Trends in US Medical School Faculty Salaries, 1988-1989 to 1998-1999

Throughout the last half-century, clinical revenues from faculty practice plans and support from affiliated hospitals have been an important source of support for medical schools.^{1 - 3} During the last decade, these revenues have been under pressure. With expansion of managed care and intensified price competition, payers have been able to negotiate discounts with faculty physician groups that have lowered reimbursement per unit of service. In addition, the architects of the Medicare Fee Schedule (MFS) designed a resource-based relative value scale for physician payment under Part B (Medical Insurance), implemented in 1992, to increase reimbursement for primary care services relative to payment for procedure-oriented and highly specialized medical and surgical services.^{4 - 5} The latter services typify the practice of academic physicians.

In light of these changes in physician service reimbursements, we would expect to see little growth in the salaries of academic physicians, if not an actual decline. Indeed, the Medical Group Management Association reported recently that the median compensation of academic specialist physicians (vs primary care physicians) increased just 1.5% from 1998 to 1999.⁶ In previous reports, compensation increases had generally been greater, although quite variable from year to year. The representativeness of the Medical Group Management Association data is limited, however, by a 25.3% response rate.

The main purpose of the present study was to examine trends in salaries of academic physicians over the last decade, using a more extensive salary database than has been used previously. We expected to find evidence for downward pressure on academic physician salaries generally and differences in salary growth rates across specialty groups. We also examined trends in salaries of PhD basic science faculty members, which while not affected directly by changes in reimbursements for physician services, might be expected to be sensitive to financial pressures felt by the medical school as a whole. For both groups of faculty, we examined whether changes in compensation varied by school ownership, type (eg, research-intensive schools and community-based schools), geographic region, and extent of managed care penetration, as well as by faculty rank and department.

We analyzed data collected through the Faculty Salary Survey (FSS) of the Association of American Medical Colleges (AAMC). Started in the mid-1960s, the FSS is the most comprehensive source of data on salaries of medical school faculty members. Each academic year, the AAMC distributes the FSS to all allopathic medical schools in the United States. Summary measures of data reported in the FSS are published annually in the Report on Medical School Faculty Salaries.⁷

In the FSS, medical school representatives report annually the pretax compensation of individual full-time faculty members, along with the faculty member's degree (MD or equivalent, PhD or other doctoral, or nondoctoral), rank (from instructor to full professor with department chairs separately identified), and primary department affiliation. Reported compensation data include both fixed and variable (supplemental and bonus or incentive) components. While school representatives report these components separately, we used data on total compensation from all components. As part of total compensation, school representatives report faculty member contributions to salary reduction programs (eg, 403B/TSA plans), but not employer contributions to retirement plans, special honoraria, and the value of fringe benefits provided by the institution, such as tuition benefits and employer-paid health, life, and disability insurance.

To assess recent changes in faculty compensation, we analyzed data from the salary surveys for academic years 1988-1989, 1993-1994, and 1998-1999 (referred to in the text as 1988, 1993, and 1998, respectively). We excluded instructors and limited the analyses of clinical science departments to faculty members with medical degrees (MD or equivalent). We limited analyses of basic science departments to faculty members with nonmedical doctoral degrees (PhD or equivalent). As a result of these adjustments and incomplete reporting, not all faculty members for whom schools reported compensation data were included in the study. We believe, however, that the cases analyzed are representative of faculty members overall for several reasons.

First, faculty member distributions analyzed here resemble distributions of faculty members in the AAMC's Faculty Roster System (FRS) (http://www.aamc.org) on key dimensions such as rank, degree, and geographic distribution. For example, in 1988, 31.7% of the MDs in clinical departments analyzed here were full professors vs 30.8% in the FRS. Most differences between the FSS and FRS distributions were less than 3 percentage points and many, less than 1 percentage point. Second, by using the FRS to represent the population of medical school faculty members, we estimate our coverage of the population to be 75% each year (approximately 70% for MDs in the clinical sciences and more than 90% for PhDs in the basic sciences). Third, the number of cases analyzed increased from 30,792 in 1988 to 42,339 in 1993 and 47,573 in 1998, consistent with the growth in faculty during this period.

We conducted 2 sets of analyses. First, we used the individual-level salary data to compute mean and median salaries for specific basic science and clinical departments in each of the 3 years. Because of the skewing effect of outlying salaries on means, medians tend to be better descriptors of average salaries. Means, however, are useful for further statistical assessment of the data.

Second, we conducted bivariate and multivariate analyses to examine the relationship between salary growth during two 5-year periods, 1988-1993 and 1993-1998, and independent variables related to faculty rank, primary department affiliation, and various school characteristics. The bivariate analyses used 1-way analyses of variance to examine the relationship between salary growth and specific independent variables. The multivariate analyses used multiple (ordinary least squares) regression to examine the relationship between salary growth and 2 or more independent variables. Independent variables whose means showed significant differences in the bivariate analyses were included as independent variables in the multivariate analyses. The salaries of faculty in basic science and clinical departments were examined separately in the bivariate and multivariate analyses.

