Thank you very much for this opportunity to testify on mandatoryretirement age regulations in the United States. I am very honoredby it.1
Ideally, the decision about when to retire should be madevoluntarily by workers in response to labor market conditions.Mandatory retirement age rules have been eliminated in most privatesector jobs as a result of anti-age-discrimination laws that wereintroduced beginning in the 1960s. Were they allowed, however,private sector employers would likely incorporate them inemployment contracts designed for ensuring and improvingworker-efficiency. Instead, private firms structure long-termincentive contracts including features of defined benefit pensionplans, other non-wage benefits, and severance packages to induceearly job terminations. Retirement incentives incorporated in suchlong-term incentive contracts appear to have spurred the trendtoward earlier retirement in the United States.
Mandatory retirement age rules still prevail in some private andpublic-sector occupations: State and local police (55-60) andfirefighters (55-60); federal firefighters (57); federal lawenforcement and corrections officers (57); and air trafficcontrollers (56, if hired after 1972); and commercial airlinepilots (60). These are "earlier-than-normal" retirement agescompared to the vast majority of other occupations.
Mandatory retirement age restrictions were introduced in theseoccupations several decades ago, primarily for ensuring their safeand effective conduct. Under today's conditions, however, theseretirement age rules appear to be outdated. The need to revisethese rules appears urgent due to impending worker shortages. Itappears desirable to introduce long-term incentive compensationstructures in these jobs similar to those in the private sector.However, the manner in which they should be introduced and whetherthey would deliver retirement choices to workers whilesimultaneously ensuring safe and effective job performance requiresfurther examination.
My testimony is comprised of several parts. First I reportfindings on private sector compensation structures that aredesigned to elicit worker efficiency and loyalty, and yet inducetimely retirements. The findings suggest that these contractsinvolve divorcing current productivity from currentcompensation--by postponing compensation from workers' early- tolate-career stages. Such contracts would not be profitable ifworkers stayed on the job to collect wages in excess of theirproductivity for too long. Thus, it appears that private firmscould put mandatory retirement age rules to good use.
However, these "incentive" contracts would become infeasible ifemployers terminated workers too early to avoid paying themseniority rents. This makes the case for anti-age discriminationrules. As explained below, it turns out that prohibitions againstage discrimination are more useful than mandatory retirement agerules in making such contracts feasible. Evidence on U.S. firmssuggests the prevalence of long-term incentive contracts. Thus, inthe United States, firms induce workers to retire early byappropriately structuring non-wage elements of compensation, whileanti-age discrimination laws provide an external commitmentmechanism against premature discharges--thereby inducing workers toaccept long-term incentive contracts.
Employment contracts are structured differently in the academicsphere. The recent removal (in 1994) of the age-70 mandatoryretirement rule has resulted in many older and highly compensatedfaculty members remaining on university payrolls. Universities havenot since been successful in inducing earlier retirements viapension plan and other incentives because these are costly to offerto faculty who are already close to retirement. However,universities may over time adopt long-term incentive contracts foryounger employees similar to those prevalent in large privatecompanies.
Second, health and longevity of the U.S population generally hasbeen improving. This may mean that jobs that could not be conductedeffectively by workers after their late fifties, now can be. Ahistorical comparison of mortality rates suggests that those agedin their early sixties today are as healthy as were those in theirmid fifties a few decades ago--when the mandatory retirement agerules were first imposed in the occupations underconsideration.
Existing mandatory retirement age rules appear unfair for somecategories of workers--such as pilots and air-traffic controllersthat are subject to such rules. Because they spend their careers injobs requiring specific, non-transferable skills, early jobseparation now results in longer spells of unemployment or forcedretirement despite possessing the ability to conduct their jobscompetently. Evidence suggests that independently of their tenurein earlier jobs older workers have greater difficulty in findingjobs in the private sector.
Third, improvements in technology imply that, other thingsequal, federal police, firefighting, and air-traffic-control jobsmay have become physically easier to conduct. Evidence alsosuggests that health is now much less important as a considerationfor the decision to retire than was the case several decades agobecause of both, lower physical job demands and improvements in thetreatment of chronic conditions.
Fourth, better technology and equipment increase the need fortrained and experienced personnel to operate and coordinateactivities. Hence, although better technology makes the jobs easierto perform, it could require more rather than fewer skilledworkers.
