Ending Mandatory Retirement for Tenured Faculty: The Consequences for Higher Education (1991)

The higher education groups who favored an exemption from ADEA were concerned that many tenured faculty would choose to work well into their eighth decade if permitted to do so. They were also worried that delayed retirements might lead to low faculty turnover. One administrator responded to our letter: "While no institution wishes to lose talented faculty, turnover through retirement . . . does allow for the infusion of new ideas and energy into an institution." Another recognized that a faculty member rarely switches fields, so that retirements create opportunities for colleges and universities to reallocate positions across departments:

In order to address these concerns, three questions are central to analysis of the effects of uncapping on faculty and on colleges and universities.

1. Would some faculty would work past the current mandatory retirement age of 70 if they could? Since most colleges and universities now require tenured faculty to retire at 70, we examined historical information about faculty demographics and retirement behavior, supplemented by data from a few colleges and universities that have recently eliminated mandatory retirement. In answering these questions, we evaluated data on the number of faculty nearing retirement age (approximately 60–70 years) in the next few years and evidence pertaining to the proportion of this group likely to postpone retirement past age 70.

2. Are faculty in some types of colleges or universities more likely to continue working into their 70s if permitted to do so? How would this

affect average retirement ages at those institutions? Some administrators and faculty reported that at certain institutions, such as the research universities, faculty are more likely to continue working past age 70. We evaluated evidence on faculty age structure and retirement patterns in a variety of institutions.

3. What would the major effects be on colleges, universities, and higher education, in general, if faculty worked past age 70? Many faculty and administrators believe that to stay in the forefront of scholarship it is important to hire new faculty. As McPherson and Winston (1988:183) note:

For example, a historian is unlikely to be a productive teacher and researcher in particle physics. At universities emphasizing research, specialization may be of even greater importance: An elementary particle physicist is unlikely to switch specialties easily or rapidly to high-temperature superconductivity. In some areas, however, faculty skills may be broader: Some introductory science courses could be taught by faculty in related disciplines. The degree of institutional change resulting from a retirement followed by a new hire depends on whether institutional policy allows departments to refill vacated positions or whether openings are transferred across departments. However, at both the departmental and the institutional level, turnover creates opportunities to bring in new faculty.

A rise in the average retirement age for current tenured faculty would reduce turnover, thereby limiting the number of tenure or tenure-track positions available for new faculty. Such a rise could be caused by an increase in the number of faculty working past age 70 after the elimination of mandatory retirement or by a large number of faculty retiring later than they do now, even if few or none waited until after age 70. We projected the potential effects of postponed retirements on colleges' and universities' resources, including their budgets and ability to hire new faculty. We also considered how our findings on these issues might be related to other research that projects a disproportionate number of future faculty retirements in certain age groups and the possibility of a future nationwide faculty shortage.

In addressing the above questions, we drew on a number of studies, sparked by the 1977 and 1986 changes in the ADEA, that analyzed faculty retirement behavior. We also considered information from national faculty data bases as a check on published faculty age structures and trends. We used information from the committee's letter of inquiry and case studies to identify potential problem areas and to illustrate the findings and conclu-

sions suggested by more systematic studies. Finally, we obtained data on recent retirement patterns from some universities at which our research suggested high proportions of faculty might work past age 70 if mandatory retirement were eliminated.

In piecing together this mosaic of existing research and data, the committee found a reasonable basis for making certain inferences about retirement patterns over the coming decade. In this chapter we report our findings on faculty demographics and retirement behavior for higher education as a whole. We also examine the evidence for claims of more severe effects on faculty hiring and budgets at some universities and in some disciplines.

The committee examined the relatively sparse evidence on faculty who choose to work past age 70, as well as the more extensive evidence on the number of faculty now working past age 65. In this section we concentrate on evidence of faculty age distributions and retirement patterns in higher education as a whole; in the next section we examine variations in retirement patterns by institutional type.

Future retirement trends depend partly on present age distributions, that is, the number of faculty who will be old enough to consider retirement (e.g., age 60 or older) at any given time. Therefore, we first looked at the age distribution of faculty and how it has changed over the past decade.

Estimates of the age distribution from three national samples of faculty members are remarkably similar (see Figure 1). The longitudinal Survey of Doctorate Recipients (SDR), conducted every 2 years by the National Research Council, includes information on the ages of faculty members with doctorates. Data from cross-sectional surveys—the U.S. Department of Education's 1988 National Survey of Post-Secondary Faculty (NSOPF) and the 1989 survey of faculty conducted by the Higher Education Research Institute (HERI) at the University of California at Los Angeles—include faculty members with and without doctoral degrees (see Appendix B for details on the three surveys, including a discussion of methodological issues and suggestions for future survey research).

The committee was cautious in its use of these survey data. For example, in examining the SDR, we learned that the unweighted numbers of respondents in the 65+ and 70+ age categories are too low as a proportion of the estimated population to support detailed projections even from weighted data (see Appendix B). This limitation prevents us from calculating faculty

Figure 1

Faculty age profiles from three surveys. Note: Summary of Doctorate Recipients (SDR), National Survey of Postsecondary Faculty (NSOPF), Higher Education Research Institute (HERI).

retirement rates using SDR data; however, the consistency between the age distributions found by the three surveys allows us to draw some conclusions.

