Routing

The Pension Benefit Guaranty Corporation,the federal agency that insures private-sectordefined-benefit pension plans, had a surplus of$9.7 billion at the end of 2000 but a deficit of$11.2 billion at the end of 2003. Pension planunderfunding stands at more than $350 billion,which increases the likelihood that more pensionplans will go under and taxpayers will eventuallybe called upon to provide a bailout.

The reasons for the PBGC’s financial difficul-ties can be found in the structure of defined-ben-efit pension plans and in the way Congress set upthe premium rules when it created the programin 1974. First, because the PBGC stands as theultimate guarantor of companies’ pension liabil-ities, plan sponsors have an incentive to investtheir assets in equities rather than fixed-incomesecurities of the same duration as the liabilities.Second, funding rules allow companies to makegradual contributions to their pension plans inthe event of underfunding, which guaranteeslong-term exposure for the PBGC. Furthermore,when faced with higher contributions, compa-nies have usually appealed to Congress to reducethe underfunding that they need to report,which reduces contribution requirements.

Unfortunately, Congress has failed to ade-

quately address the problems of the PBGC. Intemporary legislation passed in April 2004,Congress reduced the required contributionscompanies must make to their defined-benefitpension plans by an estimated $80 billion overtwo years by changing the formula used to calcu-late pension liabilities. Congress also providedadditional relief of approximately $1.6 billion tosteel and airline companies with heavily under-funded pension plans.

Rather than place the PBGC on sounderfinancial footing, those measures will likelyworsen the agency’s financial condition. For thatreason—and to reduce the likelihood that tax-payers will have to bail out the PBGC—whenCongress revisits this issue in two years’ time, itshould adopt legislation that contains the fol-lowing provisions: First, it should enforce thecurrent premium rules so that companies do notavoid paying premiums to the PBGC if theirplans are underfunded. Second, it should modi-fy those rules so that premiums are truly riskbased. Finally, Congress should allow pensioninsurance to be offered independent of thePBGC in the private sector or through a self-insurance pool whose members would be jointlyresponsible for any deficits their plans created.

How to Reduce the Cost of Federal Pension Insurance

by Richard A. Ippolito

_____________________________________________________________________________________________________

Richard A. Ippolito (richardippolito@msn.com) is the former chief economist at the Pension Benefit GuarantyCorporation.

Executive Summary

No. 523 August 24, 2004

Introduction

In 1974 Congress established the PensionBenefit Guaranty Corporation as part of theEmployee Retirement Income Security Act.The PBGC’s task is to insure defined-benefitpension plans that are underfunded at thetime of the plan sponsor’s bankruptcy (seeAppendix for definitions of pension terms).Some analysts have argued that theStudebaker bankruptcy in 1964 was the rootof the PBGC insurance program, though itsenactment was stalled until the economicdownturn of 1973–74.1

Because Congress had little idea of howmuch the insurance would cost, it set a pre-mium equal to $1 per participant. Had thatpremium been assessed in 2003 (adjusted forinflation), it would have generated about$150 million. Because of various increasesalong the way, premiums in 2003 actuallywere almost $1 billion. Yet in 2003 the PBGCreported a deficit of $11.2 billion, a number75 times higher than the originally anticipat-ed annual premium level. Moreover, undercurrent conditions, the possibility exists forthe deficit to get dramatically higher, inwhich case taxpayers will almost certainly becalled upon to bail out the system.

This paper illustrates the nature of theinsurance provided, explains the factors thatmake it expensive, and demonstrates its poten-tial to suffer catastrophic losses. Legislationrecently passed by Congress (H.R. 3108) willmake matters even worse by offering fundingrelief largely to single-employer pension plansand additional relief to airlines and steel com-panies. That need not be the case, as pensioninsurance can be altered to eliminate most ofthe variance in financial outcome and toreduce the incentives companies have toengage in moral hazard behavior against thePBGC. That result can be brought about bytransforming the federal insurance programinto a true self-insurance pool.

This paper focuses on the major programoperated by the PBGC, namely, the single-employer program, which is comprised ofcompanies that offer defined-benefit pension

plans to their employees.2 Although thePBGC is involved in an “endgame” in thesense that virtually no new defined-benefitplans have been established in more than 10years and those plans are clearly losing mar-ket share to defined-contribution 401(k)plans, an appreciation of the PBGC’s prob-lems is germane for two reasons: First, thedefined-benefit plans that remain are dispro-portionally union plans.3 Those plans, typi-cally underfunded, are responsible for thelion’s share of the underfunding and thePBGC’s claims experience. Second, althoughthe PBGC covers only defined-benefit plans,the Enron debacle of 2001, in which manyemployees had their 401(k) balances erasedbecause they overinvested in company stock,could be the Studebaker of 401(k) planinsurance. In other words, pressure couldmount on Congress to establish a PBGC fordefined-contribution pension plans. But thedefined-benefit experience speaks loudlyabout the likely costs of pursuing such aninsurance program.

The Economics of Defined-Benefit Pensions

In a defined-benefit plan, the employermakes a promise to pay workers a benefit atretirement age. A typical plan might pay anannuity starting at retirement equal to 1.5percent times years of service times finalwage. Thus the indexing of the pension tothe final wage makes it important for work-ers to stay with the company. Let’s examinewhy.

The Cost of Quitting If a worker quits a company before becom-

ing eligible for retirement, his wage for thepurposes of the pension he receives from thatcompany is frozen at whatever level it is at thetime of his departure. Although he can earnfuture pension service credits with anothercompany, his pension with his old companywill be proportional to the wage he earned atthe time he left, not the wage he would have

2

In 2003 the PBGCreported a deficit

of $11.2 billion, a number 75

times higher thanthe originally

anticipated annual premium

level.

earned at retirement had he stayed at thatcompany.

For example, consider a 40-year-old employ-ee with 20 years of service who earns $40,000.Suppose that retirement age is 60 and his pen-sion pays 1.5 percent per year of service timesfinal salary. Over the next 20 years, he expectshis salary to increase with inflation plus somereal factor. If those amounts together areexpected to be 6 percent per year, his expectedwage at retirement is $128,285.4 His expectedpension annuity, based on the 20 years of serviceaccumulated to date, would be $38,485.5 The pre-sent value of this annuity is his “ongoing” pen-sion benefit.