Finally, as the baby boomers exit the workforce, the burden onworking generations to support a larger and longer lived retireepopulation will increase. Baby boomer retirements will likelycreate skilled- and experienced-worker shortages in manyoccupations. The shortages are likely to become more acute inoccupations that impose relatively early mandatory retirement agerestrictions.
Overall, impending worker shortages in occupations withmandatory retirement age restrictions motivate the revision ofthese rules. However, as experience in the U.S. academic sphereindicates, addressing the shortages by immediately removing theserestrictions may lead to other problems--such as adisproportionately older workforce. Introducing long-term incentivecontracts to improve worker efficiency and yet provide flexibilityin retirement decisions appears to be desirable. However, suchemployment contracts "pay off" only in the long-term and should berestricted to younger workers and new hires. Worker shortages inthe near term could be addressed by revising mandatory retirementage rules upward. The decision to elimate these rules may neverbecome necessary if younger worker and new hires are offeredlong-term incentive contracts. As these contracts become morewidespread they may improve worker efficiency and induce timelyretirements. Mandatory retirement age rules will automatically fallby the wayside as workers who remain subject to them leave theworkforce.
Section 1: Mandatory Retirement Age Rules vs.Anti-Age-Discrimination Laws in the Private Sector
Anti age discrimination laws and mandatory age retirement rulesare polar opposites. Would private employers enforce mandatoryretirement age rules in the absence of anti-age discriminationrules? The answer to this question appears to be in theaffirmative for those jobs and occupations requiring firm-specificskills--that is, full knowledge of company policies, operatingrules, personnel, technology, on-going innovations and specificfeatures of the work environment. These job requirements arise inmanagerial positions where staff must learn the nature of thebusiness over several years. These requirements also apply inoccupations requiring special on-the-job training--coordinatingactivities on a construction site, scheduling to run a factorywork-shops etc. On-the-job acquisition of specific skills is alsoneeded in varying degrees from a safety and job-effectivenessperspective, as is the case with air-traffic-controllers, pilots,law enforcement officers etc.
Whenever workers are required to possess "specific humancapital," it is in the employers' interest to ensure that workersdon't quit immediately after acquiring those skills. However,because slavery is illegal, firms must induce workers to stay onthe job by incorporating appropriate incentives in employmentcontracts, which may involve implicit agreements on someelements. For example, workers may be paid less than theirproductivity during the early part of their tenure in exchange for(the implicit promise of) being paid more than their productivityduring the later periods of their tenure with the firm. Thisimplies the creation of "seniority rents."
Several studies have found evidence consistent with theexistence of long-term incentive contracting in the private sector.Overall, the evidence generally points against the "spot" marketexplanation of how compensation generally varys with age. Inparticular, the evidence suggests that compensation exceedsproductivity at older ages providing a basis for employers to treatolder workers differently.
Thus, in occupations requiring firm-specific skills, workers'compensations may rarely, if ever, match their currentproductivity. Instead, the employers seek to match prospectiveproductivity with prospective compensation over the workers entiretenure with the firm.
Apart from wages and salaries, compensation includes pensions,health insurance and other benefits. Employers generally usenon-wage elements of compensation to design work and retirementincentives. Such incentives address multiple firm objectives:Ensuring worker bonding with the firm over long periods; ensuringthat workers do not shirk on the job; inducing older workers toleave the firm at the "right" time, and so on.
Pension vesting and benefit accrual patterns of defined benefitpension plans can be designed to achieve all of these objectives.Vesting rules in DB pension plans generally require workers to bewith the firm for 5 or 10 years before pension benefits begin toaccrue. Pension accruals--the annual additions to the present valueof pension benefits from additional years of work--are alsodesigned to provide early retirement incentives.
Pension accrual patterns produce age-profiles of compensationthat are initially steeper than workers' age-productivity profiles.Pension accrual begins upon vesting and increases sharply at theearly retirement age--usually age 55. The steep increase in pensionaccrual at age 55 arises because the eligibility to retire earlyand collect benefits immediately is associated with a smaller thanfair reduction in benefits--compared to retiring at age 65 withfull benefits. Moreover, delaying retirement beyond age 55 reducespension accruals sharply-possibly making accruals negative.
These features of pension accurals create incentives for workersto retire early-keeping them from collecting seniority rents bystaying on the job for too long. Firms can fine tune theircompensation structures with other non-wage compensation elements,including severance packages. Thus, jobs involving the acquisitionof significant firm-specific skills may exhibit productivity andcompensation patterns such as those shown in Figure 1.