To conduct the bivariate and multivariate analyses, we adjusted the individual-level salaries for changes in purchasing power over the study period using the Consumer Price Index for All Urban Consumers (base year, 1988).⁸ Then, we aggregated the individual-level records into unique rank-department-school combinations (eg, full professors in anesthesiology at Yale) for each year. Because at least 2 years of data are required to calculate growth rates and to keep a consistent set of rank-department-school combinations, we excluded combinations that appeared in only 1 of the 3 years, or in 1988 and 1998, but not 1993. After adjustment, 5213 rank-department-school combinations remained in 1988 and 1993 and 5594 in 1993 and 1998. The remaining rank-department-school combinations were treated equally, regardless of size, in the bivariate and multivariate analyses.

Next, we calculated the mean salary, in real dollars, of the remaining rank-department-school combinations for each year. We used the proportionate change in real mean salary from 1988 to 1993 and from 1993 to 1998 to represent salary growth for each combination over the 5-year periods. For example, if the real mean salary of full professors in anesthesiology at Yale was $150,000 in 1988 and $165,000 in 1993, the proportionate change in real mean salary for this rank-department-school combination over the 5-year period would equal 0.10 [($165,000 − $150,000) / $150,000].

The bivariate and multivariate analyses examined the relationship between proportionate change in real mean salary of the rank-department-school combinations and independent variables related to faculty rank, department affiliation, and various school characteristics during the two 5-year periods. We first classified schools as public or private and by their geographic locations (Northeast, Midwest, South, and West). We identified research-intensive schools as those among the top 20 recipients of federal research funding (in academic year 1993-1994), and compared them with all other schools. We defined community-based schools (n = 17) according to a classification maintained by the AAMC as those that rely heavily on community hospitals and community-based physicians for their teaching program, and compared those schools with all other schools. As a measure of managed care penetration, we used data from The InterStudy Competitive Edge⁹ on the percentage of the population enrolled in health maintenance organizations (HMOs) by metropolitan statistical area. The HMO penetration data used to classify schools were recorded for 1997, the most recent year for which data were available. We coded schools in metropolitan areas where HMOs accounted for at least 40% of the market as high, 20% to 39.9% as medium, and less than 20% as low. In addition to providing clear categories, the coding scheme produced symmetrical and consistent distributions of faculty members, with approximately 25% of faculty members in the low and high categories, respectively, and 50% in the medium category, each year.

Table 1 presents mean and median faculty salaries and salaries at the 20th and 80th percentiles, in actual dollars, by department for each study year. Both mean and median salaries, in actual dollars, for faculty members increased consistently during each 5-year period and differed substantially between basic and clinical science departments (Table 1). Between 1988 and 1998, the median basic science faculty salary increased from $52,000 to $78,000 and the median clinical faculty salary increased from $101,000 to $150,000. Actual salaries varied greatly among clinical science departments in the 3 study years, but little among basic science departments. While actual median salaries in basic science departments differed by no more than $6000 in any year, actual median salaries in clinical departments ranged from $80,000 for family practitioners to $148,000 for surgeons in 1988 and from $121,000 for pediatricians to $217,000 for surgeons in 1998. Among clinicians, average salaries in departments of surgery and surgical subspecialties, radiology, anesthesiology, and obstetrics and gynecology tended to be the highest. Average salaries in the generalist departments of family practice and pediatrics and in psychiatry tended to be the lowest.

Clinical Sciences. The bivariate analyses showed that the proportionate change in real mean salary for clinical faculty was related to both rank (P<.001) and department (P<.001) during each 5-year period. The rank effect was primarily a "chair" effect. The mean salary of chairs increased at a greater rate than the mean salary of other faculty members at each academic rank during each 5-year period. This is illustrated by the ratio of the mean salary of chairs to the mean salary of other full professors, which increased from 1.35 in 1988 to 1.41 in 1993 and 1.48 in 1998.

Significant differences in salary growth related to factors other than rank and department were specific to a particular study period. From 1988 to 1993, clinical faculty salary gains differed by region (P = .005), with salary gains in the South outpacing those in the Northeast; school ownership (P<.01), with salary gains in private schools outpacing those in public schools; and school research intensity (P = .005), with salary gains at research-intensive schools lagging behind those at non–research-intensive schools. From 1993 to 1998, clinical faculty salary gains at community-based schools outpaced those at non–community-based schools (P<.001). The analyses failed to find any differences based on the HMO penetration level of the area in which a school is located.

