"As described in the technical appendix, based on year-to-year changes in national average wage levels, earnings histories were developed leading to the specified pay levels in 1983.

"Employee Benefits in Medium and Large Firms, 1984, p. 11.

Fewer than 1 percent of the participants had plans with floors providing a specified minimum monthly benefit. Twelve percent had ceilings limiting the maximum size of the benefit. These maximums are independent of ceilings imposed by tax laws, which are substantially higher than those specified in the private pension plans examined.

'Step-rate excess formulas provide a way of integrating private and Social Security benefits. See Donald Bell and Diane Hill, "How social security payments affect private pensions," Monthly Labor Review, May 1984, pp. 15-20.

10 According to the Bureau's 1984 employee benefits study, 92 percent of professional-administrative participants, 86 percent of technical-clerical participants, and 46 percent of production participants were covered by earnings-based pension formulas. See Employee Benefits in Medium and Large Firms, 1984, table 39.

11 Coming of Age: Toward a National Retirement Income Policy (President's Commission on Pension Policy, February 26, 1981), pp. 42-43. Earlier estimates are in Peter Henle, "Recent trends in retirement benefits

related to earnings," Monthly Labor Review, June 1972, p. 18; and Jane L. Ross, Maintenance of Preretirement Standards of Living After Retirement, Technical Analysis Paper No. 10 (Office of the Assistant Secretary of Planning and Evaluation, Department of Health, Education, and Welfare, 1976).

12 These alternatives parallel the varying definitions of earnings found in earnings-based pension benefit formulas. See Employee Benefits in Medium and Large Firms, 1984, tables 39 and 41.

13 For recent discussions of the replacement rate concept, see Michael J. Boskin and John B. Shoven, Concepts and Measures of Earnings Replacement During Retirement, Working Paper No. 1360 (Cambridge, MA., National Bureau of Economic Research, 1984); and Congressional Research Service, Designing a Retirement System for Federal Workers Covered by Social Security, 98th Cong., 2d sess., Committee Print 98-17 (Committee on Post Office and Civil Service, House of Representatives, 1985), pp. 305-15.

14 See footnote 10.

15 See Bell and Hill, "How social security payments affect private pensions."

16 See Coming of Age, pp. 12-14.

This study of pension benefit levels follows one of a number of alternative approaches to examining private pension plan provisions. A common approach is to review individual plan provisions, such as vesting requirements, early and normal retirement ages, benefit formulas, and pre- and post-retirement survivor options.' This approach provides a wealth of detail about plan provisions but does not permit summarization on an overall plan basis.

Such summarization is possible through examination of amounts employers spend on funding their pension liabilities, either in terms of dollars per employee per year, cents per hour worked, or percent of total compensation outlays.2 Employer cost levels, however, are commonly influenced not only by plan provisions, but also by such characteristics of the covered work force as age, length of service, and earnings history and the actuarial assumptions used in financing individual plans.3

The approach used here looks at the level of benefits available under plans in effect in 1984. It focuses on the pensions payable to workers retiring on January 1, 1984, under the latest (current service) benefit formulas of their pension plans at that time.

Aside from the pension formula itself, retirement benefits may be affected by possible coordination of private benefits with Social Security payments, limits on years of credited service, and minimums and maximums on benefits. These were taken into account in calculating retirees' pensions for this analysis. Also, many plans had more than one pension formula, and specified use of the formula providing the highest benefit for each worker's circumstances. When multiple formulas were found, each alternative within a plan was examined and, for each combination of years of service and earnings considered for study, the formula selected was the one yielding the highest pension.

Nevertheless, the study did not take account of all factors affecting a retiree's pension. For example, it did not consider benefit reductions to finance continuation of payments to a surviving spouse (joint-and-survivor annuity). Similarly, the possibility of post-retirement pension increases-either on an ad hoc basis or through an automatic cost of living adjustment formula-was ignored.

After determination of the pension benefits under individual plans, overall averages were computed. In computing these averages, individual plans were weighted by the number of active workers participating in each plan.

Benefits under a given pension plan are influenced by retirement age, length of service with the firm, and earnings history. It is, therefore, necessary to specify values for these variables to determine retirement benefits. One approach is to assume average conditions prevailing throughout the economy-average retirement age, average seniority, average earnings. This approach, however, ignores the fact that benefit formulas in individual pension plans are influenced by the characteristics of the workers that they cover."

Consequently, in the approach followed here, age 65 was chosen as the assumed retirement age because all workers are entitled to their fully accrued benefit at that age under the Employee Retirement Income Security Act. (Sixty-three percent of the participants in the pension plans studied, however, were under plans which allowed for full retirement with an unreduced pension before age 65.)

Instead of using a single assumption regarding the employee's length of service and earnings history, the multiple assumptions shown on the tables were used. The earnings levels specified represent the employee's gross earnings in the final year of work (1983). Earnings levels in each year from 1944 to 1983 were then developed from these final earnings using year-to-year changes in Social Security data

on national average wage levels."