Although the three surveys used different sampling techniques and sample sizes, all three data sets show that less than 5 percent of the faculty members are age 65 or older. Less than one-half of 1 percent of the faculty in the SDR and NSOPF samples are over age 70. The data do not allow us to estimate the exact number of current, retired, and deceased faculty over age 65, but they do suggest that large numbers of faculty begin to retire around age 65 and that most retire before the current mandatory retirement age of 70.

Over the past decade, the age distribution of faculty with doctorates (see Table 1) shows an increase in the average age of faculty. The current age distribution suggests that an increasing proportion of faculty will be approaching retirement over the next two decades, with the largest age group of current faculty entering their 60s in a little more than 10 years. Table I shows that the percentage of faculty age 65 or older increased from 2.1 to 4.0 from 1979 to 1989; this percentage change represented about

5,000 faculty members. Given the increase in the mandatory retirement age from 65 to 70 for the period from 1978 to 1982, the number of faculty aged 65 or older is still small. This suggests that most faculty members choose to retire before age 65.

In order to estimate the number of faculty who will work past age 70, one needs to know not only the number of retirement-age faculty but also how likely those faculty are to retire at earlier or later ages. The national faculty surveys provide some evidence on the number of faculty at given ages to continue working past age 70, but their data do not support calculations of the proportion of faculty at any given age who retire—that is, of retirement rates. Furthermore, most studies of faculty retirement concentrate on average retirement ages rather than the proportion of faculty retiring at higher than average ages. Studies of faculty retirement behavior do, however, cast light on the propensity of older faculty to continue working. Some faculty members who retired at age 70 might have worked longer had they been permitted to do so.

In contrast to recent national trends toward earlier retirements, college and university faculty median retirement ages have not decreased (Burkhauser and Quinn, 1989:66). However, the available data and research results suggest that few faculty have chosen to work until age 70 or older. One limitation to these data and research reports is that most faculty retirees are white males. If women or minority faculty have significantly different retirement patterns, future faculty retirement patterns will change to reflect this. Table 2 shows the age distribution of all U.S. faculty in higher education.

Furthermore, attitudes about an ''appropriate'' retirement age could change in the future, perhaps in response to some of the factors that affect retirement behavior (e.g., economic conditions, health care benefits, and state of health). Existing data show that retirement patterns do not change swiftly and may proceed at a generational pace (Burkhauser and Quinn, 1989) People who have worked for several decades with the expectation that they would retire around age 65 or 70 may be less likely to change their expectations than those just starting careers.

Some states and individual colleges and universities have already eliminated mandatory retirement for tenured faculty. Figure 2 shows the status of uncapped public and private institutions at the time of this report's publication. More than one-third of the states have eliminated mandatory retirement for tenured faculty in public colleges and universities, and some states have also eliminated mandatory retirement in private colleges and universities. In addition to institutions uncapped by state law, some public and private institutions have independently decided to uncap. Although most uncapped colleges and universities have eliminated mandatory retirement during the past 3 years, public higher education systems in three states—Florida, Maine, and Wisconsin—have been uncapped for long enough to

provide more than 1 or 2 years' data on changing faculty retirement behavior. In two of those states—Maine and Wisconsin—state law also uncapped private colleges and universities, most of which are liberal arts colleges.

Given the limited national survey data on the number of faculty over age 70, the experiences of uncapped colleges and universities provide the only available direct information on faculty retirement ages in the absence of a mandatory retirement age. The committee requested data on faculty ages and retirement ages from state higher education systems and state retirement systems in uncapped states, and it conducted case studies at public and private uncapped institutions. In all, we found few faculty chose to continue working past age 70, although faculty retirement choices at many colleges and universities may have been affected by the introduction of retirement incentive programs as well as by uncapping.

• In Florida the average retirement age for all university employees (the state retirement system cannot separate data on faculty) has remained remarkably stable at around age 63 since the state eliminated mandatory retirement in 1976. Data on the average retirement age of tenured faculty at one institution, the University of Florida at Gainesville, show that annual

average retirement ages from 1972 to 1989 varied from 61 to 64, with no upward trend over the period. The university reports that 1.6 percent of the faculty are aged 70 or older.

• The University of Maine system benefits coordinator reported that the average faculty retirement age has been between 61 and 63 both prior to uncapping in 1978 and subsequently. Only 6 of the 1,370 faculty (0.4 percent) are over age 70.

• The University of Wisconsin could not provide longitudinal data, but data from the Madison campus on the ages of the 97 faculty members who have retired since February 1989 show that the average faculty retirement age was 65 and that 14 of the retirees (14 percent) were aged 70 or over. Of the 2,368 faculty members, 26 (1.1 percent) are over age 70.

On the basis of their study of the retirement patterns of tenured arts and science faculty at 19 public and private universities and 14 private liberal arts colleges, Rees and Smith (1991) identified several factors that may explain differences in mean retirement ages among institutions. The presence of a mandatory retirement age was not one of them. In fact, for their sample of uncapped liberal arts colleges, the mean retirement age of tenured faculty is 1 year lower than the mean retirement age at capped institutions. Rees and Smith also found that the mean retirement age at uncapped colleges did not change after the end of mandatory retirement. Of course, a constant mean retirement age at an uncapped college could mask later retirement of some faculty offset by earlier retirements of other faculty.