If, on the other hand, the worker quitsnow, his final salary is $40,000. Thus hisannuity starting at age 60 based on the same20 years of service is $12,000.6 The presentvalue of that amount is his “termination”pension benefit. The difference between hisongoing and termination pension benefit isthe “pension capital loss” from quitting. Theworker in this example has a strong incentiveto stay, because if he quits he loses about two-thirds of his pension benefits.

Calculation of the present value of suchpension benefits and losses is straightfor-

ward. Figure 1 shows ongoing (solid lineschedule) and termination (bowed schedule)lifetime benefits as a percentage of currentannual wage for workers at every tenure level.The figure assumes that every worker startswork with the company at age 30, plans toretire at age 60, and has a 20-year lifeexpectancy after retirement. It also assumesthat the interest rate and wage growthtogether are 6 percent. The vertical differencebetween these two schedules shows the pen-sion loss from quitting as a percentage ofcurrent wage at various levels of service.

Absolute dollar losses are relatively smallat short tenure because workers have notaccumulated much service. Those lossesbecome large midcareer as workers accumu-late more service but still earn a wage that issubstantially lower than their anticipatedwage at retirement. As workers approachretirement, the wage converges to the retire-ment wage, which reduces the size of the cap-ital losses for the worker.

Pension Consequences of Bankruptcyand Termination

Workers as a group have a stake in thefinancial success of a firm. If a firm encoun-

3

A worker in midcareer has astrong incentiveto stay, because ifhe quits he losesabout two-thirdsof his pensionbenefits.

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31

Figure 1Ongoing vs. Termination Benefits

Years of Service

Lif

etim

e Pe

nsio

n V

alue

as

Perc

ent o

fA

nnua

l Wag

e

Pension Consequences of Bankruptcy andTermination

A worker in midcareer has astrong incentive tostay, because if hequits he losesabout two-thirdsof his pension ben-efits.

0

100

200

300

400

500

600

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31

Ongoing Benefits

Contingent Benefits

Insured Benefits

C

I

Figure 1Ongoing vs. Termination Benefits

ters financial difficulty, it sometimes termi-nates its pension plan, an outcome that iseven more likely in the event of bankruptcy.7

Upon a termination, the employees are enti-tled to their pensions, but, for the purposesof calculating their pension benefits, they arelocked into the wage they earned on the dayof termination. In effect, workers’ termina-tion benefits are the same as those describedby the bowed schedule in Figure 1.

Presumably, if a firm is successful, theplan will not terminate, and workers willreceive the full value of their ongoing pen-sion benefits. If a firm encounters sufficientfinancial stress, however, it may terminatethe plan and pay workers their terminationbenefits. We can think of the differencebetween ongoing and termination liabilitiesas “contingent benefits.”

For illustration, suppose that the plandepicted in Figure 1 has one worker at eachservice level. Upon a termination, workersabsorb the losses denoted by area C. Thoseare called contingent benefits because theyare payable only if the firm survives and theplan is not otherwise terminated. If a plansponsor has sufficient assets in the pensionplan to cover termination benefits—that is,the amount denoted by area I—the plan isfully funded and there is no claim against thePBGC. If the bankrupt sponsor has assetsthat are less than that amount, the PBGCmakes up the difference. The PBGC can tryto recover those losses in bankruptcy court,but typically the agency recovers less than 10cents on the dollar. (Although insured bene-fits are sometimes a bit less than terminationbenefits, the two concepts are so closely relat-ed that we can safely ignore the differencesfor our purposes.)

Factors That Affect PBGC Losses

Three factors explain the underfundingexposure of the PBGC. First, some unionplans, by construct, are persistently under-funded even in the best of economic condi-

tions. Second, the insurance program guar-antees nominal (not real) pension benefits,which means that the PBGC’s exposure issensitive to nominal interest rates. And third,although plan sponsors can eliminate theeffects of interest rate volatility by holdingfixed income securities, they typically do not.Instead, they hold large amounts of equitysecurities, which ensures dramatic increasesin underfunding in economic downturns,precisely when bankruptcy rates are highest. Iconsider each of those factors in turn.

Persistent UnderfundingMany plans (notably those covered by col-

lective bargaining agreements) are typicallyunderfunded for insured benefits (planassets are less than the amount depicted byarea I). That is because they are “flat-benefit”plans that, instead of promising a pensionproportional to salary level and years of ser-vice, promise a fixed dollar amount per yearof service.8 Even if the sponsor increases theflat benefit as a result of periodic adjust-ments to the collective bargaining contract,the funding rules enforced by the InternalRevenue Service do not permit the sponsorto assume that benefits will increase. Eachtime benefits are increased, “new” under-funding is recognized and paid for graduallyover a period of time.

By contrast, in salary-related plans, pro-jected benefits automatically increase withwages, and the funding rules permit spon-sors to assume wage growth. Those plansoften have sufficient assets to cover the ter-mination benefits denoted by area I in Figure1, plus some cushion to cover some portionof contingent benefits (part of area C). Firmsand union representatives could agree totransform a flat-benefit plan to a salary-relat-ed plan and thereby ensure better funding,but they mostly choose not to, which createsa large exposure to the insurer—that is, to thePBGC.

Interest Rate EffectsThe value of the insurance (and the

amount of PBGC exposure) also depends on

4

The PBGC’s exposure is sensitive to

nominal interestrates.

the interest rate. When Congress enacted thePBGC insurance program, it could have spec-ified that the insurance cover a fixed percent-age of ongoing benefits (i.e., real benefits).Instead, it set insurable benefits to some-thing closely related to termination benefits.That means that in periods during which theinterest rate is very high, the present value ofthe pension is relatively low. On the otherhand, if interest rates become very low, thepresent value of insured benefits is prettyclose to that of ongoing benefits.

The reason for that seeming anomaly iseasy to understand intuitively. If nominalinterest rates are very high, that almost cer-tainly means that anticipated inflation ishigh. When inflation is high, $100 in benefitsreceived 10 years in the future is not worthmuch in today’s dollars. By contrast, whennominal interest rates are very low, anticipat-ed inflation is also very low. In that case, thereal value today of $100 received 10 yearshence is something reasonably close to $100.

Figure 2 gives a visual representation ofthe importance of the interest rate in theinsurance. The bowed schedule with opensquare markers depicts the baseline lifetimeinsured benefits based on a 6 percent interestrate (this schedule is taken directly from

Figure 1). The lowest “more bowed” scheduleshows termination benefits when the interestrate is 12 percent, and the “less bowed”schedule shows those benefits when theinterest rate is 2 percent. The lower the inter-est rate, the higher the level of insured bene-fits as a portion of ongoing benefits.