The same final earnings levels and earnings histories were used for all three occupational groups studied-professional-administrative, technical-clerical, and production workers. Nevertheless, some of the final earnings levels presented would not have wide applicability in each occupational group. For example, it is unlikely that many technical-clerical workers in medium and large firms had final earnings as high as $40,000, nor is it likely that many professional-administrative workers had final earnings as low as $15,000 in 1983. Because pension benefit formulas are often designed for a specific group of workers with a known range of earnings, some distortion in benefits at

unlikely earnings levels is possible. Thus, when examining the results of this analysis, the focus should be on the benefits provided at earnings levels applicable to a particular occupational group.

Social Security benefits are important not only as a source of retirement income but also as a factor affecting benefits under many private pension plans. For example, a common approach to integrating private and public annuities is to reduce private pensions by a percentage of Social Security benefits. To estimate benefits under the Social Security system, it was assumed that an employee worked in covered employment for a total of 40 years.8

1See, for example, Employee Benefits in Medium and Large Firms, 1984.

2 Such data were developed in the Bureau's survey of employer expenditures for employee compensation, which has been discontinued. See, for example, Employee Compensation in the Private Nonfarm Economy, 1977, Summary 80-5 (Bureau of Labor Statistics, 1980).

3 Differences in labor force characteristics and actuarial assumptions may be accounted for by estimating what it would cost to provide surveyed pension plans to a standardized work force, using uniform actuarial assumptions. For an illustration of this approach, see Total Compensation Comparability: Background, Method, Preliminary Results (Compensation Group, United States Office of Personnel Management, 1981).

"Sample weights assigned to each surveyed establishment were also applied to provide representation of all establishments covered by the survey, not only those providing data. The resulting averages are measures of benefits payable under assumptions discussed in the remainder of this appendix. They are not, however, measures of average benefits actually

being received by retirees. For such measures, see Linda Drazga Maxfield and Virginia P. Reno, "Distribution of Income Sources of Recent Retirees: Findings From the New Beneficiary Survey," Social Security Bulletin, January 1985, pp. 7-13. Also see Findings From the Survey of Private Pension Benefit Amounts (Office of Pension and Welfare Benefit Programs, U.S. Department of Labor, 1985).

"Furthermore, average earnings of all workers are considerably less than the average for full-time employees nearing the retirement age. See Alicia H. Munnell, The Economics of Private Pensions (Washington, Brookings Institution, 1982), pp. 25-27.

"See Social Security Bulletin, Annual Statistical Supplement, 1983, p. 28. "See Bell and Hill, "How social security payments affect private pensions."

Actually, for retirees in 1984, the measuring period used to determine Social Security benefits would be the same for individuals with 25 years of service or more. See Robert Myers, Social Security (Homewood, Ill., Richard D. Irwin, Inc., 1981), pp. 54–55.

The term "three-legged stool" is com

T

monly used to emphasize that retirement income derives from three primary sources: Social Security, employer-sponsored retirement plans, and workers' savings. If workers retire before age 62, however, the first leg of the stool, Social Security payments, is missing. To compensate, some private pension plans provide extra payments until a retired worker is eligible for this Government benefit.

According to the Bureau of Labor Statistics 1988 Employee Benefits Survey of medium and large firms in private industry, 1 in 8 defined benefit pension plan participants was in a plan that provided such supplemental payments. These special benefits usually continued until the retiree reached age 62, the earliest age at which regular Social Security benefits may commence.1 Benefit payments ranged widely, but most commonly were a uniform dollar amount for all plan participants regardless of salary or length of service.

The 1988 survey studied full-time employees in a sample of 2,500 establishments, which represented approximately 107,000 establishments employing 31 million full-time workers. Separate data were developed for three broad occupational groups: professional and administrative, technical and clerical, and production and service workers. The first two groups are referred to collectively as white-collar workers, the third as blue-collar workers.2

Retirement ages

The ages at which employees may begin to receive retirement benefits from employersponsored pension plans and from Social Security frequently do not correspond. To understand these age differences, it is useful to contrast the benefit approaches of private plans with those of Government programs. In 1988, 63 percent of employees in medium and large firms in private industry participated in defined benefit pension plans.3 A defined benefit pension plan specifies a formula (for example, a percent of the employee's earnings or a dollar amount for each year of service) and also specifies age and length of service requirements that must be met before an employee is eligible for either normal or early retirement benefits. Normal retirement benefits are the full annuities yielded by the pension plan's benefit formula, without reduction due to age at retirement. Early retirement benefits are systematically reduced because they begin at an earlier age and, on average, will be received over a long period.

Traditionally, most private pension plans adopted age 65 as the normal retirement age. In 1969, 69 percent of workers in pension plans had to work until age 65 to receive normal retirement benefits, and only 9 percent could receive such benefits prior to age 60.4 This corresponds with Social Security, which, at its

inception, established 65 as the age for full benefits. The age for receipt of full Social Security benefits is currently 65, although the required age will rise beginning in 2003, as discussed later.

More recently, private pension plans have lowered the age at which retired workers may receive unreduced normal benefits. In 1988, nearly 60 percent of the participants of defined benefit pension plans could retire before age 65 with full benefits.