The few colleges and universities that have been uncapped for a long enough period that faculty could have continued working until their late 70s or early 80s report that few individuals have taken advantage of this opportunity. The committee's case studies of uncapped colleges and universities found that only one or two faculty in uncapped institutions have stayed past age 73. The oldest two retirees in the University of Wisconsin-Madison data were 74, and, as noted above, only 1.1 percent of the faculty are over age 70. The proportion of faculty over age 70 is small (no more than 1.6 percent), even at colleges and universities that have been uncapped for over a decade. We note that none of these uncapped institutions is a private research university. Johns Hopkins University, the only private research university uncapped at the time of our study, stopped enforcing mandatory retirement when the ADEA amendments passed in 1986 and formally uncapped in 1989. Of their 1,974 faculty, 4 (0.2 percent) are over age 70.

In 1978 Congress required colleges and universities to raise the mandatory retirement age for tenured faculty from 65 to 70 by July 1, 1982.

Changes in retirement patterns over this period can provide some insight into the proportion of faculty choosing to postpone retirement when they have the opportunity to do so, but one cannot simply compare average faculty retirement ages before and after 1982. Although the 1978 ADEA amendments set a deadline for raising the mandatory retirement age, a 1980 study of 278 institutions found that prior to the 1978 amendments, one-third of all institutions already had no mandatory retirement age or a mandatory retirement age higher than 65. Several states raised or abolished the mandatory retirement age prior to the 1982 deadline. Furthermore, institutional retirement policies were apparently flexible: At colleges and universities with a stated mandatory retirement age of 65, 40 percent of the faculty members reaching age 65 continued to work (Holden and Hansen, 1989:38).

In their study of faculty retirement ages at 101 colleges and universities, Lozier and Dooris (1990:14) found that the overall average retirement age at all institutions was less than 65: It was age 63.8 at institutions that raised the cap in 1982 and age 64.3 at institutions with a mandatory retirement age of 70 during the entire study period of 1981–1988. Another study of retirement patterns at 36 colleges and universities (mostly private) found that the average faculty retirement age increased by slightly less than 1.5 years, from age 64.6 to age 66.0, from 1982 to 1986 (Consortium on Financing Higher Education, 1987).

Nationwide, faculty retirement patterns have remained fairly stable for the last 15 years despite a major change in retirement law between 1977 and 1982. At the uncapped institutions for which there are data on faculty ages, the proportion of faculty over age 70 was less than 1.6 percent. The observed faculty retirement behavior may be influenced by factors other than uncapping. For example, many of these institutions offer retirement incentive programs. Data from case studies of institutions uncapped for more than 3 years show that of the few faculty who have chosen to work past age 70, almost all retired by age 73.

On the basis of our consideration of the available data and studies, we conclude: Most faculty do not choose to work until age 70, although they have the opportunity to do so, and, overall, only a small number of the nation's tenured faculty will continue working in their current positions past age 70.

National averages and the experiences of individual colleges and universities cannot describe or predict variations in faculty retirement behavior

across more than 3,200 institutions of higher education or by fields. Therefore, in this section we examine evidence on whether faculty retirement patterns differ by types of institution. In order to make our examination of institutional variation more manageable, the committee used a set of broad institutional classifications constructed by the Carnegie Foundation for the Advancement of Teaching. These classifications divide colleges and universities based on whether they are public or private, on enrollment, on research spending, on number and types of degrees awarded, and on range of subjects offered. There are six broad categories (see Appendix C for details):

• research universities—about 100 public and private universities that offer the widest range and level of degrees, including at least 50 Ph.D.s annually, and with at least $12.5 million in annual research support;

• doctorate-granting universities—more than 100 universities that offer a wide range of subjects and degrees, including 20 or more Ph.D.s annually in one discipline or 10 or more Ph.D.s annually in three or more disciplines;

• comprehensive colleges and universities—about 600 public and private institutions that offer a wide range of degrees up to the master's level (some also award a few doctorates) and that enroll at least 1,500 students;

• liberal arts colleges—more than 550 colleges that award mostly bachelor's degrees;

• 2-year colleges—nearly 1,400 colleges, three-quarters of which are public, that offer Associate of Arts degrees and adult training in a wide range of fields; and

• specialized institutions—more than 600 institutions that offer degrees in one or two specialties, such as the traditional professions (e.g., law, medicine) and fine arts.

The number of faculty in any institution or field who would work past age 70 if allowed to do so can be discussed in terms of the number of faculty reaching age 70 and the proportion who would choose to keep working. The age distribution of faculty by selected subgroups reveals that some institutions and fields face a more immediate increase in the number of faculty nearing traditional retirement ages (60–70 years). Others face an increase several years in the future, and still others are likely to have a consistent number of faculty reaching retirement age over time. Figures 3 and 4 depict the variations in the age distribution of faculty with doctorates by type of institution and by selected field of study, based on the 1989 SDR data.

In comparison with other institutional categories, comprehensive colleges and universities have a higher proportion of faculty aged 45–55 so that a greater proportion of their faculty members will reach retirement age (60 or older) in 10–20 years. Liberal am colleges and doctoral universities appear to follow the overall age distribution for higher education. Research

Figure 3 Faculty age profiles by type of institution.

universities have a higher percentage of faculty in the youngest age groups (under age 40) and the oldest age group (over 65 years) and a correspondingly lower percentage of faculty members in the middle years.