Notice the importance of a reduction in theinterest rate from 6 to 2 percent. Suppose thatthe plan is exactly fully funded for terminationbenefits at the 6 percent interest rate. Whenthe interest rate falls, covered workers havemore valuable insurance, because the insuredamount is much closer to ongoing benefits.Area A measures the additional insured bene-fits. If the plan was underfunding (or exactlyfully funded) prior to the change in interestrate, underfunding also increases by theamount denoted by area A.

In the short to medium term, the PBGCabsorbs entirely the additional underfundingas added exposure. But the sudden increasein underfunding triggers Internal RevenueService funding requirements that force plansponsors to begin making up the additionalunderfunding. Special “Deficit ReductionContribution” rules are in place to makefunding requirements more urgent in timesduring which sponsors become seriously

5

The lower theinterest rate, thehigher the levelof insured benefits as a portion of ongoing benefits.

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30

Figure 2How Interest Rates Affect Insurable Benefits

Lif

etim

e Pe

nsio

n V

alue

as

Perc

ent o

f Ann

ual W

age

Years of Service

The lower theinterest rate, thehigher the level ofinsured benefits as a portion of ongoing benefits.

0

100

200

300

400

500

600

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30

2% Interest Rate

6% Interest Rate

12% Interest Rate

Figure 2How Interest Rates Affect Insurable Benefits

A

underfunded for termination benefits. Thathelps protect the PBGC but creates dramaticvolatility in contribution requirements. Asshown below, when funding rules becomeparticularly binding, pension plan sponsorsare adept at getting Congress to relax therules, as they did in the recently passed legis-lation that allows them to evade the worstimpact of the rules and to offload more expo-sure on the PBGC.

Asset Composition and the Value ofPBGC Insurance

Finally, the value of pension insurancealso depends on the composition of the pen-sion plan’s portfolio. At any given fundingratio, a portfolio comprised of stocks pre-sents more risks to the PBGC than a portfo-lio of Treasury bonds.

A plan sponsor can protect itself againstinterest rate fluctuations by fully funding itspension fund with bonds of the same dura-tion as its liabilities. Duration is a term of artthat measures the percentage change in thevalue of a debt security as a result of a onepercentage point change in the interest rate.If a firm’s pension liability has a duration of10, then that liability increases by 10 percentin response to a reduction in the interest rateof one percentage point.

Consider a simple example. The interestrate is 5 percent. A plan promises to pay anannuity equal to $100 per annum in perpe-tuity, starting immediately. Akin to thenotion of termination benefits, I assume thatthe $100 per annum obligation is fixed innominal terms. The present value of a $100annuity paid out each year in perpetuity isequal to the annuity amount, $100, dividedby the interest rate, in this case 5 percent. Ifthe plan sponsor purchases a $2,000 bondwith no maturity date and a coupon rate of 5percent, then the bond delivers $100 in inter-est payments per annum forever, which isjust enough to pay the $100 pensionpromised per annum forever. If the interestrate falls to 4 percent, then the present valueof the annuity is $2,500 (= $100/.04). In thiscase, the value of the liability changes by

$500, or 25 percent of its value, as a result ofa 1 percent reduction in the interest rate. Thisliability has a duration of 25.

The present value of the promised pensionis higher at the lower interest rate. If the pen-sion sponsor tried to sell the obligation to paythe annuity to an insurance company, thesponsor would have to hand over, not $2,000,but $2,500. That is because when the insur-ance company tries to buy a bond to fund$100 per year forever, it must buy a $2,500 per-petual bond, which at 4 percent interest yields$100 per year. That is why the market value ofthe liability increases from $2,000 to $2,500upon the fall in interest rates.

If the sponsor in my hypothetical examplealso held a perpetuity on its asset side, thenthe increase in the liability value of the per-petuity is exactly matched by the capital gainon its assets. Obviously, if the interest ratefalls to 4 percent, the firm can sell its 5 per-cent coupon rate bond at a significant pre-mium. How much would the bond sell for?

The answer is $2,500. At this market price,the coupon rate (5 percent) times the facevalue of the bond ($2,000) equals $100,which is 4 percent of the $2,500 purchaseprice. So if the pension plan is fully funded tostart with (which I assume to be the case),then it is insulated or “immunized” againstinterest rate risk because it has “matched” itsassets to its liabilities. Because the sponsorholds a bond with the same duration as itsliabilities, the increase (decrease) in the pen-sion liability is exactly matched by the capitalgain (or loss) in the bond value no matterwhat happens to interest rates.

In reality, there is no such thing as a per-petual annuity, nor can one purchase a per-petual bond. But the same principles ofmatching are unaffected (though they aremade more complicated) for obligations thatare more realistic.

How Stocks Create Potential ExposuresTypically, firms hold some bonds of dura-

tions matched to liabilities, but rarely dothose bonds make up 100 percent of theportfolio. A more typical portfolio has 40

6

A plan sponsorcan protect itself

against interestrate fluctuationsby fully fundingits pension fund

with bonds of thesame duration as

its liabilities.

percent bonds and 60 percent stocks. Stockvalues tend to increase if the interest ratefalls, but the correlation is quite weak. Thenormal volatility in stock returns swamps thesmall correlation that exists between pensionliabilities and stock values.

If a plan holds a majority of its assets inthe form of stocks, it is “mismatched,” in thesense that it is vulnerable to interest rate risk.All else being the same, a reduction in marketinterest rates creates additional underfund-ing. The present value of promised nominalbenefits increases, but the assets that backthose liabilities do not necessarily do so.Firms hope that the average return on stockthey hold will over time exceed bond returns,in which case they can get away with con-tributing less to their pension funds. Onaverage, that reasoning is correct. But it addsconsiderable exposure to the PBGC.

To understand the role of stocks in thepension plan, one needs to remember thatthe probability of bankruptcy depends onthe overall performance of the economy.Some firms enter bankruptcy during high-growth periods in the economy, but bank-ruptcies are far more likely during or follow-ing economic downturns.

The correlation of bankruptcy risks withpoor performance of the economy as a wholegenerates important downside exposure for thePBGC. Consider a pension plan that holds allstock in its portfolio. In bad times, the portfoliovalue falls, which increases the overall amountof underfunding and the risks of bankruptcy. Apension plan that is fully funded for insuredbenefits with stocks is riskier than a fully fund-ed plan with a portfolio that comprises dura-tion-matched bonds. Consequently, it shouldpay the pension insurer—in this case, thePBGC—higher premiums for the additionalpotential exposure it creates.

