrants above the 900,000assumed in the intermediate-cost projectionsimproves Social Security’s long-range actuarialbalance by just 0.05 percent of taxable payroll.59

Given Social Security’s 75-year actuarial deficit of1.89 percent of payroll, immigration would needto top 4.68 million annually over the next 75 yearssimply to keep the system in balance. For this rea-son, hopes expressed by some that increasedimmigration could keep Social Security healthyover the long term are largely misplaced.60

Another Way of Looking at the Economy: PerCapita Growth. When one hears that thetrustees project future economic growth to bejust half that of the recent past, it is easy to envi-sion them predicting a permanent recession.Because such a long-term decline in economicwell-being is implausible to many, it becomes

easier to reject the trustees’ projections forSocial Security as well.

But the trustees’ economic assumptions appearmore reasonable when we move away from eco-nomic growth measured for the economy as awhole and focus on GDP growth per capita,which is a more accurate measure of the materialimprovement of people’s lives. Indeed, someeconomists consider per capita growth the onlyrelevant measure of economic improvement.Economist Thorvaldur Gylfason declares that “anincrease in the labour force as such does not real-ly count as a source of economic growth, becausewhat matters for a nation’s standard of living is notthe growth of national economic output per se, butrather of output per capita.”61

By that measure, the trustees’ view of theeconomic future is not quite as grim as somecritics claim. While total GDP will grow moreslowly because of reduced labor force growth,historical data for 1960–99 and the trustees’projections for 2000–2075 show that GDP percapita will continue to grow at a reasonable ifnot spectacular rate (Figure 3). And GDP percapita will understate the improvement in theearning power of the average worker, as it willbe diluted by the increased number of non-working retirees. As noted earlier, projectionsfor real wage growth at twice the rate measuredfrom 1975 to 1995 means that individual work-

0

10000

20000

30000

40000

50000

60000

70000

80000

1960

1964

1968

1972

1976

1980

1984

1988

1992

1996

2000

2004

2008

2012

2016

2020

2024

2028

2032

2036

2040

2044

2048

2052

2056

2060

2064

2068

2072

Figure 3Real Gross Domestic Product per Capita

Source: 2000 Trustees Report, Tables II.H1, III.C1; 2000 Economic Report of the President, 2000–2075, projected.

Year

ers can see the trustees’ view of their future asrelatively bright.

Examining the factors determining econom-ic growth leads one to conclude that thetrustees’ long-term projections are at leastreasonable. The Technical Panel and PwC’sanalysis largely concur. While a 1.74 percenteconomic growth rate over the next 75 years iswell below the historical average, a relativeslowdown in total economic growth is unavoid-able when workforce growth is practically nil.

Life Expectancy. The third major factorinfluencing Social Security’s future solvency islife expectancies. However, falling mortalityrates (which determine longevity) receive littleattention from those who wish to deny thatSocial Security faces problems. One reason forthe crisis deniers’ inattention to life expectan-cies may be that, in this case, their own argu-ment works against them.

Many experts believe the trustees’ projec-tions for life expectancies greatly underesti-mate the program’s problems. Worse yet, of thethree major variables influencing SocialSecurity’s financing—labor force growth, realwage increases, and mortality rates—mortalityrates have by far the most effect.

Rising life expectancies are an unqualifiedblessing for the U.S. population. But unlikechanges in labor force and productivity growth,which generate both benefits and costs for SocialSecurity, increased longevity is almost complete-ly detrimental to Social Security’s financing.After all, most reductions in death rates todaytake place not during childhood or working yearsbut after an individual has retired, resulting in alarger beneficiary population without firstincreasing the number of workers paying into thesystem.62 The trustees note:

[Although] lower death rates cause boththe income (as well as taxable payroll)and the outgo of the OASDI program tobe higher than they would otherwise be,the relative increase in outgo . . . exceedsthe relative increase in taxable payroll.For any given year, reductions in thedeath rates for people who have attainedthe retirement eligibility age of 62increase the number of retired-workerbeneficiaries without adding significantlyto the number of covered workers. . . .Consequently, if death rates for all ages

are lowered by about the same relativeamount, outgo increases at a rate greaterthan the rate of growth in payroll, therebyresulting in higher cost rates.63

The trustees assume a 41 percent reduction indeath rates as part of their intermediate costestimates. Each additional 10-percentage-pointreduction in death rates decreases the long-range actuarial balance by about 0.34 percent oftaxable payroll.64

This makes mortality rates the most powerfulof the three major variables. Figure 4 shows thatshifting mortality rates from the intermediate toeither the high- or low-cost assumption altersoverall actuarial balance by an average of 0.74percent of payroll in either direction. By contrast,real wages alter actuarial balance by only 0.51percent of payroll, and immigration and fertilitycombined alter it by just 0.42 percent of payroll.Consequently, changes in mortality rates havegreat potential to affect the system’s financing,for good or ill.

And many experts believe that the intermedi-ate assumptions for mortality rates substantiallyunderestimate future increases in life expectan-cies and therefore give an unnecessarily rosyview of Social Security’s future. Historicalexperience lends prima facie credence to thisidea. A simple regression trend line of historicalchanges in life expectancies at birth and at age65, as in Figures 5 and 6, shows that the trusteesanticipate a substantial slowdown in the growthrate of life expectancies. For instance, should thetrend from 1940 to the present continue, by 2075the average life expectancy at birth would bealmost 93 years, compared to 83 years as pro-jected by the trustees. Likewise, total lifeexpectancies for individuals reaching age 65would exceed 88 years if current trends continue,rather than slightly over 86 years as projected bythe trustees. (One should not be confused by thefact that projected life expectancies at birthexceed life expectancies for those reaching age65. Remember that the estimates apply to indi-viduals born or aged 65 in that year. An individ-ual born in any given year will presumably bethe beneficiary of medical progress that an indi-vidual aged 65 in that year will not be.) In brief,while overall historical trends were pushedupward by spectacular longevity increases in the1940s, the trustees project that total lifeexpectancies will increase at a rate less than half

12

13

0.13

0.29

0.51

0.74

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

Immigration Fertility Real Wages Mortality

Figure 4Effect of Variables on Actuarial Balance

Source: 2000 Trustees Report, Section II.G.

Note: Figures represent the average change in actuarial balance (as a percentage of taxable payroll) resulting from alter-ing each variable from the intermediate-cost to the high- and low-cost estimates. A larger change indicates a greaterdegree of influence on overall solvency.

60

65

70

75

80

85

90

95

1940

1945

1950

1955

1960

1965

1970

1975

1980

1985

1990

1995

2000

2005

2010

2015

2020

2025

2030

2035

2040

2045

2050

2055

2060

2065

2070

2075

Figure 5Life Expectancy at Birth (dashed line is 1940–2000 trend)

Source: 2000 Trustees Report, Table II.D2.

that during the past six decades.International data also indicate that the

trustees’ mortality assumptions may be opti-mistic with regard to Social Security’s financ-ing. Ronald Lee of the University of California-Berkeley and Shripad Tuljapurkar of StanfordUniversity note that “the rates of [mortality]decline projected by SSA are very substantiallylower than any other country has experienced inthe period 1975–79 to 1985–89, except for70–75 year olds in the Netherlands.”65 Lee andTuljapurkar show that in many cases thetrustees’ projections for mortality declines areone-half, to one-third, to even one-fifth as highas those experienced in other developed coun-tries. The Technical Panel cited some of theseinternational comparisons in its own examina-tion, noting that according to the trustees:

Life expectancy at birth for U.S. femaleswill not reach the level currently enjoyedby French women in 1995 until 2033; bySwedish women until 2026; and byJapanese women, until 2049. For U.S.males, the corresponding dates are 2002,2026 and 2029. It is difficult to under-

stand why the United States should lag sofar behind other countries.6 6

International comparisons, Lee and Tuljapurkarassert, “provide strong evidence that U.S. mor-tality decline is not yet pushing up against bio-logical limits or against limits imposed byalready existing medical technology. In ourview, the SSA forecasts of mortality decline arefar too low, and even the SSA upper bracket[high-cost] for rates of decline of mortality istoo low.”6 7

In response to these types of issues, the 1999Technical Panel, like earlier panels, “stronglyrecommend[ed] efforts toward stochastic mod-eling or similar techniques that are better ableto capture the interrelationship among assump-tions.”68 Stochastic methods, unlike thetrustees’ simple menu of three outcomes—highcost, intermediate cost, and low cost, can modelan infinite range of possible outcomes and esti-mate the probability of each outcome. In addi-tion, these more sophisticated methods couldbetter model the interaction between variables,which under the trustees’ current techniquessometimes do not make full sense.69

14

76

78

80

82

84

86

88

90

1940

1945

1950

1955

1960

1965

1970

1975

1980

1985

1990

1995

2000

2005

2010

2015

2020

2025

2030

2035

2040

2045

2050

2055

2060

2065

2070

2075

Figure 6Life Expectancy at Age 65 (dashed line is 1940–2000 trend)

Source: 2000 Trustees Report, Table II.D2.

15

Specifically, the Technical Panel pointed toresearch by Lee and others who have constructedstochastic fertility models applicable to SocialSecurity.70 Lee and Lawrence Carter estimate totallife expectancies in 2050 at 86.1 years, almost fiveyears greater than estimated by the trustees.71 Leeand Tuljapurkar’s analysis leads them to concludethat, “SSA forecasts . . . foresee implausibly smallgains to life expectancy over the next 75 years.”Because the authors calculate that each year’sincrease in life expectancies requires a 3.6 percentincrease in payroll taxes to maintain solvency,underestimates of future life expectancies havelarge potential effects on Social Security’s financ-ing position.72

Drawing on historical U.S. and internationaltrends as well as more sophisticated analytical tech-niques such as Lee’s, the Technical Panel concludedthat “historical rates [of mortality declines] providea prudent intermediate forecast, although they cur-rently correspond more closely—at least in lifeexpectancy at birth after a few decades—to the SSAhigh-cost assumption for mortality.”73 If the trustees’high-cost assumptions should prove closer to themark, as the panel believes they will, that alone

would increase Social Security’s total actuarialdeficit by almost one quarter.74

If myriad advancements in health and medi-cine bear fruit, Social Security’s total fundingdeficits could far exceed those predicted in thetrustees’ intermediate assumptions. Healthierlifestyles, improved diets, and research like thehuman genome project are unequivocally posi-tive. But rather than “a crisis that doesn’t exist,”Social Security’s problems could prove evenworse than many people think.

If Economic Growth ExceedsProjections, Will It

Save Social Security?The trustees’ economic and demographic

estimates for the future appear at least reason-able, so it is safe to anticipate that SocialSecurity’s deficits will appear and in roughly theproportion projected by the trustees. Figure 7shows Social Security’s projected surpluses ordeficits as a percentage of taxable payroll. Whilein surplus today, the program will slip into deficit

-7

-6

-5

-4

-3

-2

-1

0

1

2

3

2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 2060 2065 2070 2075

Figure 7Social Security’s Payroll Tax Surplus or Deficit

Source: 2000 Trustees Report, Table II.D2.

Note: Does not include income from the Social Security trust fund, which would keep the system technically sol-vent until the year 2037.

by the year 2015. Even if payroll tax surplusesbefore 2015 are truly saved in the trust fund,which there is little reason to believe they will be,Social Security is projected to run a deficit aver-aging 1.89 percent of taxable payroll over thenext 75 years.7 5

But what if today’s good economic times trulyportend the dawn of a “New Economy”? Couldhigher economic growth in the future save SocialSecurity? Many on both ends of the politicalspectrum seem to think so. While the crisisdeniers tend to fall on the political left, some inthe supply-side wing of the Republican Partyshare the view that higher economic growth isthe key to solving Social Security’s problems.For instance, commentator Lawrence Kudlowstresses this advice to the 2000 Republican pres-idential nominee:

Bush should . . . carefully explain why taxcuts, not federal-debt elimination, willstrengthen Social Security. Voters mustunderstand that in coming years only twoworkers will be available to support eachretiring beneficiary. Therefore, workerproductivity and economic expansionmust be maximized.7 6

It should be obvious why supply-side econo-mists, who emphasize cuts in marginal tax ratesas the key to increased economic growth,would be drawn to this view on Social Security.After all, funds that could be applied to SocialSecurity reform—either to establish personalretirement accounts or retire public debt—could instead be dedicated to income tax ratecuts. Former Treasury Department economistAldona Robbins echoes Kudlow’s view:

Despite dire predictions, a healthy econo-my can save Social Security. The mostobvious reason is that a faster-growing,lower-inflation economy brings morepayroll-tax revenue into Social Securitycoffers while keeping cost-of-livingadjustments, or COLAs, under control.77

This argument has both intuitive and theoreticalappeal. Intuitive, because increased economicgrowth makes almost any problem easier tosolve. Theoretical, because economic growth iscentral to the rate of return paid by a pay-as-you-go pension system like Social Security. As

the economy grows, workers’ wages increase.Since payroll taxes are levied as a percentage ofwages, increased wages translates intoincreased payroll tax revenues.

But is it that simple? If the economy growsfaster, will that necessarily translate into largerpayroll tax revenues? All other things beingequal it will; but in many cases, all other thingsare not equal, for there are several stepsbetween higher economic growth and increasedpayroll tax revenue to Social Security. As thetrustees note, “Projections of taxable payrollreflect the projected growth in GDP, along withassumed changes in the ratio of worker com-pensation to GDP, the ratio of earnings toworker compensation, the ratio of OASDI cov-ered earnings to total earnings, and the ratio oftaxable to total covered earnings.”7 8 Changes inany of these ratios could alter Social Security’sincome in the future. Let us examine three ofthem in turn.

The Employee Share of National IncomeThe first question is how total economic

growth translates into employee compensation.National income is split between capital own-ers, who earn profits on their investments, andemployees, who are paid wages and other ben-efits for their labor. Declines in labor’s sharecould mean that, even if economic growthincreases national income, real compensationto workers would not increase proportionately.