Fewer data are available on the age distribution of faculty at 2-year colleges: The SDR is not an appropriate source because approximately 75 percent of community college faculty do not have doctorates so other sources must be sought. The Commission on the Future of Community Colleges (1988:5) of the American Association of the Community and Junior Colleges reported that total enrollment in 2-year colleges grew by 240 between 1965 and 1975, and James Palmer (Center for Community College Education, George Mason University) reported (private communication) that a high proportion of the current faculty were hired at that time. The commission (1988:12) also found that "the average full-time community college faculty member is 50 years of age," and it estimated that approximately 40 percent of all community college faculty would retire by the year 2000.

These sources, as well as our case studies of individual community colleges, suggest that few community college faculty retire later than age 65 and almost no one retires after age 70.

Faculty age profiles also vary by field. For example, faculty in computer science are younger on average than faculty in other disciplines. Only a relatively small proportion of computer science and medical faculty members with doctorates will be approaching retirement over the next 15 years. In contrast, chemistry has a smaller-than-average percentage of faculty aged

35–45 and a comparatively larger proportion of faculty approaching retirement age during the next two decades. In the humanities relatively large proportions of the faculty are in the middle and older age groups.

Variations in the age distribution of faculty by type of institution and by field do not indicate what proportion of these faculty will retire at any specific age; rather, retirement rates need to be estimated in order to project the numbers of faculty who would work past age 70. The numbers of older faculty in the national data bases and from other studies are too small to divide further into subgroups. Therefore, some researchers have applied retirement rates from statistics on all faculty to the age distributions of faculty in individual fields and types of institutions.

With a constant retirement rate, the numbers of retirements depend on the number of retirement-age faculty. For example, Bowen and Sosa (1989:198) apply retirement rates drawn from a 1987 Lozier and Dooris survey of 20 universities to data from the 1977 and 1987 SDRs to project future retirements for arts and sciences faculty. They project that greater proportions of faculty at comprehensive institutions and research universities—the two types of institutions with higher numbers of faculty in the middle and older age groups—will retire in the next 10 years.

In the most recent study of this type, Lozier and Dooris (1990) used data on faculty retiring from 101 4-year colleges and universities to project that, by the academic year 2002–2003, the number of faculty retiring annually will be 25–40 percent higher than in 1987–1988. By field, they project that the number of humanities faculty who retire will peak during the period from 1988 to 1993, the number of life sciences faculty who retire will increase through the mid 1990s and then return to current levels, the number of mathematics and physical sciences faculty who retire will increase modestly but steadily throughout the period, and the number of behavioral and social sciences faculty who retire will decline during the mid to late 1990s and then return to current levels (Lozier and Dooris, 1990:58). Using Lozier and Dooris's earlier (1988) higher estimates of faculty retirement rates, Bowen and Sosa (1989) projected similar variations in numbers of retirements by field.

The number of faculty who retire (or do not retire) in a given discipline or type of institution depends not only on the number of faculty old enough to retire but also on the propensity of faculty in that discipline or type of institution to postpone retirement past age 70. These rates may not be constant, as assumed above, but the numbers of older faculty in available studies are too small to support calculations of retirement rates by fields. (The medical school faculty register of the Association of American Medical Colleges was large enough, but it was not structured to support analyses of changing faculty age distributions or retirement patterns.) The data we were able to collect from individual colleges and universities show varia-

tions in faculty retirement ages among different disciplines, but no one discipline or set of disciplines had consistently different retirement ages.

There is, however, evidence that faculty at some types of institutions are more likely to postpone retirement. Lozier and Dooris (1990) used data from a survey of more than 500 retired faculty from 101 colleges and universities, divided into liberal arts, comprehensive, and a category combining doctoral and research universities. They found an average retirement age of 64.3 for faculty at doctoral and research universities, 63.5 for liberal arts colleges, and 63.6 for comprehensive institutions (1990:17). However, combining research and doctoral universities could mask differences between those categories. A survey of faculty by the Carnegie Foundation (Carnegie Foundation for the Advancement of Teaching, 1989:87) found that faculty at research universities were least likely to report looking forward to retirement: 69 percent in comparison to 75 percent for all faculty.

Evidence on retirement patterns at individual colleges and universities also suggests that faculty at research universities retire, on average, at later ages. Rees and Smith (1991:21–22) found that the mean age at retirement was 1.45 years higher for research university faculty than for faculty at the other universities and liberal arts colleges in their sample of 33 colleges and universities. In all, 35.1 percent of the faculty who retired from capped private universities (including three universities not classified as research) did so at the mandatory age. At capped public universities, 17.8 percent of the faculty retired at age 70 or later; at uncapped public universities, 25.1 percent of the faculty did so. In comparison, few faculty at the liberal arts colleges in the Rees and Smith sample (all private) retired at or above age 70:4.3 percent at uncapped colleges and 11.5 percent at capped colleges. The Consortium on Financing Higher Education (COFHE) (1987:13) found that, at the 36 colleges and universities it surveyed, the percentage of faculty members retiring at age 70 between 1982 and 1986 varied from 0 to 83 percent. At 7 of the 21 universities that provided data to COFHE—Harvard, Princeton, Stanford, the University of Chicago, the University of Pennsylvania, the University of Toronto, and Yale—more than 40 percent of the faculty who retired did so at age 70. In comparison, only one of the seven liberal arts colleges in the COFHE sample—Williams College—had more than 40 percent of its retirees stay to the mandatory age.

