For the fourth consecutive year, the average salary of full-time faculty in the United States rose more rapidly than the rate of inflation, but only barely so.1 The average salary increased by 3.5 percent over the 1999–2000 to 2000–2001 academic year, while the inflation rate (from the Consumer Price Index) was 3.4 percent. Among continuing faculty (those who were at the same institution in 1999–2000 and in 2000–2001), the average salary increase was 5.3 percent. This increase translates into real (inflation-adjusted) salary growth of 1.9 percent for continuing faculty—the lowest rate in four years.
In an important sense, this year’s bottom line is not good. From 1999–2000 to 2000–2001, the real value of faculty salaries was scarcely higher than the academic year before—despite the strong performance of the economy. Moreover, the troubling salary disadvantage of faculty relative to similarly educated professionals persisted. Might this be the end of a stretch of four consecutive years of real salary gain? Perhaps, and for two compelling reasons. First, the recent tightening of state budgets threatens the real salaries of faculty at public universities, where more than two-thirds of full-time faculty are employed. Second, recent history has taught us that even a mild recession—which many economists predict for 2001—can adversely affect faculty salaries.2
Table 1 shows the changes in nominal and real salaries for the academic years 1971–72 to 1973–74 through 1999–2000 to 2000–2001 based on the AAUP salary survey.3 For the years before 1985–86, salary summaries are grouped in two-year intervals to permit presentation of all data from 1971–72 forward. The All Faculty section of the table shows the change in the average salary of all faculty members employed in a given year compared with the average salary of all faculty members employed in the previous year for those institutions included in the survey in both years. The Continuing Faculty section of the table presents the average change in salary for those faculty members employed at the same institution in both years over which the change is calculated.
Table 1 reveals several interesting patterns. First, the recent relationship between faculty salary increases and the economic cycle is highlighted by the relative slowdown in real academic salaries around the time leading up to and immediately after the 1991 downturn in the business cycle. Second, the All Rank increase this year is lower than the rate of salary increase at each individual rank for the first time in recent memory. Mathematically, that can occur only if there is a change in the proportion of faculty at each rank within an institution. Such a relatively low All Rank increase is consistent with higher-salaried retirees being replaced with lower-salaried faculty at lesser ranks, or alternatively, with the elimination of the positions left vacant by retirees. It is also consistent with expanded institutional reliance on lower-paid non-tenure-track instructional faculty.
Table 1 also shows evidence of a spread between the salaries of professors and those of other faculty, and between the salaries of tenure-track faculty at all ranks and the pay of instructors. These gaps are unique to the last four years of data and highlight one aspect of the growing salary differences among faculty that will be discussed below.
Although this year’s All Rank increase of 0.1 percent puts faculty salaries at their highest level since the mid-1980s, it is important to note that salary gains in the 1990s were modest overall. Just a quick glance at Table 1 shows that real wages for the average academic today are not much different from what they were nearly thirty years ago.
Furthermore, professors are paid less than their peers in other professions. In past years, I have compared the average faculty salaries reported at the institutional level in the AAUP survey to the self-reported, individual-level salaries of professionals with similar education published in the Current Population Survey (CPS). These analyses showed that professors are paid roughly 25 to 30 percent less than similarly educated professionals, and that the magnitude of this salary disadvantage has changed little over the past twenty years.4
For comparative purposes (and because CPS wage data are often poorly reported), I make use this year of a different data set—the Occupational Employment Statistics (OES) Survey—to examine the salaries of faculty and comparable professionals in 1999. These data are based on surveys sent to approximately four hundred thousand establishments (not individuals), which presumably increases the likelihood of reporting accuracy. Like the CPS data, however, wages in the OES are truncated at the top of the wage distribution, implying that calculated annual wage differences are likely to be understated.5
The results in Table 2 from the OES data suggest that the average faculty member earns roughly 26 percent ($15,299) less than the average highly educated professional. The magnitude of the estimated gap from these data is similar to previous estimates using the CPS data, and gives us greater confidence in the legitimacy of the 25 to 30 percent estimate of the pay gap between professors and other professionals. The bottom line is that by relative standards, faculty are, at least on average, underpaid.