For example, if a plan is fully funded withstock, a 40 percent decline in the value of thatstock leaves the plan only 60 percent funded.The new underfunding triggers higher con-tribution requirements, but those contribu-tions are spread over a period of time, so thatthe underfunding is prone to linger, particu-

larly if, in the case of flat-benefit union plans,collective bargaining requires the sponsor toraise benefits during this period.

Although it is tempting to think that thePBGC benefits if stock returns are high, in real-ity, it does not. The reason is that bankruptcyrisks are low when markets perform well and, ifa bankruptcy occurs when a pension plan isoverfunded, the PBGC does not receive a “neg-ative” claim. The PBGC exposure is completelyone-sided vis-à-vis stock performance. Thesponsor wins when equity returns are high, andthe PBGC stands to lose when they are low.

A plan presents essentially zero exposureto the PBGC if it is fully funded and if itholds a portfolio of high-grade corporatebonds (or Treasuries) of the same duration asits pension liabilities. Not only is the planimmunized against interest rate risk, it isinsulated from stock market risks anddefault risks. For simplicity, assume that ifthe economy is “up” there is no bankruptcyrisk. If the economy is “down” then the prob-ability of bankruptcy for the typical sponsoris p. Also, assume that if the economy is“down” the equity return is minus r. Assumethat any bonds in the portfolio are durationmatched against liabilities. In this highly styl-ized illustration, the expected loss frominsuring a plan for a period of time is

EL = pL* [(1 – f) – αrf]

where α is the share of stocks in the pensionportfolio, f is the starting funding ratio, andL* are termination liabilities.

This expression tells us that the economicprice of pension insurance is a product of threefactors: the probability of bankruptcy, p; termi-nation liabilities, L*; and the funding ratioupon termination in a downturn (the terminside the square brackets). The first terminside brackets, 1 – f, is the portion of liabilitiesthat is underfunded at the beginning of theperiod. The second term, αrf, captures the sub-sequent effect of equity performance on expo-sure. A plan that is 100 percent funded at thestart of the period and holds only (duration-matched) bonds presents no exposure.

7

The correlationof bankruptcyrisks with poorperformance ofthe economy as awhole generatesimportant downside exposure for thePBGC.

0∅ (1)

Market Risk and Pension InsuranceUnder normal conditions, ignoring the

administrative costs of operating the insur-ance, the economic premium for any insur-ance coverage is expected to equal expectedlosses, such as those depicted in the expres-sion above. In the case of pension insurance,however, the market price is higher. Why?The answer lies in the correlation of claimswith downturns in the economy.

The backers of the insurance require ahigher return to invest in the insurance forthe same reason that stockholders require ahigher expected return than bondholders. Ifall stocks moved randomly and independent-ly of each other, then holding a diversifiedportfolio of individual stocks would yield anaverage return that had a minuscule variance.In that case, stock investors would expect areturn commensurate with the return on abond. In reality, a diversified portfolio ofstocks has a very large variance becausereturns are correlated across stocks. That isthe so-called market risk, or beta risk, whichcannot be diversified away.9

Put somewhat differently, holders of high-quality bonds accept a lower return thanstockholders because they experience lessvolatility in their portfolio value.10 Stockswhose returns evince a higher amplitude thanmarket returns (high-beta stocks) have a high-er expected return than stocks whose returnshave a lower amplitude than market returns(low-beta stocks). No one wants to lose a sub-stantial portion of his portfolio in downtimes, and thus stockholders must earn ahigher premium relative to bondholders toobtain equilibrium in the financial markets.

The same principle applies to insurancemarkets. When insurable events are mostlyunrelated (for example, auto accidentclaims), average claims experience is charac-terized by a trivial variance, so long as theinsurer covers many areas of exposure. Evenif exposure is correlated, say with flood dam-age along the Mississippi River, the insurercan offer flood damage insurance in manydifferent areas of the country and indeed theworld and, thus, diversify away those risks

across a large number of insureds (if it isunusually wet in one spot, it must be abnor-mally dry somewhere else).

Pension insurance claims are not only sus-ceptible to bunching but are also (negatively)correlated with market returns. The backersof the insurance are asked to pay out largeclaims precisely at times during which theirportfolios are falling. Investors require a pre-mium to underwrite that kind of risk com-pared to other risks that are uncorrelatedwith market returns. Because the idea isdirectly related to the portfolio risks, pensioninsurance is said to carry beta risk.

There are two components of beta risk inpension insurance: One arises because pen-sion funding is a function of stock perfor-mance in all plans that hold stock invest-ments. The second arises because the proba-bility of bankruptcy itself increases in downmarkets. Even if a plan holds only bonds butis underfunded, the probability that thisunderfunding leads to a PBGC claim is high-er in down markets than up markets, andthus the insurer’s loss exposure is negativelycorrelated with stock market returns.

The beta risk itself gives rise to the possi-bility of catastrophic events. In periods dur-ing which economic performance is poor, thePBGC runs the risk that many bankruptciescan occur within a short period, each charac-terized by abnormally low funding levels.Even under normal economic conditions, theinsurer is vulnerable to a few large claimsarriving by chance, or to the downturn of anentire industry. Catastrophic exposure ischaracterized by the possibility of severe“tail” events—that is, events with a smallprobability of occurring but with very largeclaims when they occur. The cost of this kindof coverage is hard to estimate because cata-strophic events, by definition, are rare, whichrestricts the usefulness of historical data forprojecting claims.

Illustration of the Catastrophic Event Claims experience over the brief history of

the PBGC illustrates both the catastrophicand beta risk inherent in federal pension

8

In periods duringwhich economic

performance ispoor, the PBGC

runs the risk that many

bankruptcies canoccur within a

short period.

insurance. Figure 3 shows PBGC claimsthrough 2003.11 The bar columns show nom-inal claims (measured along the left verticalaxis). The line distribution shows the two-year rolling-average equity return on the S&P500 Index (right vertical axis).