The percentage of faculty who retire at age 70 may be more significant for projecting the number of faculty who will postpone retirement than data on average retirement ages. Lozier and Dooris (1990:44–47) report that 80 percent of the retirees in their study said that the mandatory retirement age was not important in their retirement decision. However, 87 percent of the faculty who retired at age 70 said that, without mandatory retirement, they would have continued working, on average, to age 74.5.

Retirement age differences between institutions may be explained by

differences in mission and student quality. Rees and Smith (1991) used regression analyses to predict the age of faculty retiring at or above age 70 for their sample of colleges and universities. They found quality of the student body (measured by average Scholastic Aptitude Test [SAT] scores) is the strongest predictor of delayed retirement, followed by low teaching loads and holding research grants. Rees and Smith suggest that faculty are more willing to continue working when they have good students, low teaching loads, and a funded research agenda.

Faculty tend to retire later at private colleges and universities than at public colleges and universities. Lozier and Dooris (1990:17) found that the average retirement age for faculty from private institutions was 65.3, compared to 63.5 with faculty from public institutions. COFHE (1987) found that 31 percent of the faculty who retired from private institutions did so at the mandatory age, compared with 20 percent for faculty from public institutions. However, some of these differences may be due to variations in teaching loads, student quality, and levels of faculty retirement income rather than to whether the institution is public or private. Rees and Smith discovered no differences in retirement patterns between the public and private research institutions in their more homogenous sample.

The committee requested data on the age of faculty who retired at age 70 or older from universities with the highest reported research and development expenditure, divided where possible by schools within the universities. Some other colleges and universities also provided us with these data. These data, like those from the 1987 COFHE study, show that there is considerable variance in retirement behavior within as well as among institutional types (see Table 3). Most of the few uncapped universities were not able to provide data on their faculty retirement patterns. Ten institutions—the University of California system, the University of Chicago, Columbia University, Duke University, Harvard University, Massachusetts Institute of Technology, the University of Pennsylvania, Princeton University, Stanford University, and Yale University—reported that more than one-third of their faculty who retired did so at age 70 or older. The University of California at Irvine, the University of Chicago (except its medical school), Columbia's Arts and Sciences Division, Harvard, and Yale reported that more than one-half of their faculty who retired did so at age 70 or older (see Table 3). The faculty retirement data thus suggest that at some research universities high proportions of faculty will choose to work past age 70, but the data do not suggest any way to distinguish which universities they will be. Some of the variation in faculty retirement patterns may be due to different retirement incentive programs offered by some institutions.

On the basis of these data, the committee concludes: At some research universities a high proportion of faculty would choose to remain employed past age 70 if allowed to do so. Faculty at research universities are more likely than faculty elsewhere to have low teaching loads, relatively high-quality undergraduates, and the facilities to support research grants—the factors Rees and Smith found best predicted delayed retirements.

Any college or university at which a significant proportion of faculty decide to continue working past age 70 will experience faculty hiring reductions, cost increases, or both. In this section we estimate the magnitude of possible effects on faculty hiring and institutional budgets at universities at which faculty are most likely to work past 70. We also consider whether reduced hiring will limit opportunities for new faculty members. (We examine the consequences for faculty and institutional quality of reduced faculty hiring in Chapter 3, and we address ways to increase hiring by encouraging faculty retirements in Chapters 4 and 5.)

In order to project the effects of eliminating mandatory retirement on faculty hiring and the budget at a given institution, one must consider the current age distribution of its faculty members and the rate at which faculty of different ages enter and leave the institution. For example, some institutions do most of their hiring at the assistant professor level, and other institutions hire more mid-career faculty members. Likewise, faculty leave institutions at different ages and for different reasons: denial of tenure, acceptance of a position elsewhere, poor health, death, or retirement.

Administrators at a few colleges and universities use ''faculty flow models'' to estimate the numbers of faculty entering and leaving their institutions, as well as the age distribution of their faculty and the size of their budget (see, e.g., Hopkins and Massy, 1981). The modeler must specify the rate at which faculty enter or leave for each period, using historic data on hiring, resignations, retirements, and deaths. Based on the age distribution and the entering and leaving rates of faculty in each age category, the models then project faculty hiring and age distributions for successive time periods. Such models can also project future salary costs on the basis of estimates of the average salary of faculty in each age group. These models allow colleges and universities to examine the effects of policy changes on the composition and costs of their faculty.

The committee projected potential effects of uncapping on hiring, using data provided by three research universities that have a mandatory retirement age of 70 and at which a significant proportion of faculty may postpone retirement past age 70 if mandatory retirement is eliminated. Two of

these institutions examined some of the potential effects using their own faculty flow models and data on their arts and science faculty. For the third institution, the committee used a model it adapted from one designed by Biedenweg and Keenan (1989) to examine data supplied by the university. At university A a committee of faculty and administrators found that 64 percent of the faculty aged 60–64 remained employed after age 65. At university B a committee of faculty and administrators reported that 60 percent of tenured faculty in arts and science who retired did so at age 70. At university C more than 50 percent of the arts and science faculty who retire do so at the mandatory age.