Differences Across Institutions
Although we are underpaid on average, there are growing differences among us, which I have highlighted in past reports. These differences and their trend suggest that the average figures (which do not control for institutional category, professorial rank, or the public, private-independent, or church-related status of institutions) reveal less both in cross section and over time than salary figures that differentiate among types of institutions.6 By extension, the average salary figures reported by institutions in the AAUP salary survey also hide differences that occur within rank and across disciplines that I will explore below.
Inequality in Salaries Across Institutional Types
The AAUP salary survey data show us that the academy, like the private sector, includes low- and high-paying employers. But, as I will explore in more detail below, the growing differences in average pay across postsecondary institutions have little to do with a change in the rank ordering of high- and low-paying institutions over time. Instead, the inequality trends result from growing differences between what high- and low-paying institutions pay.
Table 3 highlights various elements of rising inequality in faculty salaries in the 1980s and 1990s. The table uses the salaries of professor-rank faculty, but the inequality trends would remain qualitatively unchanged if comparisons were made at the associate- or assistant-professor rank. The first row of the table compares the pay of professors at private-independent (non-church-related) institutions with that of professors at public universities. The row shows that the gap between what private and public institutions pay continues to grow. This is troubling news given recent forecasts of imminent state budgetary pressures that might result in lower real salary gains, or even pay losses, among faculty at public institutions in the near term. It is worth noting that as the relative salary advantage at private institutions has risen, tuition costs have accelerated somewhat faster at public universities than at private institutions.7
The table also compares the pay of professors by institutional category, revealing that the salary advantage of being at a research (Category I) institution has grown over the past fifteen years. In addition, the table shows that the premium for working at the most "elite" research universities is far more substantial among public institutions than among private ones, where differences are more narrow. (Elite institutions are defined as the top twenty public and private universities according to the 1990 ranking of U.S. News & World Report.)
How have these growing differences affected the pay distribution in the academy? Table 4 uses an inequality measure that characterizes both the level of inequality and the distribution of faculty salaries, namely the difference in the natural log (ln) of faculty salaries at the fifth, fiftieth, and ninety-fifth percentiles in the salary distribution. This measure of in-equality has many advantages over more standard measures of salary dispersion, such as the variation in ln salaries over time, because it yields information about the spread in salaries at the top and the bottom of the salary distribution and therefore conveys information about the shape of the distribution over time.
The first row in each category group is the spread in the natural logarithm of average reported salary at the ninety-fifth versus the fifth percentile in the relevant distribution. This difference (given the natural logarithmic specification) can be thought of as a percentage difference, implying, for example, that the gap between what professors earn at doctoral-level (Category I) institutions at the top (ninety-fifth) percentile and bottom (fifth) percentile of the salary distribution has grown from 47 percent (.47 ln salary dollars) in 1990–91 to 57 percent (.57 ln salary dollars) in 2000–2001.
The gray shading highlights the occasions of rising inequality in the distribution over the 1990–91 to 2000–2001 period. The table demonstrates that the overall distribution of faculty salaries across institutions has widened at all ranks among doctoral-level (Category I) institutions that provide data to the AAUP, but not as uniformly in other institutional categories. Taken together, Tables 3 and 4 suggest that salary differences between institutions are growing faster than differences within institutions—at least when average pay at rank is used as the relevant measure.
Table 4 is perhaps most interesting for what it says about how salary distributions are shifting. Specifically, the results point to the increased contribution of the top part of the salary distribution (the fiftieth to ninety-fifth percentile) in explaining inequality. In 1990–91 the larger weight of inequality was due to differences between low- and average-paying institutions. By 2000–2001, however, the distribution had shifted. High-paying institutions, presumably as a means of luring the most talented faculty, seem to be pulling away from the pack. This conclusion is consistent with the results from Table 3, which show growing salary differences between elite and other universities.