In addition, early retirement provisions covered nearly all participants of defined benefit pension plans in 1988. Age 55 is the most common minimum age for early retirement, and almost all participants can retire with reduced benefits by age 60. Furthermore, some employers offer incentives to retire early by moderating the reductions in benefits imposed upon early retirees; fewer than a fifth of pension plan participants now face a full actuarial reduction in annuities.S

While these data suggest that private pension plans tend to encourage earlier retirement, Government policy appears to be just the opposite. In 1985, the Federal Government outlawed mandatory retirement for most workers. The next year, in response to ongoing court challenges, the Congress required employers to

credit years of service worked after age 65 toward future pension benefits. And, currently, the Congress is debating removing the limit on income that can be earned by recipients of Social Security, a restriction which, many believe, discourages older Americans from continuing to work.8

In addition, future Social Security policy will encourage people to work longer and retire later. Presently, full Social Security benefits may be received beginning at age 65, and reduced benefits at age 62. However, beginning in 2003, the normal retirement age will climb 1 month per year, reaching age 67 in 2027.10

The earliest age that Social Security retirement benefits may be received will remain at 62, although benefit reduction for early receipt will increase. Currently, benefits are reduced 63 percent per year for receipt prior to the normal retirement age of 65. This yields a 20percent reduction at age 62. Beginning in 2003, the reduction will be 63 percent per year for the first 3 years and 5 percent per year beyond 3 years. Once the normal retirement age reaches 67 in 2027, the age 62 benefit will be reduced 30 percent. In addition, the increase in benefits provided to those who postpone receipt of Social Security until after normal retirement age will rise gradually from the current 3 percent per year to 8 percent per year.

Supplemental private pensions

In 1988, 12 percent of participants were in private defined benefit pension plans that provided supplemental payments, generally designed to augment benefits received prior to receipt of Social Security. Production and service workers were the most frequent beneficiaries of such provisions, in part, because of the prevalence of supplementary benefits in several large collectively bargained plans. The following tabulation shows the percent of full-time participants in defined benefit pension plans by provisions for supplemental payments in medium and large firms in 1988:

Professional

[In percent]

Method

All Professional Technical Production particiand and and pants administrative clerical service

Normal retirement

Total

100

100

100

100

Flat dollar amount

41

35

38

Dollar amount per year of service

25

33

28

Total benefit equals a flat dollar

amount1

Supplemental payment provisions are about evenly divided between those that supplement normal retirement benefits and those that augment early retirement benefits. The approach of a specific plan generally reflects the age at which it makes available normal and early retirement benefits. For example, if a plan offered normal retirement at age 62 and early retirement at age 55, the supplement would apply to the early retirement benefit because Social Security is available at age 62. If, however, normal retirement were available at age 60, the supplement might apply to normal retirement. It is also possible to have payments supplementing both normal and early retirement benefits.

Requirements. Plans may impose special age or length-of-service requirements, or both, on eligibility for supplemental payments. Therefore, a person retiring with full benefits (but still too young to receive Social Security) may receive a supplement while a person retiring with reduced early retirement benefits (and also too young to receive Social Security) may not. Seventy percent of participants in plans supplementing normal benefits did not have to meet special eligibility requirements for the supplement; the remaining 30 percent, however, faced age or service reqirements more stringent than those for normal benefits.

Additional requirements were more common when supplemental payments were tied to early retirement benefits. For example, one large manufacturing plan offers early retirement to workers at any age with 30 years of service, or at age 55 with 10 years of service. The early retirement supplement is available to all employees with 30 years of service, but is available to employees age 55 only if their age plus service totals 85. (This means that, essentially, the supplement is limited to employees with 30 years of service.) The following tabulation examines whether plans impose additional requirements to receive supplemental benefits at early retirement-it shows the percent of participants in plans that impose the same requirements or additional requirements for receipt of supplemental benefits as for early retirement

benefits:

percent of the retiree's earnings or a flat dollar amount for each year of service, all of the supplemental payments found in the survey were expressed as dollar amounts, either flat amounts or amounts per year of service. (See table 1.) Prior earnings, therefore, do not affect supplemental payments. There are two reasons: first, many of the supplemental payments are part of unionized blue-collar plans, which typically specify benefits as a dollar amount per year of service. Second, the Social Security payments, for which the supplements are a substitute, decline as a percentage of salary for higher-paid workers, and are limited to a maximum dollar benefit. To base a supplement on earnings could

provide payments to higher paid employees beyond that which Social Security combined with private pensions would pay.

Supplemental payments were expressed in several ways. Payments could be a flat amount, such as $300 per month, or an amount times years of service, such as $5 per month times years of service. Alternately, benefits were expressed as a guaranteed total benefit, the sum of the basic payment and the supplement. For example, the supplement, when added to the basic benefit, will yield a monthly payment of $1,000, or a monthly payment of $25 times years of service.

While the incidence of supplemental payments among survey respondents was insufficient to allow an extensive survey of benefit amounts, some typical payment amounts can be used to examine the extent to which supplements substitute for Social Security. Flat dollar amount supplements typically ranged from $200