The university A and B committees took different approaches to estimating the age of older faculty likely to retire after age 70. As a low estimate of changed behavior after uncapping (i.e., most faculty continue to retire before age 70), the university A committee assumed that the proportion of faculty who remain employed past age 65 would not change and that 25 percent of the faculty who worked past age 65 would continue to work past age 70. As a high estimate (i.e., more faculty retire after age 70), the university A committee assumed that the percentage of faculty who remained employed beyond age 65 would increase from 64 percent to 75 percent, and 50 percent of the faculty who worked after age 65 would also work after age 70.

In order to test the sensitivity of these assumptions, we ran our model using estimates of faculty working past age 70 that were both higher and lower than those provided by university A. The results show the model to be relatively insensitive to retirement or retention rates, because the total number of faculty reaching age 70 at university A is relatively low during the next 15 years.

The university B committee made projections using estimates of the percentage of faculty over age 70 who would continue to work each year after uncapping: 75 percent (low) and 90 percent (high). In other words, for the low estimate 25 percent of the faculty over age 70 retire each year, and for the high estimate 10 percent retire each year. In comparison to university A, university B assumed no increase in the proportion of faculty reaching age 70. Our committee also applied the university B assumptions (75 percent and 90 percent) to University C's model and faculty data.

The magnitude of cost increases or the limit on future hiring resulting from decreased turnover depend heavily on an institution's policy choices. We examined the potential impact of uncapping on hiring and budgets with three separate options: constant faculty size, constant budget, and constant hiring. A college or university can choose to maintain the current size of its faculty by hiring faculty at a slower rate. It can also maintain a constant salary budget, which, since salaries rise with age, will limit hiring still further. (Data on average salary by 5-year age group from universities A,

B, and C and from case study institutions show older faculty earn more on average, but there is little increase between ages 61–65 and 66–70. We have therefore assumed that faculty in the model who stay past age 70 will earn salaries equal to the average for faculty aged 66–70.) An institution could continue to hire new faculty at a steady rate regardless of postponed retirements, which maintains hiring flexibility but increases costs. Lastly, it could choose a policy other than constant budget, salary, or hiring, such as hiring a few more faculty members than it needs to fill openings but hiring fewer faculty annually than it has hired in past years. Table 4 summarizes the projected effects on each institution in terms of its increased salary costs or decreased hiring in comparison with its salary costs and hiring projected if no faculty member works past age 70 and faculty size is constant. This comparison presents the projected decrease in hiring due to uncapping, but it underestimates the total reductions in hiring expected at university C, which had planned to decrease its faculty size.

Several letter survey respondents stated that postponed faculty retirements would create a decrease in hiring because an increase in faculty career length means a decrease in faculty turnover. One letter survey respondent assumed colleges and universities would attempt to maintain a constant faculty size and pointed out the cost of this policy:

The assumption of a fixed number of faculty positions is most applicable to public colleges and universities in states in which the number of faculty is set by the legislature. We estimated the extent to which hiring would be reduced if the three universities for which we have data held faculty size constant after uncapping. Projected hiring reductions in the first 5-year period after uncapping range from 5 percent at university A using its low estimate to 31 percent at university B using its high estimate (see Table 4a).

The projected number of positions that become available eventually rises at all three universities as faculty who postponed retirement begin to retire, but the projected number of positions remains below the levels expected with mandatory retirement. After 15 years the projected decrease in hiring ranges from 2 percent to 15 percent less than levels projected if all faculty retire by age 70. In the interim period, however, patterns vary. At university A after 5 years, the projected number of positions falls even further and then begins to rise after about 10 years. University B expects fewer open positions for the 5–15 years after uncapping, regardless of mandatory retirement policy, because of a decrease in the proportion of current faculty who will reach retirement age. Using university B's low estimate, university C shows 5 percent fewer positions in the first 5 years and 8 percent fewer positions in both periods using the higher estimate. These results are within the range projected by Southworth and Jagmin (1979), who modeled faculty flows for the colleges and universities belonging to the Consortium on Financing Higher Education as a way to estimate the effects of raising the mandatory retirement age from 65 to 70.

A constant faculty size does not imply a constant faculty budget. Salaries tend to increase with age; thus, an increase in the average age of an institution's faculty increases overall costs. For university A, projected real salary costs increase overall by 2 percent over the first 5 years using the high estimate of postponed retirements and 1 percent using the low estimate. For university C, projected costs increase by about 2 percent over the

first 5 years after uncapping using the high estimate. (University B did not calculate cost effects.)

Few colleges and universities in the United States would find it easy to cover the cost increases associated with a larger faculty size. A college or university that is facing higher costs because of postponed faculty retirements might look to decreases in hiring as one mechanism to balance its budget. Using this assumption in the models yields a "worst case scenario" in the sense that all expense cuts come from the academic salary budget.

For university A's high estimate, the constant budget model projects 21 percent fewer faculty hired in the first 5 years after uncapping, in comparison with 14 percent fewer in the constant faculty size model. Total faculty size decreases by 2 percent. The model projects that some of the lost positions could be reinstated 15 years after uncapping, but as in the constant faculty size model, the rate of faculty turnover remains lower than that projected with mandatory retirement. With a constant budget the projected size of the faculty also levels off at a new, lower level. For university A's low estimates, the number of faculty hired and the size of the faculty initially decrease by 9 percent and 1 percent, respectively. They then rise gradually through successive 5-year periods, matching the levels expected with mandatory retirement 20 years after uncapping.