The Distribution of Salary Changes
This analysis is also supported by the distribution of salary changes. Table 5 compiles data from the salary survey table "Percentage of Institutions and Percentage of Faculty by Average Increase in Salary Levels, by Affiliation and Category" for 1994–95, 1996–97, and 2000–01 (Survey Report Table 2 in this issue of Academe). Given differences in the underlying inflation rate in each of the three years (3.4 percent this year, 3.3 percent in 1996–97, and 2.7 percent in 1994–95), I focus not on differences in the level of increases across years, but rather on changes in the distribution of increases within years over time.
The distribution of salary change is increasingly disperse. In 2000–2001 most institutions reported salary increases in either the top range of 6 percent or more (34.7 percent of institutions) or the bottom range of 2 percent or less (33.6 percent). In both 1996–97 (when the inflation rate was markedly similar) and 1994–95, however, more institutions experienced salary changes in the middle ranges. Moreover, in the years before 2000–2001, the distribution of salary growth was more equal among research (Category I) universities than among all institutions combined. This year, for the first time in recent memory, it is less equal among research universities than among all institutions taken together.
Stability in the Salary Distribution Over Time
How stable is the faculty salary distribution? Do schools that pay low (or high) salaries at some point continue to pay low (or high) salaries ten years later? Has the overall stability of the distribution changed over time?
To examine this issue, I looked at institution-specific salary data from the AAUP survey for 1980–81, 1990–91, and 2000–2001. Using 1990–91 salary levels as a base, I compared the rank order of institutions (according to what they paid professors) in 1990–91 to their rank order in 1980–81 and 2000–2001. Figure 1 presents the outcome for institutions that were in the lowest quartile in the 1990–91 salary survey. It shows the fraction that were so positioned in 1980–81 (lighter bar) and in 2000–2001 (darker bar). Figure 2 performs a comparable analysis for schools in the top quartile of the pay distribution.
The figures reveal a relatively high degree of stability in the rank ordering of schools: those that paid high salaries to professors in 1990–91 paid similarly high salaries ten years earlier and ten years later. In addition, the rank order among research (Category I) universities showed the greatest persistence over both ten-year intervals. These data provide further evidence that the high level and rising trend of inequality among research universities is due to growing salary differences between high- and low-paying institutions and not to a changing mix of these institutions over time.
The data also demonstrate a greater persistence in the rank ordering of institutions at the high end of the distribution of salaries than at the low end, implying that over time, high-wage institutions more frequently maintain their relative standing than do low-paying institutions. Moreover, persistence in the rank ordering of institutions was generally greater in the most recent decade (1990–91 to 2000–2001) than it had been in the previous one (1980–81 to 1990–91).
It is difficult to infer from these numbers whether rank-order persistence among institutions of higher education is high or low, since we don’t have a relative base for comparison. One well-known study, however, looked at firms within industries to see if wage differences across employers last. It found that the simple correlation of pay over five- and fifteen-year periods was 0.83 and 0.65, respectively.8
These figures resemble those I obtained for ten- and twenty-year correlations in professor salaries from AAUP data. For example, rank-order correlations were in the 0.75 to 0.9 range and statistically significant for institutions in all categories and ranks (above instructor) over the 1990–91 to 2000–2001 period, with the highest correlations occurring among doctoral-level (Category I) institutions. Twenty-year correlations (1980–81 to 2000–2001) produced a range of statistically significant estimates from 0.5 to 0.8, again with persistence in salary structure notably higher among Category I institutions. In sum, the persistence of the salary structure in academe appears similar to that among firms in industry, except perhaps among research universities, where there is less change in rank over time.
Differences Within Institutions
In addition to modest change in overall levels of faculty salaries and more substantial variation in pay across institutions, the 1990s also brought about change in the pattern and level of disciplinary differences. Indeed, salary differences among disciplines are large, and they continue to increase over time. The data in Table 6 come from a sample of institutions that provided data in 1979–80, 1989–90, and 1999–2000 to the Faculty Salary Survey by Disciplines of Institutions, which is sponsored by the National Association of State Universities and Land-Grant Colleges (NASULGC). The Office of Institutional Research at Oklahoma State University conducts the survey annually. In the 1999–2000 survey, eighty-eight doctoral-level (Category I) institutions submitted data for the survey. As with the AAUP survey, the sample of institutions that submit data in time to be included in the NASULGC survey may change from year to year.