In the multivariate analyses, the relationships between the proportionate change in real mean salary and the independent variables whose means showed significant differences in the bivariate analyses remained. Again, rank and department were the only independent variables to have consistently strong effects on salary growth rates for each 5-year period. Although the regression models were significant (P< .001), they accounted for very little of the variance in salary growth rates of clinical faculty members (<7% from 1988 to 1993 and <5% from 1993 to 1998).

Basic Sciences. As with the finding for clinical faculty, the bivariate analyses showed a consistent and significant relationship between rank and proportionate change in real mean salary for basic scientists (P<.001). Professors who were also chairs saw their salaries increase during both periods at a faster pace than other assistant, associate, or full professors. As a result, the ratio of the mean salary of chairs to the mean salary of other full professors increased from 1.37 in 1988 to 1.43 in 1993 and 1.49 in 1998. No differences in salary growth were found among basic science departments during either period.

Other effects related to salary trends for basic scientists were found for geographic region (P<.001), school ownership (P<.001), and school type (P = .004), all of which were specific to a given study period. Between 1993 and 1998, basic science faculty in the Northeast recorded smaller salary gains than their colleagues in the Midwest and West, and basic science faculty at community-based schools had larger salary gains than those at other schools. Between 1988 and 1993, salary growth of basic science faculty at public medical schools lagged behind that at private schools. Neither the research intensity of a school nor the HMO penetration rate of the area in which a school is located was related to differences in salary trends for basic scientists in either period.

While the bivariate analyses showed that faculty rank and several school characteristics were related to the proportionate change in real mean salary for basic science departments, multivariate analyses again showed the explanatory power of the variables to be limited. Regression models constructed from these variables explained approximately 5% of the variance in salary growth rates from 1988 to 1993 and less than 4% of the variance from 1993 to 1998.

Given the results of the bivariate and multivariate analyses, we examined salary growth for faculty in basic science and clinical departments from 1988 to 1998 more closely. We used the individual-level salary data to calculate annualized compound growth rates for basic science and clinical departments for each 5-year period and adjusted the department mean salaries computed earlier (Table 1) for changes in the rank distribution of faculty members and in purchasing power during the study period. To adjust for the effect of changes in the rank distribution of faculty members on mean salaries, we weighted the 1993 and 1998 mean salaries for basic science and clinical departments by their respective rank distributions in 1988. We deflated the rank-adjusted means using the Consumer Price Index for All Urban Consumers (base year, 1988).⁸ Using the rank- and inflation-adjusted means, we calculated annualized compound growth rates, in percentages, from 1988 to 1993 and from 1993 to 1998 for each department group.

The annualized compound growth rates revealed important differences in salary growth for faculty in basic science and clinical departments from 1988 to 1993 and from 1993 to 1998. After adjusting for inflation, during the earlier period, average salaries of clinical faculty increased 1.9% per year, while average salaries of basic science faculty increased only 0.3% per year. From 1993 to 1998, the situation reversed. Basic scientists recorded a 1.3% inflation-adjusted, annual increase in salaries, while clinical faculty recorded a 0.2% inflation-adjusted, annual increase.

Annualized compound growth rates for individual clinical departments are shown in Figure 1. In the earlier period, faculty in departments of obstetrics and gynecology, cardiology, radiology, and surgery recorded the largest real mean salary gains, while in the latter period, faculty in family practice departments did so. The data in Figure 1 also illustrate the downward pressure on salary growth in clinical departments from 1993 to 1998 noted earlier and the rapid turnaround for particular clinical specialties. The downward pressure and turnaround were particularly pronounced in the specialties of anesthesiology, radiology, and obstetrics and gynecology. After showing large relative increases in real mean compensation from 1988 to 1993, average salaries in these departments lost ground to inflation between 1993 and 1998. In contrast, the rank-adjusted real mean salaries of faculty in departments of family practice grew at an accelerated pace throughout the study periods, increasing at a 0.6% annual rate in 1988-1993 followed by a 2.7% annual growth rate in 1993-1998.

The data reported here suggest that medical school faculty salaries generally continue to increase, at a rate at least equal to, if not above, inflation. Yet, downward pressure on salaries for faculty in certain departments, expected as a result of changes in physician service reimbursements, is also evident. The notable inflation-adjusted growth of clinical faculty salaries between 1988 and 1993 has been replaced by growth just slightly above inflation. The clinical specialties of anesthesiology, radiology, and obstetrics and gynecology have experienced a rapid turnaround in salary growth. While actual average salaries in these specialties have not declined, they are now not keeping pace with inflation. Salaries in family practice run counter to this trend. The accelerating real growth of average salaries of faculty in family practice departments would have been predicted from the introduction of the MFS and from changes in physician supply-demand ratios brought about by managed care. Despite their accelerating real growth, average salaries of faculty in family practice departments are still lower than average salaries in most clinical departments.