While, as Figure 8 shows, the employeeshare of national income has declined slightlyover the past 25 years,the longer-term trend hasbeen for workers to take a slightly larger shareof national income. The trustees assume thatthe current allocation of national income toemployee compensation will remain stable.The Technical Panel’s methodology indicatedthat total compensation would continue a slightdecline as a share of GDP, but the panel madeno formal recommendation to that end.7 9

Changes in either direction could alter SocialSecurity’s payroll tax revenues in the future.

Wages vs. CompensationWhile total compensation to workers as a share

of national income is important, it should not beconfused with workers’ take-home wages. Totalcompensation includes health benefits, payrolltaxes paid by employers, and other costs that arenot included in the worker’s paycheck and thus

16

17

50

55

60

65

70

75

80

1929

1934

1939

1944

1949

1954

1959

1964

1969

1974

1979

1984

1989

1994

1999

Figure 8Employee Compensation as Share of National Income

Source: Bureau of Economic Analysis, National Income and Product Accounts.

80

85

90

95

100

1929

1934

1939

1944

1949

1954

1959

1964

1969

1974

1979

1984

1989

1994

1999

Figure 9Wages and Salaries as Share of Total Worker Compensation

Source: Bureau of Economic Analysis, National Income and Product Accounts.

are not subject to payroll taxes. While totalemployee compensation has risen as a share ofnational income, workers’ wages as a share ofnational income have fallen substantially.

The reason is that, as health care costs andpayroll taxes have risen, a smaller and smallershare of workers’ total compensation has takenthe form of the wages and salary they receive intheir paychecks. Despite an increase in recentyears, wages have fallen from 95 percent oftotal compensation in 1950 to less than 84 per-cent today (Figure 9). This change produces theseemingly puzzling result that, while totalemployee compensation rose from 67 to 71 per-cent of national income from 1955 to 1999,take-home wages declined from 64 percent to59 percent of national income in the same peri-od.80 The trustees expect that the wage share oftotal compensation will continue to decline at arate of 0.2 percent annually over the 75-yearperiod.81 If health care costs increase more thanprojected, as some commentators insist theywill,8 2 then the wage share of total compensa-tion could fall further. A greater share of theeconomy would shift from the taxable to the

nontaxable sector, where it would not benefitSocial Security, regardless of the level of eco-nomic growth.

Taxable vs. Nontaxable EarningsFinally, total wages do not themselves form

the tax base for Social Security. SocialSecurity’s 12.4 percent payroll tax rate appliesonly to wages and salaries up to a limit, cur-rently at $76,200.83 Any wages over that levelare not subject to Social Security payroll taxes(nor are they credited toward benefits). Hence,even if economic growth raises total compensa-tion, and total compensation raises wages, thatdoesn’t necessarily mean an increase in wagessubject to payroll taxes.

Taxable earnings reached their peak at 90percent of total covered earnings in 1982 and1983 and have declined somewhat since thatpoint, such that they now comprise 85 percentof total covered earnings (Figure 10).84 Thetrustees project that this declining trend willcontinue at a slower rate until 2009, then holdsteady throughout the remainder of the 75-yearperiod.8 5 Should the trend continue past 2009,

18

60

65

70

75

80

85

90

95

1950

1953

1956

1959

1962

1965

1968

1971

1974

1977

1980

1983

1986

1989

1992

1995

1998

Figure 10Taxable Earnings as Share of Total Covered

Source: 1999 Annual Statistical Supplement, Table 4.B1.

19

Social Security’s payroll tax revenues will fallshort of projections, even if projected wagegrowth rates are unchanged.

Many on the left complain about this trend inthe distribution of income, without noting theeffect it might have on Social Security. Forinstance, the Economic Policy Institute and theCenter for Budget and Policy Priorities released“Pulling Apart: A State-by-State Analysis ofIncome Trends,” which noted the following:

Despite the strong economic growth andtight labor markets of recent years, incomedisparities in most states are significantlygreater in the late 1990s than they were dur-ing the 1980s. The average income of thelowest-income families grew by less thanone percent from the late 1980s to the late1990s—a statistically insignificant amount.The average real income of middle incomefamilies grew by less than two percent,while the average real income of high-income families grew by 15 percent.86

Likewise, Isaac Shapiro and Robert Greensteinof the Center on Budget and Policy Prioritiesanalyzed Congressional Budget Office taxreturn data, concluding that from 1977 to 1999the average after-tax income of the top fifth ofhouseholds increased 43 percent, the middlefifth increased only 8 percent, and the bottomfifth declined by 9 percent.8 7 Whether the realincomes of the bottom fifth in fact declined isopen to question; many argue that officialmeasures overstate the rate of inflation, whichwould tend to underestimate levels of wage,income, and economic growth. Nevertheless,these numbers show that whatever wage growthhas been occurring has been more heavily con-centrated on the top end of the distribution.

These writers intend their analysis to supportpolicies that would redistribute income fromthe well-off to the poor. Unintentionally, how-ever, they show the difficulty in asserting thateconomic growth will save Social Security. Ifhigher economic growth manifests itself asincreases in wages not subject to payroll taxes,then Social Security will be none the better offfor them.88 The lesson: you cannot pay taxes onan income you do not have. Baker and Weisbrothave recently confronted this issue.

The gains from economic growth may not

be as obvious as they should be, mainlybecause the majority of employeeshaven’t been sharing in them. Over thelast 26 years, the typical wage or salaryhas stagnated in real terms. . . . What thismeans is that reclaiming the majority’sshare of the economic pie is the real“challenge and opportunity of the 21stcentury”. . . . Yet the question of incomedistribution has been removed from thepolitical agenda.8 9

There is nothing technically inaccurate in thisstatement, as wage growth has indeed slowedover the past quarter century. But if SocialSecurity’s financing health depends not only onspectacular economic growth but also on a newnational commitment to income redistribution,then the “Don’t Worry, Be Happy” tenor of therest of Baker and Weisbrot’s argument hardlyseems justified.

Social Security’s trustees themselves estimatethat taxable payroll will decline as a percentage ofgross domestic product from 40.6 percent today to37.8 percent in 2035 and to only 35 percent in2075. Hence, economic growth, whatever it maybe, will not fully translate into wage growth andincreased payroll tax revenues.90

What If Payroll Tax Revenues Do Increase?Economic growth may not exceed the

trustees’ projections and, even if it does, growthmay not translate into increased revenues forSocial Security. But let’s assume that in the year2000 the economy’s growth exceeds expecta-tions, most of that extra income takes the formof taxable wages rather than nontaxable bene-fits like health care, and most of the taxablewages go to workers below the payroll tax ceil-ing. What then?

If taxable wage growth were to increase, thenpayroll tax revenues would increase as well. Allother things being equal, Social Security wouldrun a larger payroll tax surplus,9 1creating a cor-responding increase in the balance of the SocialSecurity trust fund.

This scenario would be unqualified goodnews for the system, except that workers whopay more taxes into the system are entitled tomore benefits from it. Any increase in payroll taxrevenues must be counted against correspondingincreases in benefit liabilities. Federal ReserveBoard chairman Alan Greenspan has comment-

ed on several occasions that if Social Securitywere run an on accrual basis, in which payrolltax revenues and benefit liabilities are counted atthe same time, it would be in deficit already.9 2

The first step to understanding the effect of awage increase on benefit obligations is to seehow benefits are determined in the first place.When calculating a worker’s benefits, theSocial Security Administration first determinesthe worker’s Average Indexed MonthlyEarnings (AIME). The AIME takes the work-er’s 35 highest earning years of employmentand adjusts each year’s earnings for increases inaverage wages. These wage-adjusted annualearnings are then averaged and divided by 12 toproduce adjusted monthly earnings.

Using the worker’s AIME, Social Security’sbenefit formula calculates what is called thePrimary Insurance Amount (PIA). The PIA isthe basic benefit that worker would receive, andother benefits—such as spousal benefits—arecalculated based upon it. The PIA is determinedby subjecting the AIME to what are called“bend points” (Figure 11). The bend pointsdetermine the portion of the worker’s AIME

that will be replaced by his retirement benefits.Under current law, the worker’s basic benefit orPIA would equal 90 percent of the first $531 ofhis AIME, 32 percent of his AIME between$531 and $3,202, and 15 percent of his AIMEin excess of $3,202. The bend points are thesource of Social Security’s progressivity, sincethey replace a greater portion of lower adjustedincomes than of higher.

Given this mechanism for determining bene-fits, how would an increase in wages affect thecalculations? In three ways: First, increasedwages in any particular year would raise theworker’s AIME simply due to averaging: ahigher wage in one particular year raises theaverage wage.

Second, because past wages are indexedaccording to subsequent wage increases, higherwage growth today means higher indexed earn-ings for all past years. The easiest way to under-stand wage indexing is through questions. Forinstance, indexing past wages for inflation wouldanswer the question, “What would my pastwages be if I were paid in today’s dollars?” Onthe other hand, indexing for total wage increas-

20

0

10

20

30

40

50

60

70

80

90

100

Figure 11Social Security’s “Bend Points”

$531 $3,202Average Indexed Monthly Earnings

Source: 2000 Trustees Report, p. 69.

21

es, which incorporate both inflation and produc-tivity gains, answers the question, “What wouldmy past wages be if I were paid in today’s dol-lars and I were as productive then as I amtoday?” Indexing past wages for total wagegrowth takes into account nominal increasesattributable both to inflation and to productivitygrowth. As a result, indexing the AIME for wageincreases effectively credits workers with wagesthat they never paid taxes on.

Third, the bend points used for calculating thePrimary Insurance Amount are indexed to wagesas well. This prevents the progressivity of thebend points from benefiting the program’sfinances. If the bend points were fixed or indexedonly to inflation, then increased wages wouldmean that a greater portion of workers’ AIMEswould fall under the bend points paying outlower replacement rates. For instance, assumethat wage growth pushed a low-income worker’sAIME from $531 to $551. If the bend pointswere fixed, then the worker’s PIA would bebased on 90 percent of the first $531 but just 32percent of the extra $20 per month. But since thebend points are indexed to wages, the bendpoints would move up as well and the overallreplacement rate for that worker would not fall.When these factors are all counted, an increase in

a worker’s taxable wages would create a rough-ly proportionate increase in that worker’s benefitentitlement, thereby negating much of the gainsfrom economic growth.

It is important to realize, however, that thesebenefit increases would take place only gradu-ally. When a worker’s AIME is calculated,wage increases only up to age 60 are indexed.Moreover, benefits for existing retirees increaseeach year only according to inflation; it is onlyfuture retirees whose benefits would beincreased by higher wage growth today. Sincethe earliest a worker can retire is age 62, thereis at least a two-year gap before any worker’sAIME substantially increases. Moreover, sincea worker’s benefits are based on his 35 highestearning years of employment, it could be wellover a decade before the full effects are felt.

But when wage growth’s effects are felt,Social Security will have to pay increasedbenefits. And when it does, the program’sdeficits might actually increase, as SocialSecurity Administration analysis shows.Figure 12 describes year-by-year payroll taxdeficits and surpluses based on the trustees’sensitivity analysis of real wage growth. Aspredicted, wage growth of 1.5 percent annual-ly, 50 percent higher than the intermediate-

-$800

-$700

-$600

-$500

-$400

-$300

-$200

-$100

$0

$100

$200

2000

2003

2006

2009

2012

2015

2018

2021

2024

2027

2030

2033

2036

2039

2042

2045

2048

2051

2054

2057

2060

2063

2066

2069

2072

2075

1.5 Percent

1.0 Percent

Figure 12Cash Flow under Intermediate and High Wage Growth Assumptions

Source: Social Security Administration (unpublished), based on wage sensitivity analysis in 2000.

cost projection, produces larger and longerlasting payroll tax surpluses in the near term.For instance, whereas Social Security wouldbegin running a payroll tax deficit in 2015under a 1 percent annual real wage growthassumption, deficits are delayed until 2017when wage growth is assumed to be 1.5 per-cent. While two years of additional solvencymay not seem to be much, even as late as2047 Social Security’s net annual cash flowwould be $30 billion higher under 1.5 percentgrowth than under 1 percent. But after 2047that advantage shrinks, as increased wagegrowth translates into increased benefit obli-gations. And from 2059 onward, SocialSecurity’s deficits would be larger under thehigh wage growth assumption than under theintermediate-cost assumption.

To make up those larger deficits woulddemand even higher wage growth, whichwould eventually lead to higher benefit obliga-tions. Hence, Social Security could be aptlydescribed by the Red Queen of Lewis Carroll’sThrough the Looking Glass, who tells Alicethat, “in this place it takes all the running youcan do, to keep in the same place.”93

Consequently, the results of a longstandingproductivity increase such as that envisioned byJorgenson and Stiroh, even if translated fullyinto wages, would be to defer Social Security’sproblems but not to eliminate them.9 4 Thisanalysis is supported by Steuerle, whodeclared:

different economic assumptions usuallydon’t have a substantial impact on [SocialSecurity’s] deficits. Crudely speaking,these programs are scheduled to growfaster when the economy grows faster,and slower when the economy growsslower. Although there are some lags, thisclose relationship between economic andprogram growth makes taxes and expen-ditures grow more or less in line and,hence, the difference between the two isnot affected so much by changes in eco-nomic growth.95

This position does not mean that permanentlyhigher wage growth would bring no benefits.But it is illusory to believe that increasedwage growth of any realistic degree can keepthe system solvent on an ongoing basis.

Actuarial BalanceGiven those observations, how can people

say that by raising wage growth, increased eco-nomic growth will save Social Security? Morespecifically, how can the trustees’ report saythat “each 0.5-percentage-point increase in theassumed real-wage differential increases thelong-range actuarial balance by about 0.50 per-cent of taxable payroll,”9 6 which implies thatreal wage growth of 2.89 percent annuallywould erase Social Security’s deficit over thenext 75 years?