The productivity and compensation profiles of Figure 1 areestimated for male managers based on employment and compensationdata from a Fortune 500 firm with over 300,000 employees. It showsseveral features:
- Managerial workers at this firm are compensated by less thantheir productivity during the early part of their careers andcompensated by more than their productivity later. This inducesworker retention because early quitters lose the "bond" they haveposted with the firm by accepting compensation less than theirproductivity. They also forfeit the opportunity to collectseniority rents later in their careers.
- Managers' productivity is estimated to be hump-shaped,increasing during the early part of their careers but declining atolder ages.
- Pension accrual commences in the 10th year (vesting) and spikesup sharply at age 55.
- Compensation declines gradually after age 55 although it ismaintained above productivity throughout the later phase of amanager's tenure. This excess compensation constitutes theseniority rent mentioned earlier.
Not all workers may be compensated under long-term incentivecontracts. Routine office workers, support staff, sales agents, andso on appear to be compensated on a "spot" basis rather than underlong-term incentive contracts. For example, we estimated annualproductivity and compensation to be aligned more closely forsalesmen in the above firm--as shown in Figure 2.
It should be noted that the adoption of long-term incentivecontracts does not obviate the need to monitor worker performance.Indeed, the threat of being caught shirking and discharged is anintegral part of the incentive structure.
Workers may also prefer contracts with rising compensation byage--if they can obtain them. One possible reason is theirinability or lack of discipline to save -- as is suggested by somepsychological studies of saving behavior. They may prefer toreceive lower compensation in the near term and higher compensationin the future as a forced saving mechanism.2
For long-term incentive contracts to be feasible, however,workers must also be convinced that employers will not arbitrarilydischarge them as soon as they begin accruing seniority rents.
One conjecture is that firms' incentives to "cheat" in thismanner are reduced by the need to maintain their reputations--inorder to continue hiring workers. However, purely reputationaleffects need to operate extremely strongly to effectively policeagainst mid-career job terminations by employers. In addition,evidence suggests that early worker terminations can occur throughother mechanisms -- for example, after hostile takeovers ofcorporations. In one of my studies I find that post-hostiletakeover managements discharge older workers, implying little or nonegative reputational consequences.
The weakness of reputational effects in preventing prematureworker terminations provides a possible rationale for anti agediscrimination laws. A law prohibiting terminations purely on thebasis of age can serve as external, and therefore more credible,commitment mechanism-an external check against the temptation todischarge workers prematurely--and makes long-term incentivecontracts easier to implement.
How effective are long-term incentive contracts in inducingearly retirements? Table 1 contains an answer based on data fromthe firm mentioned earlier. It shows retirement "hazard"rates--that is, the fraction of those employed at the beginning ofthe year that leave the firm within the year. The rates are shownby age and tenure with the firm.
The table suggests that because of the inducement to retireearly (at age 55) provided through the pension accrual pattern,retirement rates step up to the 10-12 percent range between ages 55and 59. Without the retirement incentives, they would remain atabout the 3 percent level that prevails prior to age 55. Note thatthose aged 55-61 who are not yet fully vested in the pension plan(that is, those with less than 10 years of service) exhibitseparation rates around 3 percent annually. Job separation rates at10-12 percent per year rather than 3 percent per year can havesubstantial cumulative effects on overall labor force participationbetween the ages of 55-61. It is noteworthy that job separationrates increase even more dramatically at age of 62 and 65. Theseincreases in retirement hazards probably occur as workers notsubject to long-term incentive contracts respond to the retirementincentives provided by the Social Security program at theseages.
Prior to the 1980s, defined benefit (DB) plans coveredtwo-thirds of workers and defined contribution (DC) plans coveredabout one-third. As is well known, defined benefit pension plancoverage has been declining and defined contribution plan coveragehas been growing during the last two decades. Evidence shows thatDB plans' usefulness has declined for both employers and employeesin an environment of rapid technological and structural changes inthe economy. Under such conditions, the desirability oflong-duration employer-employee matches has declined. Employersattempting to adapt to new technologies may require greaterflexibility in workforce composition. Employees may prefer greaterportability of pension assets if expected job-durations areshorter.3
However, the surge in DC pension plans since the early 1980s hasnot extinguished the use of DB plans: About one-third of theworkforce continues to be covered under DB plans. Moreover, becauseearly retirement incentives are not incorporated into DC plans,retirement rates among those in their early sixties havedeclined--a reversal in the trend established over severaldecades.