Since 1970 there have been only two peri-ods during which equity returns were so poorand persistent as to generate a negative two-year return. The first was 1974, which cap-tures the stock market reduction of 1973–74(the average return during those two yearswas minus 20 percent per annum). Theunderfunding created by this event and thecorresponding increase in bankruptcy rateswere important stimuli for the enactment ofthe federal pension insurance program onLabor Day 1974.12

The legislation that created the PBGC wasnot retroactive. It offered pension insuranceagainst claims arising in the future.Fortunately, the post-1974 period was charac-terized by a dramatic reversal in equity perfor-mance. Indeed, stock returns generally werequite favorable for the insurance programthroughout the 1980s and 1990s, creating the

impression that perhaps the insurance wasnot especially costly. Though premium rateswere increased occasionally, total premiumsrarely exceeded $1 billion per annum, and asof year-end 2000 the PBGC enjoyed a $9.7 bil-lion surplus position. It had not yet experi-enced the conditions that give rise to the cat-astrophic nature of the insurance, notably, asubstantial downturn in the economy.

Figure 4 shows the distribution of insuredpension liabilities by plan funding ratio as ofthe start of 2001, just prior to the stock mar-ket downturn. The figure gives little cause foralarm. Total underfunding amounted toonly $31.2 billion. The figure adequatelydemonstrates the pitfall of gauging exposureby relying on a snapshot of funding ratioswithout also taking into account the poten-tial impact of negative stock returns on pen-sion assets. For example, U.S. Airways, whichbecame a claim against the PBGC in 2003,was 104 percent funded in 2000 but only 50percent funded two years later.13

From the time the snapshot in figure 4 wastaken, equity returns fell by almost 40 percentthrough the end of 2002, virtually reproduc-

9

U.S. Airways,which became aclaim against thePBGC in 2003,was 104 percentfunded in 2000but only 50 percent fundedtwo years later.

Figure 3PBGC Claims vs. Stock Market Returns

Year

Cla

ims

($ m

illio

ns)

S&P

Return, Tw

o-Year R

olling Average

U.S. Airways,which became aclaim against thePBGC in 2003, was104 percent fund-ed in 2000 butonly 50 percent fundedtwo years later.

0

1,000

2,000

3,000

4,000

5,000

6,000

7,000

8,000

9,000

10,000

1971

1973

1975

1977

1979

1981

1983

1985

1987

1989

1991

1993

1995

1997

1999

2001

2003

2005

-0.3

-0.2

-0.1

0

0.1

0.2

0.3

0.4

Claims

S&P Return

Figure 3PBGC Claims vs. Stock Market Returns

ing the stock market crash of 1973–74. Figure5 demonstrates the effect of this downturn onplan underfunding. The figure shows two

series of underfunding, one based on IRSForm 5500 Annual Report data and the other(since 1996) based on submissions to the

10

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

$31.2 Billion

Figure 4Funding Ratios as of 12/31/2000

1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002

Figure 5Pension Underfunding, 1980–2003

Note: The Form 5500 series provide funding ratios as of January 1 of the year and count vested liabilities only. Data

are taken from Schedule B attachments. The PBGC series pertain to September 30 of the year and include all accrued

liabilities. The PBGC obtains its data from firms that by law must divulge underfunding beyond $50 million (so-called

4010 filings). Both series are adjusted to a common mortality table and both adjust to a common PBGC rate (though

the interest rates for the same year can differ because the series are calculated at two different times of the year).

Funding Ratio x 100

Perc

ent o

f L

iabi

litie

s$

Bill

ions

of

Und

erfu

ndin

g

Funding Ratio x 100

Source: 2001 Form 5500 Annual Reports (Schedule B).

0

2

4

6

8

10

12

14

16

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Total Underfunding: $31.2 Billion

More than 200% Funded

Figure 4Funding Ratios as of 12/31/2000

Figure 5Pension Underfunding, 1980 2003

Figure 4Funding Ratios as of 12/31/2000

0

50

100

150

200

250

300

350

400

450

1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002

Form 5500 Series(January 1,vested benefits)

PBGC Series4010 Filings(September 30,all benefits)

Figure 5Pension Underfunding, 1980 2003

PBGC made by plans with more than $50 mil-lion in underfunding.14 Although the seriesevince some differences, it is apparent thatpension plan underfunding increased startingin 2001 and reached about $400 billion by thefall of 2002.

Unlike its initial experience, in which thePBGC escaped with relatively few claims fol-lowing the economic downturn in the early1970s, the agency has not escaped the conse-quences of the stock market downturn thatstarted in 2001. This time it has absorbed thefull brunt of the bankruptcies and under-funding that accompanied the downsideevent (see Figure 3). Over a short time, thePBGC has accumulated $15.9 billion inclaims, more than twice as much as the $6.1billion it assumed over the entire period1980–99 ($9.5 billion in 2003 dollars). In realterms, per annum claims during the lastthree years have been more than 15 times theper annum claims over all prior years.

The downside event erased the heretoforementioned $9.7 billion surplus and replacedit with a deficit of $11.2 billion as of the endof 2003, a reversal of positions of about $21

billion over the span of three years. Moreclaims may yet materialize from this period.As of the end of 2003, underfunding stood at$350 billion, and $83 billion of that amountwas in plans sponsored by firms whosebonds were rated as “junk.”15

The PBGC’s Financial Future

The future position of the PBGC dependsin large part on economic conditions thatunfold. Like most other catastrophic insur-ers, the PBGC uses a stochastic simulationmodel to capture the distribution of poten-tial net financial outcomes. Essentially, wheninsurers lack reliable data from large num-bers of observations, the insurers use infor-mation about what is known to affect lossesto simulate future conditions. For example,insurance companies that cover hurricanedamage use weather models to simulatethousands of possible spring hurricane sea-sons to determine the distribution of possi-ble claims outcomes. The PBGC uses a com-

11

Future premiumsare likely insufficient toeven hold thePBGC deficit atcurrent levelsover 10 years.

Figure 6Distribution of the PBGC’s Financial Position in 2013 (starting position 2004)

Source: 2003 PBGC Annual Report.

Num

ber

of S

imul

atio

ns (

out o

f 5,

000)

$ Billions (present value)

The PBGC’s Financial Future

Future premiumsare likely insufficient to evenhold the PBGCdeficit at currentlevels over 10years.