These conclusions are drawn from thetrustees’ use of the idea of 75-year “actuarialbalance,” which is defined as “the differencebetween the summarized income rate and thesummarized cost rate over a given valuationperiod.”9 7 In other words, the trustees calculateSocial Security’s income rate, the taxes it willcollect as a percentage of payroll plus trust fundrevenues, and subtracts its cost rate of benefitspayable as a percentage of payroll. The differ-ence is its actuarial surplus or deficit.

One reason the idea of 75-year actuarial bal-ance leads to misleading conclusions is the timelag between the wage increase and the increasedbenefit liability it creates. For the sake of illustra-tion, imagine three 75-year time periods with dif-fering rates of real wage growth:

• Period 1: 1925–2000; annual wage growth,1 percent

• Period 2: 2000–2075; annual wage growth,2 percent

• Period 3: 2075–2150; annual wage growth,1 percent

As period 2 opens, wage growth increases from1 percent to 2 percent. Because Social Securitywould be collecting taxes on the basis of 2 per-cent wage growth while for a substantial periodpaying benefits earned during period 1, whenwage growth was just 1 percent, actuarial bal-ance during period 2 would improve.

But now go to period 3, when wage growthreturns to 1 percent. Growth of payroll tax revenuewould decrease, but during the time lag beforereduced benefit liabilities took effect SocialSecurity would collect taxes at the 1 percent wagegrowth level while paying benefits on the basisof 2 percent wage growth. Under these condi-tions, Social Security would have difficulty meet-ing benefit obligations incurred during period 2.

22

23

Benefits owed during any 75-year perioddepend upon wages earned prior to that period,and the ability to pay benefits accrued duringany 75-year period depends upon wage growthafter that period has ended. This time lag mat-ters in the current context because for the nextseveral decades Social Security will be payingbenefits based in whole or part upon liabilitiescreated during the slow wage growth yearsfrom 1973 onward. If wage growth increased,benefits would increasingly be based uponhigher wage growth than in the past, and bene-fit liabilities would rise as a result. Likewise,while 3 percent wage growth would technicallyeliminate Social Security’s actuarial deficitover the next 75 years, that level of wagegrowth must continue well beyond the year2075 if those workers who established benefitclaims prior to 2075 are to be paid.

Moreover, actuarial balance’s dependenceupon the Social Security trust fund can give adeceptive view of the time distribution of SocialSecurity’s funding shortfalls. From 2000 to 2024Social Security is projected to run a payroll taxsurplus of 0.1 percent of GDP, not including trustfund balances, while from 2025 to 2050 itsdeficit is projected to average 1.7 percent ofGDP and from 2051 to 2075 is expected to aver-age 2 percent of GDP. The current 75-year actu-arial deficit of 1.89 percent is meaningful only ifthe trust fund can accumulate resources andredistribute them over time. However, it cannot.

The Trust FundSeventy-five-year actuarial balance is a mis-

leading measure of the benefits and burdens ofthe Social Security program, for it assumes thatsurpluses today can be saved to make up fordeficits decades into the future. But mostexperts do not consider the trust fund in such away. If they are correct, then achieving actuari-al balance—either through increased economicgrowth or through policies designed to buttresstrust fund balances—could not be said to haveeffectively benefited the system or the workersand retirees who depend upon it.

The Clinton administration provides in its fis-cal year 2000 budget perhaps the clearest state-ment of what the trust fund can and cannot do:

[Trust fund] balances are available tofinance future benefit payments . . . butonly in a bookkeeping sense. . . . They do

not consist of real economic assets thatcan be drawn down in the future to fundbenefits. Instead, they are claims on theTreasury that . . . will have to be financedby raising taxes, borrowing from the pub-lic, or reducing benefits or other expendi-tures. The existence of large trust fundbalances, therefore, does not, by itself,have any impact on the Government’sability to pay benefits.9 8

In other words, the bonds in the trust fund makeup a claim on the existing pool of assets, but theydo not constitute meaningful assets in and ofthemselves. The reason? The payroll tax surplus-es that generated the bonds in the fund were notsaved. As the president put it in 1998, “Today,we’re actually taking in a lot more money fromSocial Security . . . than we’re spending out.Because we’ve run deficits, none of that moneyhas been saved for Social Security. . . .”9 9 Sincepayroll tax surpluses did not add to the econo-my’s capital stock, the bonds in the fund repre-sent a pledge by the government to redistributeincome from workers to Social Security benefi-ciaries, not an asset to make the system moreaffordable for those workers.

Alan Greenspan, who chaired the commis-sion whose recommendations led to the 1983amendments to build up trust fund balances,100

clearly lays out the case for running payroll taxsurpluses to increase the balance of the fund:

One reason to build surpluses in the trustfunds is to set aside savings and thus todivert part of the nation’s current produc-tion away from consumption, both privateand public. . . . They should boost invest-ment and thus foster the growth of thenation’s capital stock. And with more cap-ital per worker than would otherwise be inplace, productive capacity will be greaterand we will be better able to fulfill ourpromises to the retirees, while maintainingthe standard of living of future workers.101

This statement is true, given the provisoGreenspan adds: “assuming of course that thesurpluses are not offset by reductions in the sav-ing of households and businesses or by largerdissavings, that is, deficits, elsewhere in thefederal budget.” In other words, trust fundfinancing must increase net national savings if

it is to be a meaningful asset to the system. Carolyn Weaver, a member of the 1994–1996

Advisory Council on Social Security, makes pre-cisely this point: the trust fund’s fundamentalshortcoming is that “no mechanism in the lawensures that the surpluses translate into mean-ingful savings.”102 Whether and how muchnational savings increase depends not simplyupon the size of payroll tax surpluses. It alsodepends on whether those surpluses are dedicat-ed to reducing government debt, rather thanspending increases or tax reductions, andwhether individuals reduce their own savings inreaction to increased savings by the govern-ment.103 In 1990, Greenspan expressed doubt asto the net positive effects of trust fund surplusesto that point.104

Determining whether past payroll tax surplus-es increased national savings is inherently prob-lematic, demanding counterfactual suppositionsof what the government and private sector wouldhave done in the absence of those surpluses. Butregarding government spending, at least, Nobellaureate James Buchanan is confident that “asmall dose of public choice theory might havedampened the enthusiasm of those who soughtto ensure the integrity of the system” by usingpayroll tax surpluses to bolster national sav-ings.105 Trust fund surpluses can be spent on allthe projects and benefits that buttress a politi-cian’s electoral prospects, but because thesemonies derive from the sale of bonds to the fundrather than to the public they do not count towardthe budget deficit or toward general perceptionsof the public debt.106 Payroll tax surpluses’ abili-ty to mask on-budget deficits causes mostexperts to conclude that these surpluses relaxedfiscal discipline, reducing or eliminating thenational savings benefits that are their entire rea-son for being.

This plays into what fellow Nobel laureateMilton Friedman calls the “budget constrainthypothesis,” which states that “governmentsspend what governments receive plus whateverthey can get away with.”107 W. Mark Crain ofGeorge Mason University and Michael L.Marlow of California Polytechnic Universityperformed a statistical analysis of the correla-tion between changes in Social Security trustfund balances and overall federal spendingfrom 1940 to 1987; their results supportBuchanan and Friedman’s beliefs: “The evi-dence in support of the argument that excess

Social Security trust fund balances are savedand not spent,” Crain and Marlow found, “isweak or nonexistent.”108 More recently, HooverInstitution economist John Cogan found thatpast buildups of trust fund reserves have corre-lated with expansions of Social Security bene-fits, leading him to conclude that “unless amethod can be found for altering congressionalbehavior from its 60-year norm, any attempt toensure Social Security’s solvency by building alarge trust fund reserve will likely provefutile.”109 Vice President Gore’s election-yearproposal to increase Social Security benefits forwidows and working mothers provides anec-dotal evidence that this trend continues.110

If the above is correct, then the trust fundmechanism has increased paper obligations tothe Social Security system without a concomi-tant increase in the economic ability to fulfillthose obligations. As early as 1991 SocialSecurity actuaries gave this warning:

Due to continuing deficits in the rest ofthe Federal Government, we are not accu-mulating a true fund and are insteadmerely accumulating a right to futuregovernment revenues. The expected trustfund buildup will not (1) lower futurecosts, (2) lower total future taxes, or (3)generate faster economic growth (to makehigher future taxes easier to absorb).Under these circumstances, the public is,at the minimum, gaining a false impres-sion about the ability to prepare inadvance for the financial effects of thebaby boom’s retirement. In addition, theymay be gaining a false impression aboutthe financial resources that will berequired, after the baby boom retires, tofinance the program.111

In a similar vein, the Congressional BudgetOffice noted that “the size of the balances in theSocial Security trust funds—be it $2 trillion,$10 trillion, or zero—does not affect the obli-gations that the federal government has to theprogram’s beneficiaries. Nor does it affect thegovernment’s ability to pay those benefits.”112

The Brookings Institution’s Henry Aaronadmits that Social Security’s payroll tax surplus-es have not been saved for the future, but main-tains that the responsibility lies not with SocialSecurity but with the rest of the government:

24

25

While the Trust Funds have succeeded inadding to Social Security reserves, theymay have failed in adding to national sav-ing, if they caused government to runlarger deficits or smaller surpluses on therest of its activities. In short, unwise fiscalpolicy outside Social Security may haveprevented the accumulation of SocialSecurity reserves from increasing nation-al saving. If this unfortunate eventoccurred, however, the reason is not thatSocial Security reserves were invested ingovernment bonds, but because of impru-dent fiscal policy on activities of govern-ment other than Social Security.113

The point Aaron misses is that the government“other than Social Security” is the same gov-ernment that amended the program in 1983with the ostensible purpose of increasingnational savings. If trust fund surpluses do notin fact raise national savings to make paymentof future benefits more affordable, it makes lit-tle difference that the “fault” lies not withSocial Security but with the policymakers whodesigned and manage it. If the trust fund mech-anism has not effectively increased savings thenthose paper “assets” should be written off and amore effective means of saving to meet benefitobligations implemented.114

But what about the future? Now that the gov-ernment is running surpluses in the non–SocialSecurity budget, can the trust fund mechanismat last fulfill its intended function? The Clintonadministration’s Social Security proposalassumes that it can. The administration planwould dedicate payroll tax surpluses to debtreduction in hopes of raising national savingsand increasing the economy’s capacity to paybenefits in the future. The administration planwould issue bonds to the trust fund in exchangefor payroll tax surpluses, as in current practice,then use the cash derived from the bond sale toretire publicly held government debt. The trustfund would then be credited with additionalbonds equivalent to the annual interest savingsderived from that debt repurchase. One prob-lem with this is “double-counting.” The originalpurchase of bonds by the trust fund is econom-ically meaningful only if the cash from that saleis used to retire debt, and the interest paid onthose bonds already represents savings in debtservice costs. By issuing additional bonds to the

fund equal to the debt service savings, theClinton administration plan counts these sav-ings a second time.1 1 5Politically speaking, thisplan is understandable: unlike past practice, thegovernment would save payroll tax surplusesrather than spending them and would presum-ably want recognition for doing so. But simplybecause the past practice of spending payrolltax surpluses rendered trust fund bonds eco-nomically meaningless does not justify givingourselves extra credit for doing what we shouldhave been doing all along. Under the Clintonadministration plan, insofar as payroll tax sur-pluses are used to retire debt, trust fund bondsin the amount of those surpluses would be eco-nomically meaningful. Bonds deposited in thepast, as well as additional bonds credited to thefund under the administration plan, wouldremain obligations but not true assets.

A more important problem with the Clintonadministration plan is that, as in the past, thereis no assurance that payroll tax surpluses resultin increased government savings, much lessnational savings. It appears that in the adminis-tration plan additional bonds would be creditedto the trust fund regardless of whether any debtwas actually retired. General Accounting Officehead David M. Walker noted:

One disconcerting aspect of thePresident’s proposal is that it appears thatthe transfers to the trust fund would bemade regardless of whether the expectedbudget surpluses are actually realized.The amounts to be transferred to SocialSecurity apparently would be written intolaw as either a fixed dollar amount or as apercent of taxable payroll rather than as apercent of the actual unified budget sur-plus in any given year. These transferswould have a claim on the general fundeven if the actual surplus fell below theamount specified for transfer to SocialSecurity—and that does present a risk.116

Just as in the past, Congress could relax its fis-cal discipline and renege on its pledge toincrease savings without affecting the issuingof new bonds to the trust fund. This movewould again create the pretense of “savingSocial Security” without the reality.

Social Security reform that is based uponpersonal accounts, in which workers would

invest a portion of their payroll taxes in stocksand bonds, would have greater insulation fromsavings-offset issues currently affecting thetrust fund. Payroll taxes diverted to personalaccounts could not be used to mask the size ofon-budget deficits, thereby imposing greaterfiscal discipline on the government. And whilethe “wealth effect” and “Ricardian equiva-lence” would cause many high-income workersto dissave in reaction to the new deposits totheir personal accounts, for many low-incomeindividuals savings are already so low that fur-ther reductions in savings are less likely.117

Hence, while personal accounts would clearlynot guarantee a dollar-for-dollar increase innational savings, they have better firewalls toensure that the savings created by fundsdeposited in these accounts are not offset bydissavings elsewhere in the government.118

Even if for the sake of argument the trustfund is treated as a real economic asset capableof paying full benefits until 2075, in 2076Social Security falls off a financial cliff.According to the intermediate assumptions, in2076 alone Social Security would face a fund-ing shortfall of more than $7.5 trillion, equal toalmost 2.2 percent of GDP or 6.2 percent ofpayroll. In this case, 12 percent of all federalexpenditures would be taken up, not with pay-ing for Social Security, but simply for payingfor Social Security’s shortfall. Even under themost optimistic set of assumptions, an individ-ual born in 2009 could expect substantial short-falls in his promised retirement income.Perhaps the greatest failing of actuarial balanceas a measure of Social Security’s financinghealth is that it does not differentiate between asystem that just crawls past the 75-year “finishline” and one that can sustain solvency perma-nently. Workers born today will live well pastthe year 2075; for them, mere 75-year solvencymeans very little.119

In sum, even if the trustees underestimatefuture economic growth, there are a number ofreasons why higher growth may not fully filterthrough to Social Security. More important,increases in benefit liabilities that accompanyeconomic growth would offset much of what-ever gains did appear. Finally, even if growthsomehow eliminated Social Security’s long-term actuarial deficit, cash flow pressures inany particular year could be severe and the trustfund would do little or nothing to offset them.