When costs are held constant in university C's model, using university B's high estimate, projected hiring is 12 percent less than if all faculty retire by age 70, in comparison with the 8 percent less projected using the constant faculty size assumption.

When we incorporated the average salary for faculty in each age range in the constant faculty size models, we found that uncapping causes projected faculty salary budgets to increase from 1 to 3 percent, depending on the estimated proportion of faculty who postpone retirement past age 70. Continuing to hire new faculty in an attempt to cover new fields would be even more costly. If an institution continued to hire new faculty at a rate greater than the rate at which faculty were leaving, its faculty size would increase, and its costs would rise accordingly.

To estimate the magnitude of such an increase, we ran the model for university A holding the number of faculty hired in each time period equal to the number projected to be hired for the first 5-year time period with mandatory retirement. Projected real salary costs for the first to fifth year after uncapping are 2–4 percent higher than costs projected for the same

period if all faculty retire by 70 (using the low or high projection of the proportion of faculty staying past age 70). The projected faculty size is 1–3 percent larger. When university C holds faculty hiring constant in its model, the projected costs in constant dollars of total faculty salary increase by 4–5 percent of the total salary budget in the first 5 years after uncapping. The university estimates that if it relied on tuition increases to pay for constant hiring, it would have to increase tuition by 3 percent just to cover the additional salary costs. If it relied on fundraising to cover the increase for the arts and sciences faculty salaries alone, it would need to raise a $30 million endowment. Furthermore, as the number of faculty increased, the institution would also have to pay for additional office space, laboratories, and support services.

As these figures suggest, an institution not planning to expand its faculty size has little opportunity to continue to hire if faculty postpone retirement. According to the provost at one research university, continued hiring is too expensive a strategy to use over an extended period:

At universities that are expanding, however, uncapping is unlikely to have major adverse affects. For example, in response to predictions that 63,000 additional students—over one-third of current student enrollment—will enroll by 2005, University of California system officials report plans to create three new campuses and increase enrollment and faculty size at seven of its nine campuses. Although these plans may be delayed because of state budget cuts, the system as a whole and all but two of its campuses, may eventually hire faculty at or in excess of former rates regardless of whether faculty members continue to work past age 70.

These models estimate the consequences of different policy choices in isolation. Colleges and universities can respond to faculty members' re

maining employed beyond age 70 by limiting any combination of hiring and faculty costs. A university might also cover additional faculty salary costs by limiting its expenditures for other categories of the budget, such as reducing support services or new construction, although any such reductions could have costs for the faculty and the institution. Colleges and universities could attempt to limit the number of postponed retirements by instituting programs designed to make retirement more attractive. However, financial incentives to retire or the provision of office space, support services, or other benefits for retirees create additional costs (see Chapters 4 and 5).

From a historical perspective, the projected effects of eliminating mandatory retirement on faculty hiring and on salary budgets are not extraordinary. The high inflation and energy costs of the 1970s caused greater financial hardship and more severe hiring constraints than are likely to result from changing mandatory retirement policy. The average age of faculty has risen and will continue to rise more because of the aging of current faculty than uncapping. The average age of faculty in the United States has been increasing because job growth in higher education leveled off during the 1970s and because of the possibly related increase in the average age of new Ph.D. recipients (Bowen, Lord, and Sosa, 1991).

Nevertheless, the committee recognizes that colleges and universities face severely limited sources of additional revenue. Most are already engaged in extensive fundraising. Research universities, including the universities most likely to be affected by the elimination of mandatory retirement, already struggle to balance their budgets, often through tuition increases that are well above the inflation rate. Additional salary costs or the cost of a retirement incentive program would add to existing fiscal pressures: State and federal funding of financial aid has been decreasing; federal support for overhead costs on grants is being reduced; new tax laws have limited fundraising; tax-exempt borrowing has been curtailed; other new tax regulations have forced institutions to cease offering unequal benefits to staff and faculty; new accounting regulations make retirement health benefits much more costly for private institutions; and colleges and universities face pressure to limit tuition increases. Few institutions expect the 1990s to match the 1980s in terms of economic growth, endowment growth, or low rates of inflation. The combination of these changes will make it more difficult for the universities most likely to be affected to adapt to the effects of uncapping. Thus, the elimination of mandatory retirement will have adverse effects on the budgets and hiring opportunities of some research universities.

For most faculty members the effects of eliminating mandatory retirement would be positive: They gain the right to choose a retirement age without any upper age limit on the choice. For some this future benefit may be partly offset by limited job opportunities in the present: Research uni-

versities adversely affected by uncapping will be forced to reduce their hiring or to undertake extraordinary fundraising activities to increase the number of available for positions, which suggests that they will have fewer positions available for either prospective new junior faculty or more senior faculty from other institutions. Since reduced rates of hiring will be confined to only some research universities, however, reductions in hiring are more likely to limit where faculty seeking research positions find jobs than whether they find jobs. Faculty who are qualified for positions at adversely affected research universities would be likely to attract offers from other research universities.

The committee's evidence shows the largest proportion of faculty are now aged 45–49, with the next largest group aged 40–44. Most of these faculty will not be considering retirement for at least 10–15 years. The effect of this "age bulge" on retirements will not take place until the end of this century or the beginning of the next.