0

20

40

60

80

100

120

140

-130

-120

-110

-100 -9

0-8

0-7

0-6

0-5

0-4

0-3

0-2

0-1

0 0 10 20 30 40 50 50 60

Mean: $ (18.7)Std. Dev. $ 22.71% $ (82.5)5% $ (60.3)10% $ (49.0)25% $ (31.8)50% $ (16.2)75% $ (3.7)90% $ 8.4

Figure 6Distribution of the PBGC s Financial Position in 2013 (starting position 2004)

plex model that accommodates stochasticmovement in stock returns, interest rates,employment changes, bankruptcy rates, andother factors to create various scenarios thatcould affect the agency. Typically, it bases itsprojections on 5,000 simulations.16

Figure 6 shows the distribution of project-ed possible outcomes in 2013 as reported inthe 2003 PBGC Annual Report.17 The expectednet position is an $18.7 billion deficit in pres-ent value terms. In other words, future pre-miums are likely insufficient to even hold thedeficit at current levels over 10 years. Theprojected deficit (expressed in present valueterms) is about 18 times higher than annualpremiums.

The expected deficit, of course, is merelythe average of all the positions that are possi-ble 10 years hence. There is about a one-in-five chance that the PBGC will have suffi-ciently favorable experience to accumulate asurplus position in 10 years. There is a 10 per-cent chance that the deficit will be at least$49 billion, a 5 percent probability that thedeficit will be at least $60.3 billion, and a 1percent probability that it will be at least$82.5 billion.

Recent Legislation

Recent temporary legislation has worked tofurther weaken the PBGC’s financial positionby lowering sponsors’ required contributionsthat are intended to reduce the underfundingshown in Figure 5. The Internal RevenueService minimum contribution rules providethe PBGC’s most significant defense againstfinancial ruin. Dramatic underfunding (de-fined as less than 80 percent funding ratio)triggers the Deficit Reduction Contribution,which is especially helpful in reversing seriouslevels of exposure.

In the context of the claims history shownin Figure 3, the additional funding requiredin “bad times” is the primary way in whichplan sponsors accept some of the beta riskinherent in the insurance. That is, if sponsorsdecline to fully fund their pensions and

choose to hold stock instead of duration-matched bonds, they expose themselves tothe possibility of triggering stringent fund-ing rules upon encountering poor stock mar-ket performance. In a study I did with StevenBoyce, we showed that the contribution rulesreduced the overall cost of pension insuranceby about half, partly by reducing underfund-ing and partly by shifting some of the betarisk away from the PBGC and toward premi-um payers themselves.18

Although the deficit reduction rules havebeen in play since 1986, they have not beenespecially relevant because favorable stockmarket returns have reduced the frequency ofunderfunding, especially dramatic under-funding. As shown in Figure 3, the first truetest of those rules came with the stock marketdecline of 2001–02. The decline coincidedwith reductions in interest rates, which them-selves increase underfunding. When facedwith the requirement to absorb their part ofthe beta problem by increasing contributions,sponsors appealed to Congress to reduce con-tribution requirements. In temporary legisla-tion that lapsed on January 1, 2004, Congressreduced the funding rules by taking the nor-mal interest rates used for this purpose—thoseof long-term Treasury bonds—and allowingsponsors to use rates equal to 120 percent ofthose rates for the years 2002 and 2003. Whilethat measure may have reduced underfundingfor the purposes of making contributions, itdid not affect plans’ underfunding as mea-sured by the market.

That temporary legislation is one reasonwhy U.S. Airways, though its funding ratiofell to 50 percent at the time that its planbecame a claim against the PBGC, legallymade no contributions for the four yearsprior to the claim date. The PBGC took aclaim of $2.2 billion.19 Similarly, BethlehemSteel was 45 percent funded when it wastaken as a claim in 2003, and yet the last cal-culation of underfunding for purposes ofmaking contributions put the company at 84percent.

In April 2004 Congress passed new tempo-rary relief legislation (H.R. 3108). Instead of

12

If sponsorsdecline to fully

fund their pensions and

choose to holdstock instead of

duration-matchedbonds, they

expose themselvesto the possibility

of triggeringstringent funding

rules uponencountering

poor stock market

performance.

permitting 120 percent of long-term Treasurybond rates, H.R. 3108 allows sponsors to usecorporate bond rates, which are about 100basis points higher than long-term Treasuries.Corporate rates are higher because corporatebonds have varying chances of default. Thoserates are inappropriate for PBGC purposesbecause the PBGC would not default on itsinsured promises. It is widely believed thatCongress will bail out the PBGC regardless ofthe deficit it accumulates. That guaranteemeans that the promised flow of annuitiesshould be discounted by long-term Treasuryrates.

Put somewhat differently, the only way aninsurance company can guarantee to pay anannuity flow is to hold duration-matchedrisk-free bonds. Use of corporate bonds forthis purpose could create the possibility ofdefault on some part of the payments becauseof default risks inherent in corporate bonds.

In addition, riders to H.R. 3108 exemptplan sponsors in the steel and airline indus-tries from most of the deficit reduction con-tributions that are still required from thosesponsors even after the application of thehigher interest rates. The estimated reliefthose two industries will receive from the rid-ers is approximately $1.6 billion. Becausemost recent PBGC claims come from thoseindustries, the blanket waivers impose largeadditional exposure on the PBGC.

How Can the Problem Be Fixed?

As long as sponsors of underfunded pen-sion plans are not held responsible for theexposure they impose on the PBGC, ultimate-ly either the premium level must increase, inwhich case some of the cost will be shifted towell-funded pensions in the short run, or, ifexposures create claims that reach catastroph-ic levels, taxpayers will be called upon to pro-vide a bailout through the PBGC.

Shortfalls of Premium RulesIn principle, the PBGC is supposed to col-

lect a variable rate premium equal to $9 foreach $1,000 of underfunding. But plan spon-sors can avoid premiums if their actuariesclaim that the plan is at the so-called fullfunding limit, which arises if plans con-tributed more than the minimum requiredin past years. It is not relevant if the plan con-tinues to be underfunded. Moreover, recentlegislation increased the interest rate used tocalculate these premiums as well.

The ineffectiveness of the premium rules ascurrently enforced is starkly seen if we multi-ply the $9 rate against market underfunding.For example, in 2002 the PBGC calculatedthat, based on market values, underfundingstood at $400 billion. If it had collected 0.9 of1 percent of that amount as prescribed by law,the PBGC would have collected $3.6 billion invariable rate revenues in 2002. In reality, totalpremiums equaled $787 million, of which$586 million came from a fixed assessment of$19 per plan participant. That means that thePBGC collected only about $200 million fromthe variable rate premium, or about $0.50 per$1,000 of underfunding, or about 5.5 percentof the advertised $9 charge.