The crisis deniers’ arguments fail to rebut theconclusion that unless reform is enacted, SocialSecurity’s crisis is coming and may well beeven larger than predicted.

“It’s Not Exactly the Endof the World”

The crisis deniers insist that the economy willgrow significantly faster than predicted and thatSocial Security will be saved. As we have seen,these assertions are dubious. But their argumentdoes not depend on higher growth, they say.“Even if the dismal growth forecasts turn out tobe true, and the program eventually runs adeficit,” say Baker and Weisbrot, “it’s not exact-ly the end of the world.”120 Former Clintonadministration National Economic AdvisorLaura D’Andrea Tyson makes the same case:while over the long term “a financing shortfalldevelops . . . it amounts to only 2 percent of totalpayrolls, or less than 1 percent of gross domesticproduct over the next 75 years.”121 Literallyspeaking, of course, Social Security’s problemswould not be “the end of the world.” But thatdoes not mean the fiscal strain maintainingpromised benefits would be insubstantial.

Commentators such as Baker and Weisbrotargue that Social Security is easily affordablebecause its 75-year actuarial deficit of 1 percentof GDP is similar in size to the military buildupfrom 1976 to 1985, which raised the Pentagon’sbudget from 5.2 to 6.2 percent of GDP.122 Buteven a moment’s consideration shows this anal-ogy to be false. First, the time periods beingcompared are entirely different. The 1976–85military buildup may have cost 1 percent ofGDP over 10 years, but over any 75-year peri-od it would be just a small fraction of a percent,far smaller than Social Security’s projecteddeficits. Alternately, Social Security’s shortfallas a percentage of GDP over the 10-year period2066–75 averages 2.1 percent of GDP, twice aslarge at the military buildup. Even in the 10-year period before the trust fund’s official insol-vency in 2037, Social Security faces payroll taxdeficits 70 percent larger as a percentage ofGDP than the 1970s–80s military increases.1 2 3

Moreover, some commentators interpret themilitary buildup of the 1970s and 1980s as aninvestment designed to end the Cold War. Withthe Soviet Union now defunct, defense spending

26

27

is at a postwar low of just 3 percent of GDP, gen-erating billions of dollars in savings that havebeen devoted to balancing the budget, increasingspending on nondefense programs, and reducingtaxes. Social Security’s deficits, by contrast,never end. There is no future point at which thecurrent system is expected to return to balance,no time at which extra spending devoted to thecurrent system will pay off with lower costs tothe taxpayer. Hence, the valid comparison isbetween a 10-year military spending increasefollowed by lower military spending in the futureand a spending increase twice as large for SocialSecurity that goes on forever.

Barring increased savings and market invest-ment to raise Social Security’s rate of return,there are three ways to put right the program’spending shortfalls. The first is to increase pay-roll taxes; the second, to raise income taxes;and the third, to reduce other governmentspending. While projected payroll tax deficitscould be made up through a combination of thethree, examining each in isolation gives a meas-ure of the size of the problem.

Payroll Tax Increases For Social Security to be self-financing, payroll

taxes must rise to match expenditures. Beginningin 2015, payroll taxes will no longer be sufficientto pay full benefits. From 2015 to 2037, SocialSecurity’s payroll tax deficits would be made upusing the trust fund. Realize, however, thatredeeming trust fund bonds entails income taxincreases or spending reductions equivalent to therequired payroll tax rate. In 2038, when the trustfund became exhausted, payroll taxes wouldimmediately increase from the current 12.4 per-cent to 17.9 percent of wages, and reach 19.5 per-cent in 2075. It is unclear whether and at whatpoint after 2075 tax rates would stabilize. Andthese projections assume that employment levelsare unaffected by the rising payroll tax burden.Were unemployment to rise, payroll tax revenueswould fall and deficits would reappear.

Income Tax IncreasesThe Clinton administration’s plan to issue

new debt to the Social Security trust fundwould effectively shift a substantial portion ofthe program’s funding from payroll to incomestaxes. Assume that income tax revenue willgrow at the same rate as the economy and thatSocial Security’s payroll tax deficit will be

made up through income tax increases.124 Until2015, no additional income tax revenue is need-ed to finance Social Security, since the programwill be running a payroll tax surplus untilthen.125 Following 2015, the additional incometax revenue needed would initially be small;only $11 billion in 2016, less than 1 percent ofincome tax revenue.

But the extra revenue required to maintainsolvency grows quickly. By 2025 we wouldneed an additional 12.5 percent of income taxrevenue; by 2035 income tax revenue wouldneed to increase by 17.5 percent just to paypromised benefits. This debunks the claim thatSocial Security is easily affordable through2037. By 2075, income taxes would need toincrease by one-fifth.

Spending CutsOver the past half-century federal government

tax receipts have generally varied between 17and 19 percent of GDP. Given this relatively nar-row band of government spending, 1994–1996Social Security Advisory Council memberSylvester Schieber says, “If we begin with anassumption that total government claims on theeconomy are narrowly limited and that SocialSecurity is scheduled to make a bigger claimthan currently, then some other governmentexpenditures must shrink.”126 But could wedivert additional government resources to SocialSecurity without harming other programs? Somepeople apparently think so. For instance, Bakersays that, “repayment [of trust fund bonds] willnever be very large relative to the size of theeconomy. It almost certainly will be less thangovernment spending on prisons, for exam-ple.”127 Baker is correct: repaying the bonds inthe trust fund would cost less than federal spend-ing on prisons—for about the first nine monthsof 2015, the first year Social Security runs a pay-roll tax deficit.128 By 2020, the cost of repayingthe bonds in the trust fund would be 16 timesgreater than spending on prisons, and rising.

Assume that federal spending rises at thesame rate as economic growth, thereby keepingit constant as a percentage of the total economy.If the increased funds needed for SocialSecurity will be taken out of that pool, otherspending must be cut. How much would needto be cut from other government programs tokeep Social Security solvent? In the early years,not very much. For instance, in 2016, the pay-

roll tax shortfall is just 1 percent of total gov-ernment spending. But by 2035, it would takean almost 10 percent reduction in federalspending to pay full Social Security benefitswithout increasing taxes. By 2075, federalspending would have to be cut by over 12 per-cent, just to make up Social Security’sdeficits.129 Of course, much of this governmentspending is itself unnecessary or inefficient andcould be eliminated without great harm. Butthat view of government spending is not gener-ally shared by those who deny Social Security’sproblems. In fact, in many cases it seems theydeny the Social Security crisis preciselybecause reform would curtail discretionaryspending by the federal government.

Default?Turning to payroll tax increases, income tax

increases, or spending cuts to pay full SocialSecurity benefits would entail substantial sacri-fice by taxpayers. Is it possible that instead offinding new funds to pay promised benefits, thegovernment would reduce benefits to the levelaffordable within available funds? Vincent J.Truglia, managing director of Moody’sSovereign Risk Unit, thinks so. Truglia conduct-ed a cross-country analysis of the burden thataging populations will place on industrializedcountries. All developed countries maintain highdebt ratings on their government bonds, but partof the reason is that Moody’s anticipates defaulton implicit debt promises to beneficiaries of gov-ernment entitlement programs such as SocialSecurity. If health and pension programs are notreformed, Truglia warns, ratings on explicit gov-ernment debt might have to be lowered, and “thetime horizon for any potential rating actionwould have to be sooner rather than later.” TheUnited States’ relatively low benefit levels andrelatively high birth rates place it in a far betterposition than some other developed countries,though should birth rates fall closer to Europeanlevels that circumstance could change substan-tially. Nevertheless, “Moody’s expects almostevery industrialized nation to ‘default’ on its pen-sion promises,” the United States included. TheUnited States, Truglia points out, has alreadydefaulted on Social Security benefits in changeslegislated in the 1980s and 1990s.

Payments that were previously exemptfrom income tax suddenly became, for a

large number of wealthier pensioners,taxable income. The government couldhave accomplished the same result bydecreasing benefits to those same pen-sioners, but probably chose the tax routebecause it better obscured the final out-come—lower net payments to certainpensioners. This is just one example. Thelist of pension reforms involving reduc-tion in present-day benefits, never mindfuture benefits, is long indeed.130

Benefit cuts have already been implemented, ifby the back door, and clearly could do so again.

This fear of default appears credible whenwe consider the trustees’ low-cost assumptions,whose optimism goes far beyond the increasedwage growth predicted by the crisis deniers.The trustees’ best-case scenario forecasts futurewage growth more than twice that over the past30 years; unemployment 30 percent lower; fer-tility rates 10 percent higher; immigration 50percent higher than at present; average lifeexpectancies in 2075 lower than in Japan today;and a GDP in 2075 66 percent higher thanunder the intermediate cost assumptions. Evenin this rosiest of futures, where absolutelyeverything goes Social Security’s way, the pro-gram still faces a deficit of $7.6 trillion (in 2000dollars) between 2020 and 2075.131 Assumingonly higher wage growth, as the crisis deniersdo, Social Security’s shortfall is 3.6 times larg-er.132 To advocate waiting until the 2030s totake action, in hopes that these rosy eventscome to pass, is nothing short of wishful think-ing.

Dependency RatiosAdvocates of Social Security reform often

point to declines in the ratio of workers toretirees as certain evidence for the need forreform. For instance, the Concord Coalitionpoints out that, “in 1960, there were more than5.1 workers per beneficiary. Today the ratio isthree to one. By 2030,when the boomers have allretired, there will be scarcely two workers foreach beneficiary.”133 When fewer workers mustsupport a greater number of retirees, the relativefinancial burden on each worker increases.

But many deny that a falling worker-to-retiree ratio poses any problems. For instance,Baker and Weisbrot point out that, “In 1955there were 8.6 workers per retiree, and the

28

29

decline from 8.6 to 3.3 did not precipitate anyeconomic disaster.”134 What they fail to pointout is that a worker retiring in 1955 received areturn of more than 20 percent on his payrolltax contributions, while returns for a workerretiring today fell to just around 2 percent.135 Orthat in 1955 Social Security payroll taxes werejust 4 percent of the first $4,200 in wages, asopposed to 12.4 percent of the first $76,200 inwages today.136 An economic disaster? No. Buta decline in the attractiveness and effectivenessof the program? Most certainly.

Others make a more general case, dismissingthe worker-to-retiree ratio entirely and focusingon a larger “dependency ratio,” which is theratio of workers to all dependents—childrenincluded—not simply workers to retirees.Rising life expectancies mean more retirees tosupport, but falling birth rates mean fewer chil-dren. Century Foundation president RichardLeone says:

When people sound the alarm about theaging of the Boomers, they always referto the growing “burden” on those still inthe work force. In fact, the best way tomeasure this “burden” on workers is tocompare the size of the entire dependentpopulation and the resources availableper person. One key ratio is that of youngand old dependents to workers. In 1993,it was about 70 to 100. It will rise to 83per 100 in 2030, the peak. But in 1964 itwas 96 per 100. Odd, isn’t it, that noone—including the Boomers’ parents—recalls the 1960s as an era of economicdeprivation?137

But there is a very simple reason why govern-ment finances that handled baby boomer chil-dren with relative ease will be strained by thebaby boomers as retirees: the federal govern-ment spends far, far more on older people thanon the young. Not to mention the fact that, atthe time the baby boomers were children, thefederal government spent far less on childrenthan it does today.

Federal spending averages $17,700 for eachperson aged 65 and over, versus just $2,100 perchild.138 The Congressional Budget Office stud-ied the relevance of overall dependency ratiosto future entitlement spending, concluding:

The possible relative decline in the popu-lation of children would not make up forthe costs associated with the projectedsurge in the elderly population. In con-trast, state and local governments mightwell benefit from a relative decline in thenumber of children. But any reduction inthe budgetary pressure on state and localgovernments is likely to be small com-pared with the increased pressure the fed-eral government will face.139

Even if advocates of the dependency ratio argu-ment advocated cutting all federal spending onchildren—which they most certainly donot140—it would not be enough to make up forcoming deficits in Social Security. These sim-ple facts led 1994–1996 Advisory Council onSocial Security member Sylvester Schieber andStanford economist John Shoven to concludethat “anyone who suggests that substituting achild dependent for an elderly dependent wouldhelp offset fiscal demands is either badly mis-taken or simply disingenuous.”141

Stranger still, many of those making thedependency ratio argument simultaneouslyargue that we should consider Social Security’srising costs in isolation from those of other pro-grams. Baker and Weisbrot, for instance, dis-miss efforts by entitlement reform groups suchas the Concord Coalition to “lump SocialSecurity and Medicare together,”142 whichjointly could raise payroll tax rates to over 25percent and erase all future income gains toworkers.143 Such a position is hardly defensiblewhen one is simultaneously “lumping” hypo-thetical reductions in federal children’s spend-ing with Social Security to make the wholepackage appear affordable.144

In sum, the claim that the tax increases orspending cuts needed to maintain SocialSecurity benefits would not be onerous is unre-alistic, particularly in a political environment inwhich tax rates are already considered burden-some and merely reducing the rate of growth ofan existing spending program is deemed anunacceptable “cut.” These claims greatly under-estimate the extent to which the need for addi-tional revenue would affect the tax burden onworkers or the ability of the government tomaintain other desired programs, particularlywhen Social Security will already have to com-

pete with the increasing costs of other govern-ment programs catering to the elderly.