Table 6 shows average professor and assistant professor salaries for disciplines in the top and bottom quartiles (ranked by professor salary in each year). High-paying disciplines include law, business, the health professions, computer and information sciences, engineering, the physical sciences, and mathematics. Lower-paying disciplines include the visual and performing arts, library sciences, agricultural sciences, education, home economics, and communications. The high- and low-paying quartiles do not include an identical subset of disciplines across years (although there is little change); rather, the selection criterion is based on professor-salary rankings in each year of analysis.
Table 6 shows that differences across disciplines have been growing, and at a more rapid rate in the 1990s than previously. Indeed, the Oklahoma State data reveal that the gap between faculty salaries in high- and low-paying disciplines in the 1990s grew at a rate three times as rapid as in the previous decade, implying differences of roughly 35 percent between the top- and bottom-paying disciplines.
Figure 3 charts the overall variation (measured as the standard deviation in the natural log of average salary) in the distribution of salaries by discipline in 1979–80, 1989–90, and 1999–2000. It shows a clear pattern of widening dispersion. The historically high level of variation and the dramatic increase in variation in the 1990s are also evident.
To what do we owe this pattern of widening pay differences by field? To the extent that academic salaries are a function of the labor market, they will reflect relative demand-and-supply patterns both in academia and in competing labor markets. Surely, the ascension of salaries in the computer and biotechnology fields has to do with the demand for talented researchers in the private labor market, much as the relatively low salaries in the performing and visual arts have to do with excess supply and relatively weak pay in those fields outside academia. Indeed, I would hazard that disciplinary differences in the 1990s have as much to do with relative demand-and-supply shifts outside the university as with demands for talent inside. And I do not think we can underestimate the contribution of rising inequality in the distribution of U.S. incomes generally to the growing inequality within our own ranks.
Differences Between Men and Women
Previous AAUP survey reports have documented troubling and persistent differences in the salaries of men and women in academe over time. These differences persist if you control for rank, institutional category, and the public or private-independent status of the institution. Although gender salary differences are small overall, as I have reported in earlier surveys, these differences are significantly larger at research universities and for faculty at the rank of full professor.9
In public institutions, male professors earn on average 6.5 percent more than female professors, according to data from this year’s salary survey. Male professors in private institutions earn 5.9 percent more than female professors generally and 10 percent more than women in both private and public research (Category I) universities, where the male salary premium has for long been the largest. So the male salary differential persists in the 2000–2001 salary survey data. How uniform is this differential? Does it exist even in institutional settings where salaries to women professors are purported to be the highest?
To examine these questions, I looked at a subset of private and public institutions that are ranked highest in terms of what they pay women professors.
Table 7 summarizes these results. Not surprisingly, these institutions have high salaries for male professors as well, and accordingly, the differentials in pay are comparable to the overall averages. Exceptional cases are Rockefeller University and the College of William and Mary, which report higher average salaries for women than for men.10
The goal of this year’s report has been to characterize, in yet new forms, the salary differences that continue to grow among faculty. The consequences of these differences are not trivial. Although it is hard to imagine the richest and most desirable universities facing a shortage of well-qualified faculty, it is not difficult to imagine excess demand for computer scientists, mathematicians, physicists, biologists, economists, and others among institutions that are constrained in their ability to pay and therefore compete. I venture that disciplinary differences growing within our institutions also affect the quality of our environment and the decisions we make.
As I write this report, the outlook for the U.S. economy looks dim, with the most optimistic predictions arguing for a sharp slowdown in growth that will no doubt trickle down to states and affect the pay of public-sector faculty. This likelihood argues against any real progress in salaries in the immediate term. The prosperous 1990s do indeed seem to be over, and unfortunately, most among us have experienced little real salary gain.