Market Pricing through a Self-InsurancePool

Suppose that the government terminatedits role in the PBGC and required all pensionsto belong to a self-insurance pool. To obtain aclean start, suppose that the federal govern-ment handed over sufficient monies to elimi-nate the projected mean expected $18.7 billiondeficit (Figure 6). This includes $7.5 billionmore than the PBGC’s current $11.2 billiondeficit because current conditions are mostlikely consistent with claims that outstrip rev-enues over the next 10 years. The governingboard of the pool would set policy and premi-ums (board members in a pool arrangementwould presumably be elected by pensionplans, where votes are proportional to pensionparticipants in each plan).

Sponsors in the pool are held jointly liablefor any deficit that develops. That ensuresthat all sponsors will aggressively work toalign premiums with exposure and to ensure

13

As long as sponsors ofunderfundedpension plans are not heldresponsible forthe exposure theyimpose on thePBGC, ultimatelyeither the premium levelmust increase ortaxpayers will becalled upon toprovide a bailoutthrough thePBGC.

a solvent system. As a way of ensuring theelimination of cross-subsidies, pensionsshould be permitted to find coverage in theprivate sector after some period of time.Sponsors of underfunded plans will thenhave an interest in reducing their reliance onpayments from well-funded plans so as tokeep them as a source of some help in solvingthe underfunding problem.

Predictably, the board would reset thevariable rate premium each year (or more fre-quently) to reflect current economic condi-tions. In effect, it would set premiums eachperiod proportional to the average probabili-ty of bankruptcy in period t, pt, in expression(1) above. I reproduce the formula with a sub-script i to denote the expected loss from theith plan and t to denote year:

ELit = ptLit*[(1 — fit) — αitrtfit]

Recall that this formula assumes that thedownside stock return is minus r, whereas inreality it could take on many values. Althoughthe expression is oversimplified, it is adequatefor understanding how a market premium isapplied in principle.

Note that the bankruptcy risk of the par-ticular sponsor does not enter the expression(there is no i in the subscript to p). That isbecause the insurance is term renewable. Justas a person’s life insurance premium does notincrease when he is diagnosed with cancer (ina term renewable policy), neither does the pre-mium increase for a firm that becomes finan-cially unstable. But just as a cancer victim isnot permitted to purchase more life insuranceafter diagnosis, neither should a sponsor beallowed to increase benefits upon reaching acritical probability of bankruptcy. The overallpremium rate, however, can increase ordecrease as the overall risk of bankruptcyacross all insureds increases or decreases.

In periods after poor economic perfor-mance, rates increase to reflect high expectedrates of bankruptcy. The insureds themselvesabsorb market volatility instead of offload-ing it to a third party (namely, the taxpayer).In short, they absorb the beta risk. The oppo-site happens when market returns are high.Thus, on average, risk-based premiums movecountercyclically with economic conditions.

That policy dramatically increases volatil-ity of aggregate premiums but commensu-

14

On average, risk-based

premiums movecountercyclically

with economicconditions.

Figure 7Simulated Distribution of the PBGC’s Financial Position in 2013

Source: Boyce and Ippolito.

0∅ (2)

$ Billions (present value)

Num

ber

of S

imul

atio

ns

(out

of

5,00

0)

On average, risk-based

premiums movecountercyclically

with economicconditions.

0

50

100

150

200

250

300

350

-130

-120

-110

-100 -90 -80 -70 -60 -50 -40 -30 -20 -10 0 10 20 30 40 50 50 60

Projected Position: Current Rules

Position with $18.8 Billion Bailoutand Prospective Economic Pricing

Figure 7Simulated Distribution of the PBGC s Financial Position in 2013

rately reduces the volatility of the pool’spotential economic conditions. The onlyvariation in outcome is attributable to idio-syncratic risk. In periods of below-averageclaims, a fund builds up. In periods of above-average claims, the fund is drawn down (andtemporary premium surcharges are assessedif necessary).

Steve Boyce and I simulated how insur-ance works under that policy.20 Figure 7shows that the pool arrangement wouldaffect the insurer’s projected net financialposition. Notably, we eliminated the fixedcharge in the PBGC pricing structure andreplaced it with a variable rate premiumassessed against market value underfunding.(We did not reflect charges as a function ofportfolio composition.)

The bar distribution in the figure depictsthe projected PBGC financial position 10years out under current conditions (takendirectly from Figure 6). The line distributioncaptures the outcomes when the insurance isoperated as a self-insurance pool and thepool resets the variable rate against marked-to-market underfunding each period toaccommodate current market conditions.The reduction in volatility brought about byaggregate economic pricing is dramatic. Theoutcomes under this policy have no beta riskand evince a distribution that is more com-patible with a private-sector solution to theinsurance problem. All that remains is a rela-tively small amount of idiosyncratic risk.

Reduction in Moral HazardIf the board merely enforced a premium as

described in (2), sponsors would have animmediate incentive to fund their plans.They could then reduce their premiums tozero if they fully funded their plans withduration-matched bonds. If they partiallyfunded those plans with stocks, they coulddramatically reduce their premiums if theyheld assets that exceeded liabilities so as tocreate a cushion against reductions in stockvalue.21 Plans that wanted to expose the poolto lots of risk would be required to pay com-mensurately higher premiums. Indeed, with

a well-functioning variable rate premium asdescribed in (2), there is no need for a com-plex set of funding rules. Sponsors have aneconomic incentive to fund their plans.

Once taxpayers were removed as ultimateguarantors of the insurance, the plans them-selves (and most notably the better fundedplans) would have an incentive to align pre-miums with exposure, and plan sponsorswould have to face up to the problems thattheir own underfunding creates.

Conclusion

Pension underfunding is entirely control-lable by companies that sponsor pensions.Under current conditions, there is almost nocharge for imposing exposure on the PBGC.Most of the premium revenue is a fixed chargeper participant without regard to underfund-ing. Although the agency advertises a $9charge per $1,000 of underfunding, in effect,through various loopholes, it actually collectsonly about $0.50 per $1,000. Implicitly, thetaxpayer backs the insurance agency andalmost certainly will be called upon to bail outthe agency if it encounters catastrophic losslevels.

If mandatory pension insurance wereoffered in the private sector, or through amandatory self-insurance pool whose mem-bers were jointly liable for deficits, then pre-miums would reflect exposure. Sponsorswould learn to economize on the amount ofunderfunding they carry.

A large part of the problem would disap-pear even if the insurance were operated as itis now, with one change. Eliminate the loop-holes that permit sponsors of underfundedplans to evade the variable rate premium andrequire sponsors to calculate market valueunderfunding. That change would dramati-cally increase revenues to the PBGC, reducingthe need for a bailout and greatly reducingthe level of underfunding.