In the current context, offerring efficiency enhancingcompensation structures to federal and state and local workerssimilar to those adopted in the private sector appears to bedesirable--to the extent such incentive contracts are not offerredtoday. This recommendation is motivated by the need to elicitworker efficiency, and is independent of the fact that publicoperations are not driven by a profit maximizing motive. Moreover,such incentive compensation structures would provide greaterretirement flexibility and help achieve employers' objectives ofsafety and effectiveness in job performance.
However, any revision of public sector employment contractsalong these lines would require a careful examination of whethersuch compensation structures are feasible, the manner in which theyshould be introduced, and how effective they could be assubstitutes for the mandatory retirement age rules currently inforce.
Section 2: Mandatory Retirement Ages, OccupationalDevelopment, and Personnel Abilities
For the occupations under consideration, adopting a singlemandatory retirement age is not necessarily better or cheaper thanadopting flexible compensation-based incentives to retire.
A fixed retirement age potentially introduces two types oferrors from the perspective of retaining qualified workers. First,those who are less qualified than others would remain on the workforce because they are younger than the mandatory retirement age.Second, those who are better qualified than others are forced toretire because they are older. The mandatory retirement age couldbe set to minimize the sum of both types of errors. For example, ifthe mandatory retirement age were set at 40, we would force manyqualified workers to retire prematurely. Similarly, if theretirement age were set at 80 many workers who are no longercompetent would be retained. These errors would be minimized bysetting the mandatory retirement age between these extremes.
The mandatory retirement age rules in the occupations underconsideration were set several decades ago. Assuming that they wereinitially set optimally to minimize the sum of the two types oferrors-they are probably obsolete today for a number ofreasons.
Improving Health and Fitness
The significant progress achieved in medical innovation andhealth care have increased the longevity of the U.S. population ingeneral.4 People in theirearly sixties today enjoy similar health and lifestyles today withgreater frequency as did those in their mid fifties several decadesago. One indication of the better health of today's workers is thedownward trend in mortality rates.
For example, mortality information from the Social SecurityAdminstration suggests that 55-year-old men in 1960 faced the samelikelihood of dying within the year as do 62 year-old men today.And today's 66 year old men have the same chance of dying as did 60year old men in 1960. For women, mortality improvements aresomewhat smaller. Women aged 55 in 1960 experienced the sameaverage mortality as do 60 year old women today. And the samefraction of 60 years old women died in 1960 as do 64 year old womentoday.
Evidence that health among 50-60 year olds is improving can alsobe gleaned from surveys of self-reported health. Figures 3 and 4show calculations based on the Panel Survey of Income Dynamics(PSID).5 The calculationsreported below are based on weighting each household to convert thesurvey's sample into a representative U.S. householdpopulation.
Figure 3 shows that by 1997, a sizable majority of men and womenwere in good or better health. Over a 14 year period between 1985and 1997, the fraction of men aged 56 through 65 who reported beingin good or better health increased by almost 5 percentage points to71 percent. Figure 4 shows that for women, the share of those ingood or better health increased by about 7 percentage points to67.4 percent.
Although these data are based on self-reported health by surveyrespondents and spouses, a study (based on a different survery)shows that such responses are representative of the type ofinformation used in professional evaluations of health anddisability status.6
These data go back only through 1985. Projecting them furtherback in time would presumably reveal even more substantial gains inthe health and fitness of those in their mid-fifties and earlysixties. Indeed, other studies have indicated that health is nowmuch less important as a consideration for the decision to retirethan was the case several decades ago.7 This is because of both, lower physical jobdemands and improvments in the treatment of chronic conditions.
Finally, the data indicate sizable gains in longevity and healthfor the general U.S. population. I do not have directevidence of similar health gains for the subset of the populationthat forms the base for recruitment into the occupations underconsideration. To the extent that such gains have occurred,revising mandatory retirement ages upward by a few years may befeasible.
Technology Induced Demand and Projected WorkerShortages
The technology used in executing jobs in many of the occupationsunder consideration is much better today compared to 3 or 4 decadesago. The largest improvement has occurred in communications andinformation technology, and in all occupations; firemen have betterheat resistant and fire-retardant materials; pilots have planesthat are easier to fly an land; police officers have betterinvestigative, forensic, and interdiction techiques, better bodyarmour, DNA analysis, computerized laboratories etc. Not only doesnewer technology allow jobs to be executed faster and moreefficiently, they can be executed with lesser exertion ofeffort.