The Stock Market vs. Pay-As-You-Go Social SecurityThere is a flip side to the crisis deniers’ argu-

ment that the trustees projections are pes-simistic. They argue that if the trustees are rightand the future economy does grow at a mere 1.7percent annually, then market investments likestocks and bonds cannot produce historicalrates of return. Therefore, Social Securityreform based on the market investment of pay-roll taxes becomes a less attractive option. Forinstance, Jeff Faux of the Economic PolicyInstitute declares:

If the projected growth rate of the econo-my declines by half, as the SocialSecurity trustees assume, the projectedreturns from the stock market must alsodecline. A stock market consistent withthe Social Security projections wouldgenerate a return of about 3.5 percent. Butstocks are highly risky.145

Should the trustees’ projections turn out to betrue and the economy go into a long-term slow-down, the crisis deniers say, returns from stocksand bonds would surely fall as well. If so, thenworkers and retirees would receive an evenlower return from market investments thanfrom Social Security.

As baby boomer retirements and low birthrates reduce labor force growth to just 0.2 per-cent annually, total economic growth willdecline as well. The critics’ argument is simple:Slower economic growth means lower corpo-rate profits, and profits drive stock prices.Under such conditions, critics like Baker argue,stocks can’t return anything like the 1929–97average of 7.2 percent. Returns any higher than3.5 percent for the S&P 500 index, Baker says,are “simply inconsistent with the SocialSecurity trustees’ growth projections.”1 4 6

Two rebuttal points are worth making at theoutset. First, the true return from a funded pen-sion system, as Martin Feldstein points out, isnot simply the return on the assets it holds butthe real, pre-tax return on nonfinancial corpo-rate capital.147 James Poterba has estimated the

return to capital in 1959–96 at 8.5 percentannually.148 Part of this return would flow to thepension system’s investments in the form of thereturn on the stocks and bonds that it holds,while the remainder flows to the government inthe form of increased corporate income taxes.Feldstein argues that this increased revenueshould be credited to the pension system, muchas income taxes levied on Social Security ben-efits are currently credited to the SocialSecurity and Medicare programs.149 Hence, thefull return to a funded system of personalaccounts would exceed the simple return on itsinvestments, even if the simple return werebelow historical averages.

Second, unless these critics assume that theentire world’s stock markets will grow at simi-lar low rates there is no reason workers couldnot simply invest their personal account contri-butions in overseas mutual funds, dozens ofwhich are available to U.S. investors at lowcost. The full returns cited by Feldstein wouldnot be obtained, since the taxes on foreign cor-porations’ income flows largely to foreign gov-ernments, but foreign investment neverthelessoffers workers the opportunity to gain higherreturns on their investments, if needed.

Let’s Assume They’re RightNevertheless, let us assume for the moment

that these critics are correct. Let us assume that,over the next 75 years, stock market returns willequal the sum of economic growth and a 2 per-cent average dividend, totaling an average ofjust 3.74 percent annually.150 Under these direconditions, surely the current Social Securitysystem would provide a higher return than mar-ket investment. Unfortunately not. Even underthis worst-case scenario market investmentwould substantially outperform Social Security.

The Social Security Administration’s DeanLeimer calculated that while Social Securityoffered above-market returns to participants dur-ing its start-up period, such as the 25 percent realannual return offered to those born in 1880 or the10 percent return to cohorts born prior to 1905,workers retiring today will receive only around 3percent returns on their payroll tax contributions.Workers born in 1960 will receive just 2 percent,while those born in 2040 are expected to receiveonly 1 percent returns.151

Why are returns from Social Security drop-pings? As the General Accounting Office states,

30

31

“in a fully mature pay-as-you-go system, long-term average implicit returns roughly equal thegrowth of total wages covered by the systembecause both contributions and benefits arebased directly on covered wages.”1 5 2 Theauthors of the GAO report are referring toNobel laureate Paul Samuelson’s formulationthat the underlying rate of return in a pay-as-you-go system like Social Security is equal togrowth of wages subject to payroll taxes pluslabor force growth.153 Based on the trustees’intermediate projections for these variables,Social Security’s return will average just 1.37percent annually over the next 75 years, 2.36percentage points lower than Baker andWeisbrot estimate for stocks (Figure 13).154

And as Baker and Weisbrot point out, evensmall differences in rates of return can make alarge difference in outcomes over the long run.For instance, a worker who invested $50 permonth for 40 years at 3.74 percent interestwould retire with 75 percent more than a simi-lar worker who invested at 1.37 percent interest.Even investing in government bonds, projectedto return 3 percent annually over the next 75years, a worker would receive a return substan-tially higher than that from Social Securitywhile owning a risk-free asset that carries no

administrative cost.Indeed, it would be strange if private invest-

ments did not outperform Social Security. Mosteconomists consider the U.S. economy to be“dynamically efficient,” which means that thereturn on capital exceeds the growth rate of theeconomy. Under these conditions, a fundedpension system investing in real economicassets should always outperform a pay-as-you-go system over the long term even if the fund-ed system invests in riskless assets like govern-ment bonds.155

In fact, the spread between Social Security’spay-as-you-go return and the return from afunded system could be even greater. On theone hand, to the degree that the funded systemraised national savings the government wouldreceive increased corporate tax revenues, whichcould be credited to the system. On the otherhand, there is substantial evidence that pay-as-you-go pension systems reduce national sav-ings, thereby cutting economic growth. ACongressional Budget Office survey, while not-ing the difficulty of precisely measuring sav-ings effects, says, “The Social Security systemmost likely has had a negative impact on privatesaving. The best empirical estimates, thoseusing cross-section data, indicate that each dol-

0

1

2

3

4

5

6

2000

2003

2006

2009

2012

2015

2018

2021

2024

2027

2030

2033

2036

2039

2042

2045

2048

2051

2054

2057

2060

2063

2066

2069

2072

2075

Stocks

Social Security

Figure 13Implicit Return from Social Security vs. Assumed Equity Returns

Sources: Social Security returns defined as sum of labor force and wage growth, derived from 2000 Trustees Report,Table II.D1. Stock returns as calculated by Baker and Weisbrot, Social Security: The Phony Crisis, pp. 90–93, definedas sum of projected GDP growth and assumed 2 percent dividend.

lar of Social Security wealth reduces otherassets by between zero and 50 cents.”156

Of course, in practice some workers earnhigher rates of return from Social Security thanothers, because of factors such as income,longevity, and marital status. While the pro-gressivity of gains from Social Security is stilldebated, even if the program’s benefits areassumed to be progressive most workers wouldreceive a lower rate of return from SocialSecurity than under Baker and Weisbrot’sworst-case scenario for stock market invest-ments. For instance, the Social SecurityAdministration calculated inflation-adjustedrates of return of single women of various earn-ing levels born in 1973. Even a low-incomewoman, who benefits from Social Security’sprogressive benefit formula, would receive areturn of just 2.8 percent annually. Moreover,the SSA’s rate-of-return calculations for indi-viduals and couples show that only single-earn-er couples, who are generally higher-income,receive a return exceeding 3.5 percent (becauseof Social Security’s benefits for nonworkingspouses). Two-earner couples, single women,and single men all receive returns of below 1.8percent.157 So even if the critics are correct andfuture stock returns are below historical aver-ages, market investment returns would still besignificantly above those from Social Security.

Critics will respond that such rate of returncomparisons are invalid because they do notinclude the so-called “transition costs” necessaryto adopt Social Security reform based on person-al retirement accounts.158 But against these transi-tion costs must be weighed the cost of the currentsystem’s unfunded liabilities, which over the next75 years exceeds $20 trillion (in 2000 dollars).Beyond the 75-year period, funding shortfallswould increase. If the present value of transition-ing to a funded system of personal accounts is lessthan that of the current system’s unfunded liabili-ties, it makes sense from a financial point of viewto make the change even if the return from per-sonal accounts will be no higher than from SocialSecurity.159

Is the Stock Market Overvalued, andDoes It Matter?

Leaving the above discussion aside, theargument made by Baker and Weisbrot and oth-ers depends heavily on the assumption that thestock market is currently overvalued. This

assumption is hardly unwarranted. After all,spectacular gains of recent years raised theprice-to-earnings ratio of the total stock marketas of January 2000 to 44, versus an average P/Eratio over the past 20 years of just 19.160 Manyanalysts believe that stock prices exhibit “rever-sion to the mean,” such that a period of above-average returns tends to be followed by a peri-od of subnormal returns.161 James Poterba andLawrence Summers found, however, thatsophisticated equity markets like those of theUnited States exhibit substantially less meanreversion than less well-developed markets.162

Nevertheless, it is in no way irrational to pre-sume that stock returns in the near future willbe lower than those of the recent past.

In response, the Technical Panel recommendedthat when the Social Security Administration pro-jects stock market returns for the future it use alower equity risk premium to reflect the currentvaluation of the market.163 The equity premium isthe additional return that investors demand on topof the risk-free rate of return to compensate themfor the additional volatility posed by stocks.Assuming the riskless return to be governmentbonds’ projected real interest rate of 3 percent, thepanel’s recommendation would imply 6 percentreal average stock returns over the long run.164

This figure is below the 1802–1997 historicalaverage of 7 percent,165 but is still substantiallyhigher than the 1.4 percent implicit rate of returnfrom Social Security.

But there are several reasons to believe thatcurrent market valuations will not lead to along-term and severe underperformance bystocks as predicted by Baker, Weisbrot, andothers. First, Baker himself argues that a “grad-ual decline in the stock market is not a veryplausible scenario.”166 But if a large market cor-rection is imminent, it will be current investorswho experience the largest losses. Future work-ers investing their payroll taxes in personalretirement accounts would experience smalllosses, if any. After all, workers with personalaccounts would have invested only a fewmonths or years worth of payroll taxes, withdecades in which to make up the losses.Workers with personal accounts would practicedollar cost averaging, purchasing more shareswhen they are inexpensive and fewer when theyare costly. Consequently, a worker starting apersonal retirement account does not invest agiven percentage of his total lifetime income in

32

33

the stock market at the market’s current price.Rather, he invests merely a portion of thatmonth’s income, which could be just a fractionof a percent of his lifetime total contribution.

Second, Baker and Weisbrot’s assumption of3.5 percent stock returns implies an equity riskpremium of just 0.5 percent over the trustees’projected government bond rate of 3.0 percent.Historically, investors in stocks have demanded apremium of approximately 7 percentage pointsover the government bond rate to compensate forthe extra risk they would be taking on. Fewinvestors would take on the extra risk associatedwith stocks without additional compensationabove the risk-free rate of return, which wouldalso point toward a near-term correction.Commentators such as James Glassman andKevin Hassett of the American EnterpriseInstitute do foresee such a low equity risk premi-um, based on Wharton School professor JeremySiegel’s finding that stocks are less risky thangovernment bonds over the long term.167 ButGlassman and Hassett conclude that a very lowequity premium is justified only at prices vastlyhigher than those today.168

Third, the track record of even marketexperts at predicting when stocks are overval-ued is not just mixed: it is poor. Burton Malkielof Princeton University cites research showingthat active stock managers have tended to moveout of stocks when prices were relatively lowand to buy when prices were high, precisely theopposite of the intended strategy.169 Moreover,Siegel shows that even if we knew beforehandthat a certain point in time would constitute amarket peak, stocks would still be the wisestinvestment over the long term (Figure 14).Siegel calculated the value of $100 invested instocks, bonds, or Treasury bills at six stockmarket peaks during the 20th century. In allcases, Siegel found that stock investors wouldhave done by far the best, ending with an aver-age of between 2.7 and 4.0 times more moneythan those who invested in bonds or bills, evenwhen purchasing what appeared to be over-priced stocks.170

In summary, even if we knew for a fact thatthe stock market is today overvalued, a wiselong-term investment strategy would stillinclude stocks in its portfolio.

510

141125

565

177

79

0

100

200

300

400

500

600

August 1929 Average After Peaks of 1901, 1906, 1915, 1929, 1937, 1966

Stocks

Bonds

T-Bills

Figure 14Value of $100 Invested at Market Peak and Held for 30 Years

August 1929 Average after Peaks of 1901, 1906, 1915,1929, 1937, 1966

Source: Jeremy J. Siegel, Stocks for the Long Run, pp. 29–31.

The Economy and the Stock MarketLeaving the current value of the stock market

aside, the basic connection between the stockmarket and the economy is not as clear-cut asmany critics claim. On a theoretical level,Baker and Weisbrot confuse the rate of returnfrom a capital investment with the rate ofgrowth of total capital profits. A simplifiedexample illustrates the difference. Imagine thateach worker invests a fixed percentage of hisincome and receives a fixed 10 percent returnon that investment. Hence, the investments’total profits for that year will equal the productof the (labor force) • (average wages) • (invest-ment rate) • 10 percent. As wages and the laborforce increase, total profits will grow, at a rateapproximating the growth rate of the economyas a whole,171 even if the savings rate and therate of return are unchanged. Of course, if thelabor force or wages growth declines then prof-it growth would also fall. Thus slower growthof profits and slower growth of GDP are con-sistent with an unchanging rate of return oncapital investments.172

A similar process takes place with wages. Totalwages equals the product of the (labor force) •(wage rate), and the rate of total wage growth isdetermined by changes in these factors. The rateof total wage growth is slated to decline, becauseof nearly nonexistent growth of the labor force.Yet, the trustees project that the wage rate willgrow faster than it has over the past three decades,as reflected in their estimates for the real wage dif-ferential. Hence, just as lower rates of total wagegrowth do not entail lower wage rates, lower ratesof profit growth do not necessarily entail lowerprofit rates on investments.