Although it is hard to point to a singleinstance in which a federally subsidized insur-ance program was successfully privatized or

15

If mandatorypension insurance wereoffered in the private sector, orthrough amandatory self-insurance poolwhose memberswere jointly liablefor deficits, thenpremiums wouldreflect exposure.

reformed in such a way as to eliminate cross-subsidies and federal bailout guarantees, thereis a potential avenue by which the PBGC expe-rience can seriously inform public policy onpensions. Defined-benefit plans, which are theresponsibility of the PBGC, account for onlyabout one in four covered workers in the pri-vate sector. The largest area of new pensiongrowth is 401(k) plans. Those pensions, how-ever, have their own potential for risks. Insome cases, workers in those plans invest heav-ily in company stock of the firm that employsthem. In the event of that company’s bank-ruptcy, as in the case of Enron Corporation,covered workers can lose their entire pensionbalances. Perhaps it is prudent to think aboutdiversification requirements for those plansbefore the PBGC has a new subsidiary cover-ing nondiversified 401(k) plans.

Appendix: Definition of Pension Terms

Defined-Benefit Plan: A pension thatpays an annuity at retirement age propor-tional to service in the company and usuallyproportional to final salary.

Defined-Contribution Plan: A pensioninto which the firm deposits some percent-age of pay in workers’ individual accounts;sort of a tax-preferred savings account.

401(k) Plan: A special kind of defined-contribution plan in which employees canchoose to save. The firm often matchesemployee contributions according to somepredetermined formula.

Termination Benefits: The present valueof workers’ accrued pension annuity earnedon the basis of current service if the pensionterminates immediately.

Ongoing Benefits: The present value ofworkers’ accrued pension annuity earned onthe basis of current service on the assumptionthat the firm will continue the pension plan.

Contingent Benefits: The differencebetween ongoing and termination benefits.Also known as the pension capital loss fromquitting.

Underfunding: The amount of termina-tion benefits in excess of assets in the pensiontrust fund.

Funding Ratio: Pension assets to termi-nation benefits.

Duration: The percent change in thevalue of a bond as a result of a one percentagepoint change in the interest rate. A bond ofduration 10 loses 10 percent of its value if themarket interest rate increases by 1 percent.

Immunized Portfolio: A term used todescribe a portfolio of bonds that has thesame duration as pension liabilities. A fullyfunded immunized portfolio does not devel-op overfunding or underfunding if the inter-est rate changes.

Beta Risk: The term used to describeundiversifiable risk that is linked to the over-all stock market return.

Idiosyncratic Risk: Random risk thatcan be diversified away.

Notes1. James Wooten, “The Most Glorious Story ofFailure in the Business: The Studebaker-PackardCorporation and the Origins of ERISA,” BuffaloLaw Review 49 (2001): 683.

2. The PBGC also operates a smaller program,which covers union workers in multiemployerplans such as the Central States Teamsters’ plan.This program is also in deficit, but it is not sensi-tive to individual company bankruptcy risk, andso it warrants a different kind of analysis, which isnot pursued here. It is vulnerable to industry risk,for example, replacement of union workers withnonunion workers at construction sites or theexport of textile jobs to other countries.

3. About 50 percent of workers still covered bytraditional defined-benefit plans are in plans thatare collectively bargained.

4. The future value of $40,000 growing at 6 percentannually for 20 years is equal to FV = $40,000 * (1 +0.06)20 = $128,285.

5. 0.015 * 20 * $128,285 = $38,485.

6. 0.015 * 20 * $40,000 = $12,000.

7. Terminations almost always occur in Chapter 7(dissolution) and often accompany Chapter 11bankruptcies as well, especially if the plan is under-

16

funded. If underfunding exists and the bankruptcyjudge deems termination necessary to the success-ful reorganization of the company, the PBGC musttake the plan as a claim.

8. Not all collectively bargained plans are flat-benefit plans, but virtually all flat-benefit plansare collectively bargained.

9. The nomenclature “beta risk” comes from theNobel Prize–winning paper that introduced theidea of market risk. See William Sharpe, “CapitalAsset Prices: A Theory of Market Equilibriumunder Conditions of Risk,” Journal of Finance 19(September 1964): 425–42.

10. There is of course inflation risk in bonds, butone can imagine holding I-bonds issued by theU.S. Treasury, which pay a real interest rate plusinflation. Those bonds have no inflation risks, soone expects a low return on them.

11. Historical PBGC data are found in the PensionInsurance Data Book, which is issued annually bythe PBGC.

12. See Richard A. Ippolito, “A Study of theRegulatory Effect of the Employee RetirementIncome Security Act,” Journal of Law and Economics31 (April 1988): 85–125. See also Wooten.

13. Air Line Pilots Association, “A Media Primer

on Pensions and the Airline Industry,” http://www.alpa.org.

14. The 5500 data are as of the first of the year andmeasure vested liabilities only. The PBGC seriespertains to measures as of September 30 of theyear; includes all liabilities, including unvestedbenefits; and is almost four years more currentthan 5500 data.

15. The numbers in this paragraph are found inPension Benefit Guaranty Corporation (2004),2003 PBGC Annual Report.

16. The model is described in detail in StevenBoyce and Richard Ippolito, “The Cost of PensionInsurance,” Journal of Risk and Insurance 69, no. 2(2002): 121–70.

17. See Pension Benefit Guaranty Corporation.

18. See Boyce and Ippolito.

19. See Pension Benefit Guaranty Corporation.

20. See Boyce and Ippolito.

21. To overfund for termination benefits, flat-benefit plans may have to switch to a salary-relat-ed formula, or Congress would need to modifythe tax code to permit flat-benefit plans to antic-ipate periodic increases.

17

Published by the Cato Institute, Policy Analysis is a regularseries evaluating government policies and offering proposalsfor reform. Nothing in Policy Analysis should be construed asnecessarily reflecting the views of the Cato Institute or as anattempt to aid or hinder the passage of any bill before Con-gress. Contact the Cato Institute for reprint permission.

Additional copies of Policy Analysis are $6.00 each ($3.00each for five or more). To order, or for a complete listing ofavailable studies, write the Cato Institute, 1000 MassachusettsAve., N.W., Washington, D.C. 20001 or call tollfree 1-800-767-1241 (8:30 a.m.-4:30 p.m. easterntime). Fax (202) 842-3490 • www.cato.org