Ernst Benjamin, the director of the AAUP survey, reports that the National Center for Education Statistics (NCES) did not collect faculty salary data for the 2000–2001 academic year. In previous years, many institutions sent the AAUP data on NCES forms; the AAUP also obtained copies of NCES forms from NCES state data coordinators. Consequently, this year’s report contains about 82 percent as many institutions as the 1999–2000 report. As Table 8 shows, the falloff was much greater for community colleges (which are generally less well represented in the survey) than for four-year institutions. The decline in participation may modestly affect the overall reliability of the report, especially the community college data. It also diminished the number of institutions on which the average salary increase is based. But the increase in the proportion of reports based on AAUP survey forms has increased the number of institutions providing continuing-increase data and fringe-benefit data by rank. As a result, that data may be somewhat more reliable than in previous years.
This is the last year that I will be responsible for writing this report. For five years, I have worked closely with Ernst Benjamin, the AAUP’s director of research, in putting together the report. Ernie has provided countless insights, and I have learned from working with him over the years. My heartfelt thanks to him for his help and support. Patrick McCauley, the associate director of the AAUP’s salary survey, is once again to be thanked for the hard work he put in to ensure the timely release of accurate salary numbers. Galina Lewis, the research associate for the survey, also deserves thanks.
In addition, I wish to thank the members of the Committee on the Economic Status of the Profession who suggested several of the topics pursued in this report, which I researched and wrote while on leave at Stanford University. Special thanks go to those members of the committee who commented in detail on my outline and helped along the way by suggesting related literature and interesting anecdotes. The committee members are W. Lee Hansen (Economics, University of Wisconsin–Madison); Anne Harrison (Finance and Economics, Columbia University); James May (California State University–Monterey Bay); Lonnie Stevans (Business, Hofstra University); Craig Swan (Economics, University of Minnesota–Twin Cities); Jeffrey Waddoups (Economics, University of Nevada, Las Vegas).
LINDA A. BELL
(Economics, Haverford College), Chair
Committee on the Economic Status of the Profession
1. Most of the information in this report is based on the AAUP survey of higher education institutions in the United States. In 2000–2001, 1,443 institutions (representing 1,722 campuses) are represented in the survey. Data from these institutions are included in the basic results in Table 1 and many of the other tables in this report. AAUP staff compiled the data on which the tables in the report and the appendices that follow are based. Back to Text
2. See Ernst Benjamin, "Special Report," TIAA-CREF Benefit Plan Counselor (March 2001): 6–7. Back to Text
3. See the notes to the survey report tables on pages 36–48 regarding institutional representation. To meet the Academe publication deadline, this report was written before the completion of data gathering. The tables in the report may therefore differ slightly from the appendix tables. Back to Text
4. See, for example, Table 2 on page 15 of the March–April 2000 issue of Academe. The Current Population Survey is produced by the U.S. Bureau of Labor Statistics. Back to Text
5. Annual wages are imputed from hourly wages based on the assumption of full-time and full-year annual hours. The Occupational Employment Statistics Survey is produced by the U.S. Bureau of Labor Statistics. Back to Text
6. In this article, the designation "private-independent" does not include church-related institutions, which are listed separately in the survey report tables that follow the article. Back to Text
7. For example, tuition costs were 81 percent higher at private universities than at public institutions in 1992, but only 75 percent higher in 1999. Back to Text
8. Erica Groshen. "Do Wage Differences Among Employers Last?" (Unpublished paper, Federal Reserve Bank of Cleveland, 1991). Back to Text
9. For example, see the report in the March–April 2000 issue of Academe, which focuses on gender differences within rank across public and private institutions, and by institutional category. Back to Text
10. This is not an exhaustive list of institutions. Indeed, among public institutions there are sixty-nine that reported paying higher average salaries to women professors; among private institutions, there are forty-two such cases. The public institutions in this group report an average salary of $62,000 for women professors, and the private institutions report average salary of $66,000 for women. Back to Text