If investments are currently overpriced, theywill not necessarily produce historical rates ofreturn over the long run. But nothing in theorysays that a slower growing economy necessari-ly implies lower returns on capital investments.

Empirical study bears out these conclusions.Research by Philippe Jorion, professor offinance at the University of California-Irvine,shows that it is easier to say “slower economicgrowth equals lower stock market returns” thanto prove it. Jorion points out:

It is widely believed that the performanceof stock markets is related to economicgrowth. Indeed this relationship is routine-ly used to advocate investments in foreign

markets, in particular emerging markets,which have enjoyed fast rates of economicgrowth in the last decades. Astonishingly,there is no cross-country evidence to sup-port this link.173

Jorion’s study of 31 countries around the globe,ranging from established markets like theUnited States and United Kingdom to newereconomic powers like Japan and Germany todeveloping countries like Chile and Pakistan,indicates that slower economic growth in thefuture need not entail lower returns to stockmarket investments.

Jorion acknowledges the bookkeeping ideathat “asset prices should grow at the same rateas cash flows,” but warns that in the real world,“this relationship . . . may be blurred by a num-ber of factors.”174 Jorion’s theoretical modelshows that returns on capital investments“should be related to real GDP growth per capi-ta, instead of total GDP growth. Indeed, therecan be substantial variations in labor growthacross countries, which creates differences intotal GDP growth without necessarily affectinggrowth per capita.”1 7 5

Jorion’s empirical investigation confirms thetheory. Drawing on research on global equitymarkets conducted with Prof. Will Goetzmannof Yale,176 Jorion examined the relationshipbetween economic growth and stock returns for31 countries. While Jorion found “no observ-able relationship between stock market returnsand GDP growth,” his statistical analysisrevealed that “stock market returns are positive-ly correlated to GDP per capita growth.”177 Forinstance, developing economies grew 1.4 per-centage points faster on average than did theeconomies of developed countries, but equityreturns of developing economies averaged 2.6percentage points below those in the developedworld. How could this be? Developingeconomies expanded total GDP through rapidlabor force growth, not productivity improve-ments. In accordance with Jorion’s model, theirGDP growth per capita—and their stockreturns—lagged behind those of developedcountries. Consequently, Jorion concluded,“Lower capital gains are really associated withlower per capita economic growth,” not lowertotal economic growth.1 7 8

This link between per capita GDP growthand equity returns is relevant, since the eco-

34

35

nomic slowdown projected by Social Security’strustees stems almost entirely from reducedlabor-force growth, which Jorion found to haveno effect on equity prices. Productivity growth,which has the greatest effect on per capita GDPgrowth, will remain at the 1969–98 average of1.5 percent and per capita GDP growth will berespectable.

In summary, those who argue that historicalrates of return are no guarantee for the future arecorrect. And a highly valued market indeedposes a double-edged sword for advocates ofpersonal accounts: while rising stock prices fanpublic enthusiasm for investing, an overvaluedmarket can lead to lower returns in the nearfuture. But to assume returns over the next 75years at less than half the historical average,while simultaneously predicting a near-term cor-rection that would make historical returns possi-ble, pushes an otherwise reasonable case too far.

Of course, the real benefit from reformingSocial Security through personal accounts is notsimply higher rates of return; it is increased sav-ings, the building of wealth, and the independ-ence that comes from personal ownership andcontrol. But historically high market returnswould smooth the transition to a system of per-sonal accounts and lead to a lower tax burdenand higher retirement incomes in the future.

Conclusion

“When Federal finances are in a jam,”Steuerle, Spiro, and Carasso state, “unexpectedeconomic growth usually helps. . . . However,Social Security is unable to take advantage ofeconomic growth in the same way as other pro-grams.”1 7 9 The critics termed herein as “crisisdeniers” wish it were otherwise. They arguethat Social Security’s projected payroll taxinsolvency in 2015 and massive deficits there-after are merely the function of pessimistic eco-nomic assumptions and that higher economicgrowth will surely save the program.

However, Social Security’s demise is not, asMark Twain said of his reported death, “greatlyexaggerated.” Just the opposite may be the case.A government-appointed nonpartisan panel ofexperts concluded that the trustees’ projectionsactually underestimate the program’s deficitsby 25 percent. Moreover, even vastly increasedeconomic growth will not be enough to keep

the system solvent, because much of the gainsfrom increased payroll tax revenue are lost asbenefit entitlements increased alongside.

But even if everything the crisis deniers sayis true—that the trustees’ projections areextremely pessimistic, and that faster economicgrowth will keep the system solvent forever—Social Security reform based on personalaccounts would make sense. For even without a“crisis,” Social Security is still a lousy deal. Thebipartisan 1994–1996 Advisory Council onSocial Security estimated that even if SocialSecurity could pay full promised benefits for-ever without raising taxes by a penny, a typicalsingle worker born in 1973 would receive anannual return of just 1.7 percent.180 Personalaccounts holding only ultra-safe inflation-adjusted Treasury bonds, currently paying 3.9percent annually, would more than doubleworkers’ retirement incomes. More importantthan higher rates of return, personal accountsgive workers more security and control overtheir retirement savings, while helping thembuild wealth for themselves, their communities,and their children.

Certainly, there will be differences of opin-ion regarding reform: Should Social Securitybe defined contribution or defined benefit?Should investment be controlled by workers orby the government? And if workers are toinvest, what portion of their payroll taxesshould they control, what should they beallowed to invest in, and how should otheraspects of the program be modified?

But differences of opinion regarding theproper type of reform should not distract fromthe need for far-reaching change in the nation’spublic pension system. Social Security will notsave itself. If Social Security is to fulfill itsfounders’ goals for generations to come, diffi-cult decisions will have to be made. The princi-pal decision should be to move from a systemthat simply redistributes wealth to one that cre-ates wealth by saving and investing. The “crisisthat doesn’t exist” is alive and well, and growsmore formidable with every day that reform isdelayed.

Notes

1. President William J. Clinton, Remarks to “The GreatSocial Security Debate,” sponsored by the Concord

Coalition and AARP, Kansas City, Mo., April 7, 1998.Transcript available at <www.concordcoalition.org/entitlements/kcss0498.html>.

2. Rep. Jerrold Nadler, Constituent newsletter, 1997, Issue 2,p. 1.

3. Jane Bryant Quinn, “Should We Go Private?” Newsweek,December 7, 1999, p. 89.

4. Jane Bryant Quinn, “A Challenge, Not a Crisis,” News-week, July 3, 2000, p. 26.

5. Editorial, “Social Security: Here We Go Again,” BusinessWeek, February 1, 1999, p. 138.

6. For instance, a March 1999 Zogby International pollfound personal accounts favored by 65 percent of whites,75 percent of African Americans, and 89 percent ofHispanics. Likewise, 63 percent of Democrats, 70 percentof Independents and 75 percent of Republicans favoraccounts. Seventy-one percent of males favor personalaccounts, compared to 67 percent of females. For moreinformation on public opinion and Social Security reform,see the Cato Institute/Zogby International poll on SocialSecurity reform conducted August 1999, available at<www.socialsecurity.org>.

7. See the Cato Institute/Zogby International poll, whichfound 78 percent of respondents disagreeing with propos-als to maintain Social Security’s solvency by raising pay-roll taxes, 84 percent opposing benefit reductions, and 65percent opposing an increased retirement age; and respon-dents by a five-to-one margin preferred that any SocialSecurity funds to be invested in the market be invested byworkers rather than by the government.

8. Dean Baker and Mark Weisbrot, “Stirring Up SocialSecurity,” San Francisco Chronicle, May 21, 2000, p. 1/Z1.

9. The trustees’ latest projections are included in Boardof trustees, Federal Old-Age and Survivors Insurance andDisability Insurance Trust Funds, 2000 Annual Report(Washington: Government Printing Office, 2000), here-after referred to as the 2000 Trustees Report.

10. Dean Baker and Mark Weisbrot, Social Security: ThePhony Crisis (Chicago: University of Chicago Press, 1999).

11. Dean Baker and Mark Weisbrot, “Social SecurityScaremongering,” Washington Post, December 13,1999, p. A25.

12. Robert B. Reich, “The Sham of Saving Social Secur-ity First,” American Prospect, November 1, 1999. Avail-able at <www.prospect.org/columns/reich/rr980600. html>.

13. Christian Weller and Edie Rasell, “Getting Better All theTime, Social Security’s Ever-Improving Future,” The Eco-nomic Policy Institute, Issue Brief #140, March 30, 2000, p. 1.

14. 2030 Center, “Strengthening Social Security forYoung Workers,” undated, p. 2.

15. Jane Bryant Quinn, “A Challenge, Not a Crisis,”Newsweek, July 3, 2000, p. 26.

16. William J. Clinton, Remarks to “The Great SocialSecurity Debate,” sponsored by the Concord Coalition

and AARP, Albuquerque, New Mexico, July 27, 1998.Transcript available at <www.concordcoalition.org/ enti-tlements/abqss0798.html >.

17. Quoted in Katharine Q. Seelye, “The 2000 Campaign: TheVice President; Gore Assails Bush on Taxes and Calls RivalInexperienced,” New York Times, March 13, 2000, p. A16.

18. Quoted in David R. Francis, “Save Social Security?It’s Already Solvent,” Christian Science Monitor,September 20, 1999, p. 17. Langer argues that the “long-term” entails basing projections on GDP growth since1930. In fact, the trustees do not assume a GDP growthfigure based on past growth at all, but construct one fromprojections for changes in the components of GDP, suchas labor force growth, productivity growth, etc., based on30 years of historical data for these variables.

19. David Langer, “Cooking Social Security’s ‘Deficit’”Christian Science Monitor, January 4, 2000, p. 9.

20. Mark Weisbrot, “Social Security: Can the Trustees BeTrusted?” Knight-Ridder/Tribune Media Services, May19, 1999.

21. 2000 Trustees Report, p. 223.

22. General Accounting Office, “Social Security ActuarialProjections,” containing PricewaterhouseCoopers’ “Reporton the Actuarial Projection of the Social Security TrustFunds” (Washington: Government Printing Office, January14, 2000). Hereafter referred to as PwC Report.

23. Ibid., p. 4.

24. Ibid.

25. See 1999 Technical Panel on Assumptions andMethods, “Report to the Social Security Advisory Board,”November 1999, hereafter referred to as the 1999Technical Panel Report. The report is available at<www.ssab.org>.

26. This figure is calculated by applying the changes inassumptions the Technical Panel recommended regarding the1999 Trustees Report to the overall set of assumptions con-tained in the 2000 Trustees Report. The 2000 Trustees Reportmodified the prior year’s assumptions regarding real wagegrowth and mortality declines, though in neither case did thetrustees go as far as the Technical Panel had recommended.

27. Baker and Weisbrot, “Social Security Scaremonger-ing,” p. A25.

28. The section “If Economic Growth Exceeds Projec-tions,Will It Save Social Security?” shows why this may notalways be the case and how that would affect SocialSecurity.

29. Bureau of Labor Statistics, Series PRS85006092.

30. 2000 Trustees Report, pp. 151–52.

31. “The Great Productivity Delusion,” Grant’s InterestRate Observer, March 31, 2000, p. 1, citing research byJames Medoff and Andrew Harless. Medoff and Harlessnote that the Bureau of Labor Statistics assumes that non-production employees work an average of 37.6 hoursweekly, when many in the information technology sector

36

37

spent far more time in 1999 preparing for the Year 2000problem. Consequently, output would have been dividedby an unrealistically low number of hours worked, pro-ducing exaggerated measure of productivity.

32. Some have criticized the trustees’ use of the recent pastas the base for their future estimates, but the TechnicalPanel concurred that recent economic performance merit-ed more weight than the more distant past. The panel rec-ommended incorporating relatively old data but adjustingeach year such that more recent years carry greater statis-tical weight. Applying such geometric weighting to pro-ductivity data for the period 1951–97 would lead to aslight increase in assumed productivity, on the order of0.18 percent annually.

33. For instance, Business Week editor in chief Stephen B.Shepard declares that “two broad trends, globalization andinformation technology, are undermining the old order,forcing business to restructure. . . . The result: a radicalrestructuring that is making us more efficient. Thesetrends can combine in powerful ways to raise Americans’standard of living, create jobs, spur entrepreneurial effort—and do all this without boosting inflation. To the believ-ers in the New Economy, we have here the magic bullet—a way to return to the high-growth, low-inflation condi-tions of the 1950s and 1960s. Forget 2% real growth.We’re talking 3%, or even 4%.” Stephen B. Shepard, “TheNew Economy: What It Really Means,” Business Week,November 17, 1997, p. 38.

34. Robert J. Gordon, “Has the ‘New Economy’ Renderedthe Productivity Slowdown Obsolete?” paper presented toCongressional Budget Office’s Panel of Economic Advisors,June 14, 1999, p. 1. Available at <http://faculty-web.at.northwestern.edu/economics/gordon/334 pdf>.

35. Congressional Budget Office, “The Budget andEconomic Outlook: Fiscal Years 2001–2010,” January2000, Appendix A.

36. 1999 Technical Panel Report, p. 21.

37. Congressional Budget Office, “The Budget andEconomic Outlook: Fiscal Years 2001–2010,” AppendixA. The CBO warns that “future events may show that theacceleration was entirely cyclical and will be reversed dur-ing the next business cycle. Or the future may demonstratethat CBO has underestimated the trend, which could con-tinue at recent rates indefinitely. The uncertainty sur-rounding the projection of productivity, though neversmall, seems even larger than usual.”

38. Henry Aaron, “Trustees Reports on Social Security:The Dangers of Legislative Micro-Management ofStatistical Reports,” Testimony before the Subcommitteeon Social Security of the Committee on Ways and Means,U.S. House of Representatives, April 11, 2000.