Once those faculty in the bulge reach traditional retirement ages (i.e., age 60+), available evidence and projections indicate that the number of faculty retirements will increase, regardless of mandatory retirement. The committee, Lozier and Dooris (1990), and Bowen and Sosa (1989) all found that the likelihood of increased faculty retirements about 10 years from now is relatively insensitive to a range of possible future retirement rates. Colleges and universities will have increased faculty turnover. Cases of faculty continuing to work despite age-related declines in performance, although rare, could also increase. However, the lower numbers of faculty over age 50 suggest retirement levels will be relatively low in the coming decade. A decrease in the rate of retirements owing to uncapping at colleges and universities where the Largest proportion of faculty are not yet near retirement could exacerbate expected low hiring levels.

A number of studies have projected a national shortage of faculty by combining information about the overall faculty age structure with estimates of future student enrollments, student/faculty ratios, and rates of departure from academia (Atkinson, 1990; Bowen and Schuster, 1986; Bowen and Sosa, 1989; El-Khawas, 1990). Some administrators and faculty have suggested that encouraging faculty to work past age 70 could alleviate impending shortages. Others have suggested that the effects of eliminating mandatory retirement are too small to affect faculty shortages. Still others have expressed a preference for the ability to hire now, rather than in the future when some researchers have projected that the numbers of prospective now faculty members will be lower and overall demand will be higher.

Because of variations in age distributions at individual colleges and universities, disciplines, and geographic regions, the most appropriate focus for analysis and policy making in response to these concerns is at those levels. Some colleges and universities—for example, those drawing on a

regional base for students in areas with declining college-age population—may need fewer faculty in the future and not plan to replace all retiring faculty. Others, such as the University of California system, have proposed hiring a substantial number of additional faculty at certain campuses in response to projected enrollment increases.

Despite the growing number of older faculty members in U.S. institutions of higher education, evidence from uncapped colleges and universities suggests that few tenured faculty now continue to work past 70. Current faculty retirement patterns suggest most faculty choose to retire before the mandatory retirement age. The committee concludes: Higher education as a whole is likely to experience few changes in faculty behavior or demographics as a result of the elimination of mandatory retirement, and a significant number and proportion of faculty will choose to work past age 70 at a few research universities.

Faculty at some research universities are more likely than faculty elsewhere to have low teaching loads, high-quality undergraduate and graduate students, and research support. Our analysis of the demographic and financial consequences of postponed faculty retirements leads us to believe that the ability of research universities to hire new faculty or control salary costs could be significantly lessened for a transition period of 5 to 15 years. Under certain circumstances there could be less severe long-term effects.

The effects of uncapping will not be the same at all colleges and universities or on all faculty. At some institutions more than one-quarter of the current tenured faculty will reach retirement age in the coming decade; at other institutions none of the current faculty members will reach age 70 for 20 years. There are institutions at which more than one-half of the retiring faculty do so at the mandatory retirement age of 70 and institutions at which faculty have consistently chosen to retire by age 65. Therefore, the committee cannot predict the effects of eliminating mandatory retirement at each of more than 3,200 colleges and universities.

A college or university that is trying to plan for uncapping cannot rely primarily on the available aggregate data on faculty age distribution and retirement behavior. Yet we found only a few administrators and faculty who had studied their institution's faculty age distributions, retirement patterns, hiring needs, and costs as a way of understanding how their institution would be affected by the elimination of mandatory retirement. Most had not reviewed faculty handbooks, benefits procedures, or retirement programs to consider whether uncapping would require any revisions of college or university policy. The committee thus concludes: Administrators and faculty can best assess the potential impact of uncapping at their

own colleges and universities by studying their faculty age distributions, retirement patterns, and hiring needs in order to estimate the potential effect of uncapping.

At most colleges and universities, planning will not require the establishment of committees or a long-term study. Retirement patterns alone can indicate whether the elimination of mandatory retirement will have any effect on an institution. For example, prior to 1982 several faculty at one of our case study liberal arts colleges retired after age 70, but since that time faculty members have retired at earlier ages, and most current faculty members report that they plan to retire before age 65. The committee recognizes that analyzing data on faculty ages and retirement patterns is more complicated at colleges and universities with a larger faculty and, in some cases, multiple campuses than at a liberal arts college. Models like those we used in estimating effects on faculty turnover and salary budgets could assist administrators at these colleges and universities in estimating whether faculty are likely to choose to work past age 70.

Moreover, colleges, universities, higher education researchers, and groups representing higher education should all continue to monitor faculty retirements for changes in historical patterns. (We consider the current availability of resources for doing this in our discussion of faculty data bases in Appendix B.) We urge higher education systems and organizations to undertake their own monitoring and planning efforts.

We have expressed serious concern that some research universities will have fewer opportunities to hire and will face additional costs as a result of postponed retirements if mandatory retirement is eliminated. Other colleges and universities may also face low expected faculty turnover for the next decade or more, regardless of mandatory retirement policy, if their faculty age distributions reflect the national distribution with a disproportionately large number of faculty in the middle age ranges. In Chapter 3 we consider whether an increasing proportion of older faculty and decreased hiring opportunities will harm the quality of affected colleges and universities. In that chapter and in Chapters 4 and 5, we consider whether colleges and universities will be able to mitigate any adverse effects of postponed retirements or reduced faculty turnover.