We expected that salaries of basic scientists would be affected by pressures on clinical reimbursements in the same way as salaries of clinicians, at least to some degree. However, this does not appear to be the case.

Faculty members, in both basic science and clinical departments, may view with some cynicism the finding that their average salaries have increased less rapidly than the average salaries of department chairs. In truth, changes in the role of the department chair and labor market–related pressures to compensate top administrators adequately may be dictating these salary increases.^{10 - 11} The demands of managing medical school departments appear to be more difficult than ever. Recruiting and retaining the best leaders for these jobs remains a challenge for schools.

We do note that the average salaries of faculty members at public medical schools tended to grow more slowly than those of faculty members at private medical schools from 1988 to 1993. This period corresponds to a time of flat or declining state appropriations for public medical schools.³ Also, it is not surprising that the average salaries of clinical faculty at community-based schools have grown faster recently than those of faculty at non–community based schools. Community-based-schools tend to have a higher concentration of generalists, precisely the physician group that is favored by changes in fee schedules and new systems for health care delivery.

We did not find an expected relationship between salary growth and HMO market penetration levels. In retrospect, positing such a direct relationship may have been too simplistic. The impact of HMO market penetration on physician salaries is mediated by several factors, including the rate of expansion, degree, and responses of schools and faculty practice groups. It may be that as schools develop more experience with HMOs and other forms of managed care, they adapt to the demands of these organizations. In addition, the MFS might be a factor here. The schedule directly affects reimbursement to all physicians to the extent to which they serve Medicare patients. Furthermore, private payers, including managed care organizations, have increasingly adopted, in whole or part, the resource-based, relative-value scale employed by the MFS.¹² The generalized impact of the MFS and related federal payment policies may have contributed significantly to the overall deterioration in the growth of academic physician salaries in 1993-1998 and blurred any differential marketplace effects.^{4 - 5}

Overall, our findings are consistent with data reported in other studies regarding physician income over the last decade. In terms of salary growth rates, primary care physicians have fared better during the last decade than specialists, particularly hospital-based specialists. Among physicians generally, the American Medical Association reported that the nominal median net income of physicians in general or family practice, general internal medicine, and pediatrics increased at a greater rate than that of all physicians from 1987 to 1997.¹³ Similarly, data reported by the Medical Group Management Association show median compensation of primary care providers increased, in total, 22.3% from 1991 to 1995, while median compensation of specialists increased 6.55% during the same period.¹⁴

This study has several limitations. First, our measure of managed care market penetration, the percentage of the local population enrolled in HMOs, might be too insensitive to capture the intended effect. A measure that combines HMO and other forms of managed care, for example, preferred provider organizations, discounted fee-for-service, and utilization review, might better identify areas where price competition has become most intense. Also, we measured HMO market penetration at 1 point in time, 1997. The effect of HMO market penetration on salaries might be reflected better by a measure that captures the magnitude and rate of change in HMO enrollment over time. One study that examined the issue, however, reported that physician earnings were associated more closely with the level of managed care penetration than growth of managed care.¹⁵

Second, our construction of time periods is one of many possible constructions. Changing starting years, ending years, or both, as well as the number of years covered, can affect salary growth rates. This is particularly likely when a large change in salaries occurs near the margins of time periods. Our construction of time periods, however, allowed us to use the most recent salary data and to develop periods of standard length (5 years). Third, because of the lack of data, we were unable to examine the effects of other key variables, especially standard demographic measures such as sex and race, on growth in faculty salaries.

Fourth, and perhaps most important, the study does not take into account changes across these periods in the amount of time clinical faculty spend in patient care activities. A growing body of anecdotal evidence suggests that clinical faculty are spending more time in patient care (at the expense of time in teaching and research) than they had previously to maintain departmental and institutional revenues in the face of constrained reimbursements. Thus, one could argue under this premise that the decline in salary growth rates during the last decade would be much more severe than shown here, if faculty members and institutions were not making these adjustments.

We examined trends in salary growth rates of medical school faculty for MDs in clinical science departments and PhDs in basic science departments from 1988 to 1998 and found evidence that actual average salaries continue to increase for all groups. However, while the real growth rate of average salaries of basic science faculty members is increasing, that of average clinical faculty salaries is declining. This has reached the point that, overall, average clinical faculty salaries now are growing at a rate slightly above inflation, while certain clinical departments and specialties are losing ground to inflation. Salaries of clinical faculty members in family practice departments run counter to this general trend. In future studies, we plan to examine the trends described herein more closely, compare more directly salary levels of academic physicians by specialty with salaries of their counterparts in private practice, and explore how any differentials in compensation have changed for these 2 groups over time.