39. Dale Jorgenson and Kevin Stiroh, “Raising theSpeed Limit: U.S. Economic Growth in the InformationAge,” Brookings Papers on Economic Activity, forthcom-ing, p. 43.

40. PwC Report, pp. 58–59.

41. For instance, Pigeon and Wray point to Europe andJapan, both with slow-growing labor forces, arguing that

it may correlate with increased productivity growth. SeeMarc-Andre Pigeon and L. Randall Wray, “DemandConstraints and Economic Growth,” The Jerome LevyEconomics Institute, Working Paper no. 269, May 1999.

42. 1999 Technical Panel Report, pp. 20–21.

43. PwC Report, p. 67.

44. Baker and Weisbrot, “Social Security Scaremonger-ing,” p. A25.

45. The Trustees Report generally refers to high, low, andintermediate projections for all variables together.However, for certain variables the trustees perform sensi-tivity analyses, which examine the outcome of changingone variable, such as wage growth, while holding all othervariables constant at their intermediate-cost levels. Thereal wage differential and actuarial balance vary on a one-to-one basis. Hence, for instance, a one percentage pointincrease in real wage growth would lead to a one percent-age point improvement in the program’s actuarial balance(2000 Trustees Report, p. 138).

46. 2000 Trustees Report, Table II.G.4.

47. Ibid., Table II.D.1.

48. In 1997, Steve Goss, Deputy Chief Actuary for theSocial Security Administration, estimated SocialSecurity’s actuarial balance for perpetuity (i.e., its surplusor deficit extending indefinitely) as a deficit of 4.7 percentof payroll. In conversation with the author, Goss estimat-ed that using assumptions contained in the 2000 TrusteesReport would produce a similar projection or perhapsslightly closer to 5 percent of payroll. Real wage growthis estimated at 1 percent in the 2000 Trustees Report;given the roughly one-to-one relationship between realwage growth and actuarial balance (see n.45), maintainingactuarial balance indefinitely would require at least 5.7percent real annual wage growth. Goss’s calculations areunpublished; available from the author on request.

49. Eugene Steuerle and John M. Bakija, Retooling SocialSecurity for the 21st Century (Washington: UrbanInstitute Press, 1994), p. 63.

50. Alan Greenspan, Testimony before the SelectCommittee on Aging, U.S. Senate, March 27, 2000.

51. 2000 Trustees Report, Table II.F.19. Figures reflect onlyOld Age and Survivors Insurance (OASI) beneficiaries.

52. 1999 Technical Panel Report, pp. 21–22.

53. Ibid., p. 21.

54. PwC Report, pp. 36–37. Fertility rates for U.S. ethnicgroups are as follows. Whites: 1.8, African-Americans:2.2, Hispanics: 3.0.

55. Population Division of the United Nations Secretariat,World Fertility Patterns 1997 (New York: United Nations,1997). Available at <www.undp.org/popin/wdtrends/fer/fer.htm>.

56. Based on sensitivity analysis of fertility rates, 2000Trustees Report, pp. 133–34.

57. 2000 Trustees Report, pp. 146–47.

58. 1999 Technical Panel Report, p. 23.

59. 2000 Trustees Report, p. 137.

60. For instance, see Ben J. Wattenberg, “The EasySolution to the Social Security Crisis,” New York TimesMagazine, June 22, 1997, p. 30; and Peter Francese,“Social Security Solution,” American Demographics,February 1992, p. 2.

61. Thorvaldur Gylfason, Principles of Economic Growth(New York: Oxford University Press, 1999), p. 21.Emphasis in original.

62. Reduced infant mortality rates increase total lifeexpectancies, but have little effect on Social Securitybecause children who die before reaching adulthood gener-ally neither pay taxes nor collect benefits from the system.

63. 2000 Trustees Report, p.136.

64. Ibid., pp. 134–36.

65. Ronald Lee and Shripad Tuljapurkar, “Death andTaxes: How Longer Life Will Affect Social Security,”Demography, February 1997. Available at <www.mvr.org>.

66. 1999 Technical Panel Report, p. 64. The UnitedStates has lower life expectancies than many of thesecountries for various reasons, including the makeup ofits population, its dietary patterns and crime rates, andso forth. But what is important is trends, not absolutelife expectancies, and the trustees project a slowingtrend for longevity increases that many analysts findunlikely.

67. Lee and Tuljapurkar, p. 8.

68. 1999 Technical Panel Report, p. 31.

69. For instance, the trustees’ low-cost scenario envisionsincreases in both fertility rates and female labor force par-ticipation. While it is possible for women to both bearmore children and work more hours, it is more likely thatthese two variables would be negatively correlated. See2000 Trustees Report, pp. 60, 148–49.

70. 1999 Technical Panel Report, p. 75.

71. See Lawrence Carter and Ronald D. Lee, “Modelingand Forecasting U.S. Mortality: Differentials in LifeExpectancy by Sex,” in Dennis Ahlburg and KennethLand, eds., Popula-tion Forecasting, a special issue of TheInternational Journal of Forecasting, November 1992, pp.393–412; and 2000 Trustees Report, Table II.D.2.

72. Lee and Tuljapurkar, pp. 67–82.

73. 1999 Technical Panel Report, p. 64. Emphasis inoriginal.

74. Ibid., pp. 8, 64–67. The trustees’ high cost assump-tion for mortality in the 1999 Trustees Report is for a 54percent decline in death rates, which would worsen actu-arial balance by 0.44 percent of payroll.

75. As discussed later, 75-year actuarial balance is inmany ways a poor measure of the program’s sustainabili-ty. Also see Neil Howe and Richard Jackson, “The Mythof the 2.2 Percent Solution,” Cato Institute Social Security

Paper no. 11, June 15, 1998.

76. Lawrence Kudlow, “What Now? . . . A Roundup ofAdvice for Bush,” National Review, April 3, 2000.

77. Aldona Robbins, “Would Prolonged Economic GrowthSave Social Security? Yes. Tax Reform Would Fuel theGrowth That Will Ease the Transition to a Pre-fundedSystem,” Insight on the News, October 25, 1999, p. 40.

78. 2000 Trustees Report, p. 191.

79. 1999 Technical Panel Report, pp. 50–52.

80. National Income and Product Accounts.

81. 2000 Trustees Report, p. 191.

82. For instance, see Neil Howe and Richard Jackson,“Have We Turned the Corner on Medicare Costs?” TheConcord Coalition, May 30, 2000. Howe and Jacksonterm projections of future Medicare cost increases as “out-landish” in their optimism regarding cost containment.

83. The ceiling on payroll taxes is adjusted each yearaccording to wage growth. Medicare’s payroll tax appliesto all wages and salary.

84. Social Security Administration, Annual StatisticalSupplement, 1999 to the Social Security Bulletin(Washington: Social Security Administration, 1999),Table 4.B1. Covered earnings are those derived from jobssubject to payroll taxes; some workers, such as thoseworking for state and local governments, are often notsubject to Social Security taxes and thus not “covered.”

85. 2000 Trustees Report, p. 191.

86. Jared Berstein, Elizabeth C. McNichol, LawrenceMishel, and Robert Zahradnik, “Pulling Apart: A State-By-State Analysis of Income Trends,” Center on Budgetand Policy Priorities/Economic Policy Institute, January2000, p. vii.

87. Isaac Shapiro and Robert Greenstein, “The WideningIncome Gulf,” Center on Budget and Policy Priorities,September 5, 1999.

88. As the following section explains, wage growth con-centrated above the payroll tax ceiling would actuallyharm Social Security’s finances by raising benefits with-out a concomitant increase in payroll tax revenues.

89. Dean Baker and Mark Weisbrot, “Social Security: APhony Crisis,” Buffalo News, March 12, 2000, p. 1H.

90. 2000 Trustees Report, Table III.C2.

91. Social Security is currently running payroll tax sur-pluses and is projected to continue doing so until 2015.

92. See Greenspan, Testimony before the Senate SelectCommittee on Aging.

93. Lewis Carroll, Alice’s Adventures in Wonderland andthrough the Looking Glass (London: Penguin Books,1998), p. 143.

94. Higher wage growth would reduce the present valueof the program’s deficits, but this is a function of the delayin insolvency and not of a reduction in the size of those

38

39

deficits. Because Social Security is a pay-as-you-go pro-gram, deficits cannot be prepaid and present value esti-mates are less meaningful.

95. Eugene Steuerle, “Are Estimates for Years to ComeMerely Science Fiction?” Tax Notes, The Urban Institute,October 25, 1999.

96. 2000 Trustees Report, p. 138.

97. Ibid., p. 208.

98. Office of Management and Budget, Budget of theUnited States Government, Fiscal Year 2000 (Washington:Government Printing Office, 2000), Analytic Perspectives,p. 337.

99. William Jefferson Clinton, Speech at GeorgetownUniversity, Washington, February 9, 1998. Available at<www.ssa.gov/history/clntstmts.html>.

100. Greenspan chaired the National Commission on SocialSecurity, informally known as the Greenspan Commission,appointed in 1981. The Commission’s final report is availableat <www.ssa.gov/history/reports/gspan.html>.

101. Alan Greenspan, Testimony before the SenateFinance Committee, February 27, 1990.

102. Carolyn Weaver, “Controlling the Risks Posed byAdvance Funding—Options for Reform,” in CarolynWeaver, ed., Social Security’s Looming Surpluses:Prospects and Implications (Washington: The AEI Press,1990), p. 171.

103. Ibid., p. 169.

104. Greenspan also argued that the fund’s contribution tonational savings, even if not offset by dissaving in the on-budget and private sectors, is more accurately calculatedby the principal rather than the interest on the fund, sincethe interest payments to the fund represent a true papertransaction between two parts of the government. Giventhat almost two-thirds of trust fund surpluses over the next15 years are attributable to interest payments rather thanpayroll tax surpluses, Greenspan concludes that “while thecontribution of Social Security to national savings is siz-able, over both the medium and long term, it is clearlymuch smaller than the conventional calculations suggest.”Alan Greenspan, testimony before the Senate FinanceCommittee, February 27, 1990.

105. James M. Buchanan, “The Budgetary Politics ofSocial Security,” in Weaver, ed., pp. 45–56.

106. The Bureau of the Public Debt distinguishes betweendebt held by the public and debt held by the government,such as government bonds held in the Social Security trustfund. Currently, around 39 percent of total governmentdebt is held by the government. Public discussion, howev-er, focuses almost exclusively on debt held by the public,ignoring government-held debt. For instance, a Gore cam-paign press release says the Vice President “will invite theAmerican people to join in a bold plan to completely elim-inate the national debt” (“Securing America’s Future:Gore’s Plan for Expanding Prosperity for America’sFamilies,” Gore Campaign press release, June 13, 2000).Under Vice President Gore’s proposal, however, total gov-ernment debt will increase as reductions in publicly held

debt are more than matched by increases in debt held bythe government.

107. Milton Friedman, Tax Limitation, Inflation and the Roleof Government (Dallas: The Fisher Institute, 1978), p. 5.

108. W. Mark Crain and Michael L. Marlow, “The CausalRelationship between Social Security and the FederalBudget,” in Weaver, ed., p. 129.

109. John Cogan, “The Congressional Response to SocialSecurity Surpluses, 1935–1994,” Hoover Institution,Essays in Public Policy, August 1998.

110. “Al Gore Proposes New Reforms to ModernizeSocial Security and Strengthen It for the Future,” cam-paign publication, available at <www.algore2000.com/agenda/agenda_social_security_reform.html>.

111. Report of the 1989–1991 Advisory Council on SocialSecurity (Washington: Government Printing Office,March 1991), pp. 172–73.

112. CBO Director Dan L. Crippen and Deputy DirectorBarry B. Anderson, Testimony before the House Ways andMeans Committee, February 23, 1999.

113. Testimony of Henry J. Aaron before the SenateCommittee on the Budget, January 19, 1999.

114. That is not to imply that the obligations created by trustfund financing are invalid; they should be honored. It is sim-ply to say that we should not assume that the trust fundmechanism made it easier to meet those commitments.

115. This double-counting is easy to observe when oneconsiders that under the administration plan debt to thetrust funds increases more quickly than debt to the publicis reduced. Hence, gross government debt (publicly helddebt plus debt in government trust funds) actually increas-es under what the administration has touted as a debtreduction plan.

116. David M. Walker, Comptroller General of the UnitedStates, “Social Security: What the President’s ProposalDoes and Does Not Do” (Washington: GeneralAccounting Office, February 9, 1999), p. 10.

117. The so-called wealth effect proposes that increases inthe value of stock holdings boost consumer spending,while Ricardian equivalence, proposed by Robert Barro,argues that changes in government’s net savings are coun-tered by opposing movements in private savings, effec-tively unraveling the savings effects of government fiscalpolicy. See Martha Starr-McCluer, “Stock Market Wealthand Consumer Spending,” Federal Reserve Board ofGovernors, April 1998; R. J. Barro, “Are GovernmentBonds Net Wealth?,” Journal of Political Economy,November/December 1974, pp. 1095–1118. MartinFeldstein, however, argues that since most low-incomehouseholds have such low levels of savings that liquidityconstraints prevent them from consuming more, dissavingas a response to personal account balances would be rela-tively light. See Martin Feldstein, Testimony before theSenate Finance Committee, March 16, 1999.

118. On savings offsets for Individual RetirementAccounts, see Orazio P. Attanasio and Thomas C.DeLeire, “IRAs and Household Saving Revisited: Some

New Evidence,” National Bureau of Economic Research,October 1994.

119. See Neil Howe and Richard Jackson, “The Myth ofthe 2.2 Percent Solution,” Cato Social Security Paper no.11, June 15, 1998.