They also call for a workforce with a wider range ofskills--which implies that the availability of new technology isnot necessarily labor-saving overall. It requires more training tooperate and maintain newer equipment and requires more experiencedpersonnel to coordinate job activities. On the other hand, as thebaby-boomer generation retires, many jobs requiring trained andexperienced workers will begin to experience shortages. Those jobswhere retirements are mandatory at younger ages will experienceacute shortages earlier.
Personnel shortages in key jobs that must be executed on timeprovoke the imposition of mandatory overtime. However, forcedovertime over long periods imposes additional burdens and is likelyto lower worker morale. A high-stress atmosphere is likely toinduce additional accelerated retirements and make worker shortageseven more acute. Hence, worker shortages in crucial occupations canbe self-reinforcing unless dealt with in a timely manner. Theprospect of increased shortages because of the impending surge inretirements along with an increasing demand for security andaviation efficieny in a new post-9/11 world makes it necessary torevisit mandatory retirement age rules in the occupations thatcurrently enforce them.
Private sector enterprises get by without the imposition ofmandatory retirement age rules: They successfully hire high skilledworkers under long-term incentive contracts. Indeed, antiage-discrimination laws--the polar opposite of mandatory retirementage rules -- appears more important for making such contractsfeasible. An important element of private long-term incentivecontracts are defined benefit pension plans and other non-wagecompensation to achieve firms' objectives of eliciting workerefficiency and ensuring timely retirements. Similar compensationarrangements could be usefully considered in the public sector aswell, despite the lack of a profit-maximizing objective.
Mandatory retirement age rules in certain private, federal, andstate and local occupations have been in place for several decades.Their revision appears worthy of consideration for severalreasons.
The improvement in general health and abilities of those agedbetween 55 and 65 in general may imply that mandatory retirement atthese ages is unfair for a growing number of workers who retain theability to execute their jobs competently but cannot transfer theirskills to other occupations--such as pilots and air-trafficcontrollers. Evidence suggests that older displaced workers find itmuch harder to find jobs compared to younger workers and sufferlarger wage declines upon re-employment. 8
Better technology makes the conduct of these jobs physicallyless taxing. Moreover, newer technologies are likely to require alarger workforce with a broader set of skills to fill thesepositions. And, the onset of baby-boomer retirements is likely tocreate acute shortages of experienced personnel, especially inoccupations with mandatory retirement set at younger ages.
Summarily eliminating mandatory retirement age rules to preparefor upcoming worker shortages may not, however, be the correctpolicy response. Doing so may create other problems as in the U.S.academic institutions where retirement rates have plummetted aftermandatory retirement at age 70 was abrogated. That has led toslower turnover of teaching staff and aging faculties. Becauselong-term incentive contracts pay off only when introduced at thebeginning of worker careers, implementing such contracts for olderworkers becomes expensive.
Hence, existing mandatory retirement age rules should be revisedin two steps. To deal with impending shortages, existing mandatoryretirement ages could be advanced by a few years. Long-termincentive contracting should be introduced for younger workers andnew hires. The workforce subject to long-term contracting should bedesigned to both provide retirement choices to workers and satisfyemployers' objectives of work-safety and efficiency. The revisedmandatory retirement age rules will be automatically phased outover time as the workers to which they apply retire over time.
1. I am Jagadeesh Gokhale, SeniorFellow at the Cato Institute in Washington D.C. I have conductedstudies on labor market contracting in the private sector and theeffects of long-term employment contracts and worker tenure on themarket for corporate control. I have also written on demographicand retirement issues relating to the sustainability of the federalbudget.
2. R. H. Frank and R. M. Hutchens,1993, Economic Journal, Vol. 21.
3. See Friedberg and Owyang, NationalBureau of Economic Research, Working Paper No. 10714.
4. Frank Lichtenberg "Sources of U.S.Longevity Increase: 1960-1997," National Bureau of EconomicResearch, Working Paper No. 8755, February, 2002.
5. The PSID is conducted bytheUniversity of Michigan's Survey Research Center. This survey'ssample contained just over 10 thousand U.S. households in 1985 andit attrited to about 6,700 households by 1997.
6. Hugo Benitez-Silva, MosheBuchinsky, Hiu Man Chan, Sofia Cheidvasser, and John Rust, NationalBureau Economic Research Working Paper No. 7526.
7. Costa (1994), NBER Working PaperNo. 4929.
8. See David Shapiro and Steven L.Sandell, 1985, Southern Economic Journal, Vol. 52.