120. Baker and Weisbrot, “Social Security: A PhonyCrisis,” p. A25.

121. Laura D’Andrea Tyson, “Social Security Is WorkingJust Fine, Thank You,” Business Week, June 26, 2000, p. 32.

122. Baker and Weisbrot, “Social Security: A PhonyCrisis.” Data on military spending from Economic Reportof the President 2000, Table B-77.

123. 2000 Trustees Report, Table II.C.1.

124. The Office of Management and Budget estimatesincome tax revenues for 2000 at $952 billion. Revenuesgrowing at the same rate projected by the trustees for theeconomy would reach $2.06 trillion in 2015, $5.43 trillionin 2035, and $36.97 trillion in 2075. Social Security’s pay-roll tax deficits in those years will be $11 billion, $951 bil-lion, and $7.36 trillion, respectively. Source: Office ofManagement and Budget, Budget of the United States,Fiscal Year 2001, Historical Tables, Table 2.1; 2000Trustees Report, Table III.B4.

125. In addition to payroll taxes, Social Security alsoreceives a portion of income taxes collected on benefits.These are a relatively small amount and are ignored forthese purposes.

126. Sylvester Schieber, “The need for Social Security reformand the implications of funding benefits through personalsecurity accounts,” Benefits Quarterly, Third Quarter1997, pp. 29–39. Schieber calculates the federal government’sshare of GDP based upon three-year averages in order tosmooth the effects of the business cycle on tax receipts.

127. Dean Baker, “Say What?” In These Times, May 30,1999, p. 2.

128. The federal government spent $3.22 billion on pris-ons in 1999, 0.23 percent of current federal spending.Social Security’s payroll tax deficit in 2015 will beapproximately 0.29 percent of federal spending, assumingthat spending rises at the same rate as economic growth.We could include state spending on prisons as well,though Baker argues that Social Security should beviewed in isolation from other programs (see Baker andWeisbrot, Social Security: The Phony Crisis, pp. 4–6).Including state prison expenditures, it would be 2017 or2018 before trust fund bond repayments exceeded gov-ernment spending on prisons. Bureau of Justice Statistics,U.S. Department of Justice.

129. The Office of Management and Budget estimatestotal federal spending for 2000 at $1.79 trillion. Spendinggrowing at the same rate projected by the trustees for theeconomy would reach $3.78 trillion in 2015, $9.69 trillionin 2035, and $61.99 trillion in 2075. Social Security’s pay-roll tax deficits in those years will be $11 billion, $951 bil-lion, and $7.36 trillion, respectively. Office ofManagement and Budget, Budget of the United States,Fiscal Year 2001, Historical Tables, Table 3.1; and 2000

Trustees Report, Table III.B4.

130. Vincent J. Truglia, “Can Industrialized CountriesAfford Their Pension Systems?” Washington Quarterly,Summer 2000, p. 201. Emphasis added.

131. See 2000 Trustees Report, Tables I.E.1, II.D.1,II.D.2, III.B.1, and III.B.2.

132. The trustees’ low-cost scenario projects a 75-yearactuarial deficit of 0.38 percent of taxable payroll (TableI.G.2); the sensitivity analysis combining the low-costassumption for wage growth with the intermediate-costassumptions for all other variables produces an actuarialdeficit of 1.38 percent of payroll (Table II.G.4).

133. The Concord Coalition, “Federal Budget Primer, PartII: Demographics Is Destiny,” available at <www.ssa.gov/history/clntstmts.html>.

134. Baker and Weisbrot, Social Security: The PhonyCrisis, p. 32.

135. See Dean R. Leimer, “Cohort-Specific Measures ofLifetime Net Social Security Transfers,” Social SecurityAdministration, Office of Research and Statistics,Working Paper no. 59, February 1994.

136. 2000 Trustees Report, Table II.B1.

137. Richard Leone, “Don’t Worry, Generation X; Whythe Demographic Nightmare of the TwentysomethingsIsn’t Likely to Come True,” Washington Post, April 30,1996, p. A13.

138. Congressional Budget Office, “Letter to theHonorable John R. Kasich regarding federal spending onthe elderly and children,” July 27, 2000.

139. Congressional Budget Office, Long-Term BudgetaryPressures and Policy Options, May 1998, p. 6.

140. For instance, the Century Foundation supportsincreased government payments to low-income singleparents raising children. Baker argues that tens of billionsof dollars in federal budget surpluses should be devoted togreater investments in education, among other things. SeeDean Baker, “The Great Surplus Debate,” The AmericanProspect, May–June 1998, p. 80.

141. Sylvester Schieber and John Shoven, The Real Deal:The History and Future of Social Security (New Haven:Yale University Press, 1999), p. 251.

142. Baker and Weisbrot, Social Security: The PhonyCrisis, p. 4.

143. See, for example, Neil Howe and Richard Jackson,“The Graying of the Welfare State,” National TaxpayersUnion Foundation, Policy Paper 117, August 30, 1999.Howe and Jackson are analysts for the Concord Coalition.

144. See Baker and Weisbrot, Social Security: The PhonyCrisis, p. 32, where the authors declare that “the increasein the future burden of caring for a larger elderly popula-tion will be offset to a large extent by the reduced costs ofeducation, child care, and other expense of caring fordependent children.”

145. Jeff Faux, “Making Social Security Work,” statement by

40

41

Jeff Faux, President, Economic Policy Institute, to the WhiteHouse Conference on Social Security, December 8, 1998.

146. Dean Baker, “Saving Social Security With Stocks:The Promises Don’t Add Up,” The CenturyFoundation/Economic Policy Institute, 1997. See also,Baker and Weisbrot, Social Security: The Phony Crisis,pp. 90–104. It is not just advocates of reform through cre-ation of personal accounts who rely on historical stockreturns when projecting returns for the future. The1994–1996 Advisory Council on Social Security madesimilar projections (see Report of the 1994–1996 AdvisoryCouncil on Social Security, Volume I: Findings andRecommendations [Washington: Government PrintingOffice, January 1997], p. 170), and even Vice PresidentGore’s campaign refers to the historical 5.3 percent realreturn on a mixed stock/bond fund as “conservative”(Gore campaign press release, “Gore Offers ‘RetirementSavings Plus,’” June 20, 2000).

147. See, for instance, Martin Feldstein, “PrivatizingSocial Security: the $10 Trillion Opportunity,” Cato SocialSecurity Paper no. 7, January 31, 1997.

148. James Poterba, “The Rate of Return to CorporateCapital and Factor Shares: New Estimates Using RevisedNational Income Accounts and Capital Stock Data,” NationalBureau of Economic Research, April 1999, pp. 9–10. Thestandard deviation of returns on capital was 1.0 percent.

149. See Martin Feldstein and Andrew Samwick,“Allocating Payroll Tax Revenue to Personal RetirementAccounts to Maintain Social Security Benefits and thePayroll Tax Rate,” National Bureau of EconomicResearch, June 2000. Corporate tax revenues can be cred-ited to Social Security only insofar as they are generatedby increased investment. Feldstein and Samwick assumethat three-quarters of the deposits to personal accountswill constitute new savings; the remainder will be offsetby dissavings elsewhere. See also n. 114.

150. See Baker and Weisbrot, Social Security: The PhonyCrisis, pp. 91–93. Baker and Weisbrot’s calculation isbased on the 1999 Trustees Report’s projected 1.5 percentannual growth of GDP. Average GDP growth in the 2000Trustees Report is estimated at 1.74 percent, an increasedue to changes in the calculation of the consumer priceindex.

151. See Leimer, “Cohort-Specific Measures of LifetimeNet Social Security Transfers.”

152. General Accounting Office, “Social Security: Issues inComparing Rates of Return with Market Investment”(Washington: General Accounting Office, August 1999), p. 25.

153. See Paul A. Samuelson, “An Exact Consumption-Loan Model of Interest with or without the Contrivance ofMoney,” The Journal of Political Economy, December1958, pp. 467–82.

154. Baker and Weisbrot, Social Security: The PhonyCrisis, p. 93.

155. On whether the U.S. economy is dynamically effi-cient, see Andrew B. Abel et al., “Assessing Dynamic Effi-ciency: Theory and Evidence,” Review of EconomicStudies, January 1989, pp. 1–20. In a dynamically ineffi-

cient economy, where the capital stock is so large that thereturn on capital is below the economic growth rate, allindividuals could be better off over the long run by con-suming a portion of that capital today (for instance, byestablishing a pay-as-you-go pension system) rather thanby saving. For a comparison of funded and pay-as-you-gosystems, see “Reinventing Social Security for the NewEconomy,” by Thomas F. Siems of the Federal ReserveBank of Dallas, forthcoming Cato Institute SocialSecurity Paper.

156. Congressional Budget Office, “Social Security andPrivate Saving: A Review of the Empirical Evidence,”July 1998, p. 30. “Social Security wealth” is the presentvalue of the benefits to which a worker is currently enti-tled. Social Security wealth is not built upon real savings,but workers treat it as de facto wealth and reduce real sav-ings in response to an increase in their Social Securitywealth. Twelve of 14 cross-sectional studies surveyedconcluded that Social Security reduced savings, thoughmost found that it did so on less than a proportional basis(i.e., a one-dollar increase in Social Security wealth led toless than one dollar’s decrease in real savings). The CBOalso analyzed time-series and cross-country surveys, butfound them inconclusive and inapplicable to the currentcase.

157. General Accounting Office, “Social Security: Issuesin Comparing Rates of Return with Market Investments”(Washington: General Accounting Office, August 1999),pp. 27–29. Rates of return for other income levels are asfollows: average earnings, 1.8 percent; high earnings, 1.2percent; maximum taxable earnings, 0.4 percent.

158. See, for instance, Alicia H. Munnell, “ReformingSocial Security: The Case Against Individual Accounts,”National Tax Journal, December 1, 1999, p. 803.

159. On transition costs, see William Shipman, “Factsand Fantasies about Transition Costs,” Cato SocialSecurity Paper No. 13, October 13, 1998; MiltonFriedman, “Speaking the Truth about Social SecurityReform,” Cato Briefing Paper no. 46, April 12, 1999.

160. Data compiled by Robert Shiller for his book, IrrationalExuberance (Princeton, N.J.: Princeton University Press,2000). Data available at <www.econ.yale.edu/~shiller/>.

161. See Jeremy Siegel, Stocks for the Long Run (NewYork: McGraw-Hill, 1998), Chapter 1.

162. Poterba and Summers found evidence of mean rever-sion in U.S. stock returns during the 1926–85 period, supple-mented by evidence from prior to 1926 and from 17 foreigncountries. However, the authors noted “a clear tendency formore mean reversion in less broad-based and sophisticatedequity markets.” Hence, stock returns prior to 1926 and thosefrom countries with less well-developed capital marketsexhibit greater mean reversion than in the U.S. at present. SeeJames M. Poterba and Lawrence H. Summers, “MeanReversion in Stock Prices: Evidence and Implication,”National Bureau of Economic Research, August 1987.

163. The SSA does not project stock or corporate bondreturns as part of the annual Trustees Report, but doesincorporate projected returns when constructing cost esti-mates of legislation incorporating market investment.

164. Whether and when any possible market correctiontook place could greatly affect the distribution of thosereturns over time.

165. See Siegel, p. 14.

166. Dean Baker, “Double Bubble: The Implications ofthe Over-Valuation of the Stock Market and the Dollar,”Center for Economic Policy and Research, May 31, 2000.

167. See Siegel, Chapter 1.

168. See James Glassman and Kevin Hassett, Dow36,000 (New York: Times Books, 1999).

169. See, for instance, Burton G. Malkiel, A RandomWalk Down Wall Street, 6th ed. (New York: W. W. Norton& Co., 1996), pp. 186–88.

170. Siegel, p. 31.

171. The reason: economic growth is a function of thenumber of workers and output per worker.

172. Over time it would be expected that slower laborforce growth would lead to a somewhat lower return oncapital, as capital became relatively more abundant andlabor less abundant. But even over the long term such aneffect would be relatively minor. For instance, Feldsteinand Samwick estimate that even with a system of person-al accounts raising the capital stock, after 70 years the pre-tax return on capital would fall only from 8.5 percent to6.9 percent, though this would be balanced by higherincomes to labor. See Martin Feldstein and AndrewSamwick, “Maintaining Social Security Benefits and TaxRates through Personal Retirement Accounts,” NationalBureau of Economic Research, Cambridge, Mass., March1, 1999, p. 9, n. 16.

173. Philippe Jorion, “Global Stock Markets andEconomic Growth,” paper presented at “The EquityPremium and Stock Market Valuations” conference heldat the Anderson School of Management, University ofCalifornia, Los Angeles, April 30, 1999, p. 1.

174. Ibid., p. 5.

175. Ibid., p. 7. Jorion’s model assumes the workforce andthe population to be identical, which in the real world isclearly untrue. Mathematically speaking, Jorion’s modelsays that capital returns should vary with changes in out-put per worker, not GDP per capita. Applying this distinc-tion to the current discussion would strengthen the con-clusions drawn in this paper.

176. See Philippe Jorion and William N. Goetzmann,“Global Stock Markets in the Twentieth Century,” Journalof Finance, June 1999. Available at <www.gsm.uci.edu/~jorion/>.

177. Jorion, p. 12. Emphasis added.

178. Ibid., p. 9. Emphasis added.

179. Eugene Steuerle, Christopher Spiro, and Adam Carasso,“Squandered Opportunity? Increasing Social SecurityLiabilities in Flush Times,” The Urban Institute, April 15, 2000.

180. Report of the 1994–1996 Advisory Council onSocial Security, Volume I: Findings and Recommen-dations, Appendix II, p. 219–20. A single male withaverage earnings is projected to earn a 1.48 percentreturn; a single female, 1.94 percent. Both assumptionsassume that tax rates and benefit levels continue accord-ing to current law. If taxes are increased as need to paypromised benefits, annual returns for these individualswould fall to an average of 1.5 percent.

42