Schloss, Ernest P. PhD; Flanagan, Daniel M. MBA; Culler, Cheryl L.; Wright, Anne L. PhD
The general topic of employee turnover has been part of the management literature for decades,1 with more than a thousand published studies during the 20th century.2 Much of this literature has dealt with the psychological, sociological, and economic reasons why employees leave organizations, and how to remedy the perceived problems that have led employees to leave their jobs.3,4 Only more recently have organizations begun to quantify the direct and indirect costs of employee turnover.5
Several studies have shown that employee turnover is a major business expense. For example, a major technology company recently reported that it spent as much as $500,000 to $1 million per person replaced.6 Most of these studies consider only the direct costs associated with separation of the former employee, such as hiring and recruitment expenses, and the costs for training the new employee.2,4 However, direct costs typically only account for between 15% and 30% of total turnover costs.6,7 Indirect costs, which are hidden and more difficult to quantify, are related to such items as inefficiency of the new, leaving, and remaining employees.7
With the retirement of faculty members hired during the 1950s and 1960s and the impeding retirement of baby boom professors, the issue of faculty turnover has assumed greater import for universities,8 as well as their medical schools.9–11 However, there has been little research into the costs associated with faculty turnover in academia and, specifically, in academic medicine. In the handful of studies that have considered faculty turnover in medical schools, estimated costs varied depending on which factors were included in the model. In addition, estimates were influenced by the type of department considered (clinical versus basic science), whether the institution was public or private, area of specialty, and locale.12 Thus, estimates have ranged from $113,000 per faculty member in a Southern public medical school12 to $926,000 in a state-run health care system in the North.13 Although this literature provides some initial estimates of the magnitude of costs associated with faculty turnover in academic medicine, many studies include only recruitment costs,12 and the precise costs included in the models are rarely specified14; yet, those costs comprise only a small part of the expense associated with faculty turnover.
Our analysis, by contrast, calculates some hidden costs of turnover of clinical faculty at the University of Arizona College of Medicine (UA COM). It uses actual costs incurred in two clinical departments for different types of faculty, and it includes estimates of net (after expenses) department/faculty practice plan clinical income lost during the time the position was unfilled. A list of the factors considered in these calculations is provided so that the assumptions underlying the model are transparent, and to provide a template to facilitate comparable analyses in different institutions.
Defining turnover rates
Although there is some debate in the literature about how to compute turnover rates,6,15–18 we defined turnover as the number of faculty members in each department who left the UA COM in any given year (numerator), divided by the total number of faculty members who held appointments in the department during the same year (denominator). These numbers were obtained (by A.W.) from the college of medicine faculty database for five fiscal years, from July 1, 1999 through June 30, 2004, to obtain an average annual turnover rate for each of the 19 departments.
Only salaried faculty who were >50% time, whose primary appointments were in the UA COM, and who held a rank of assistant, associate, or full professor were included. Faculty members on all tracks (tenure, clinical, and research) were included in these calculations to provide a complete picture of turnover at the institution and to more accurately estimate losses in clinical revenues. The 19 faculty members who switched departments but remained in the college (including 17 who formed the nucleus for a new department) were not considered to have left because their expertise was still available to the college of medicine. We obtained information on the faculty member’s rank and gender to assess variation in turnover in relation to these characteristics.
Calculating turnover cost
The cost of replacing faculty who left was estimated using searches completed in 2005 in two of the largest clinical departments: medicine and surgery. These departments were selected because they could provide information on three categories of faculty: generalists, surgeons, and nonsurgical subspecialists. A spreadsheet was created by the department administrator (D.F.) and human resource coordinator (C.C.) in the department of medicine that included all categories of costs associated with faculty turnover (described below), which was used as a guide by the department of surgery in its own calculations. We tallied costs separately for each category of faculty in both departments. Costs were not calculated for basic science departments because too few faculty left during the study period to obtain reliable estimates.
We ascertained three types of costs for each search (Table 1): lost clinical income costs, recruitment costs, and hiring costs. Lost clinical income costs were defined as the net clinical income lost to the clinical department and the faculty practice plan from the time the faculty member left until the position was filled with a new faculty member. (This calculation, although inclusive of physician income for services rendered in the hospital, did not include lost revenue to the AMC’s major teaching hospital, because this was beyond the scope of the present study.) We calculated this cost as the difference between the revenues that would have been generated each month the position was vacant according to historical patterns, minus the expenses associated with that faculty member (i.e., base salary, incentives, employee-related expenses [ERE], other fringe benefits, malpractice, assessed practice overhead, and assessed dean’s and university “taxes”) per month. We estimated revenues from the physician’s expected monthly clinical collections calculated from budget and/or historical financial data. In the department of medicine, for example, estimated monthly collections for a generalist were $17,000 and monthly expenses were $10,000, resulting in a net loss to the AMC of $7,000 per month per generalist faculty member lost. This difference was multiplied by the number of months the position remained vacant, to calculate the lost clinical income cost to the department for each faculty member who left.
Recruitment costs included the hourly salary and ERE of each search committee member, multiplied by the number of hours spent on recruitment activities (phone and on-site interviews, candidate meals, attending guest lectures, reviewing applications, etc.) for each search. Other recruitment costs included lodging, airfare, transportation, and meals for candidates and their families. Costs associated with advertising all positions were added to these costs to calculate the total recruitment costs for each department.
Hiring costs included expenses associated with bringing each new faculty member to the AMC, such as sign-on bonuses, start-up costs, relocation expenses, and laboratory renovation. We also included expenses associated with staff time spent orientating each new faculty member and credentialing new clinical faculty members (for both hospitals and health plans), by calculating the average hourly salary of staff members involved in faculty orientation and credentialing and multiplying this number by the total number of hours spent on each new faculty recruit. (The cost of “spousal hires”—the expenses associated with recruiting the professional spouse of a desired faculty recruit—were not included, because these data were not available.)
These three costs were summed (by A.W.) for each type of faculty member replaced in 2005 for the two departments. This sum was then divided by the number of faculty members of each type to estimate the average cost of faculty turnover by faculty type in the medicine and surgery departments.
Table 2 shows the average turnover rates in all clinical departments by rank for each fiscal year from July 1, 1999 through June 30, 2004. Turnover was highest among assistant professors, averaging 10.1% per year, and lowest among full professors, averaging 2.9% per year. There was substantial variability in turnover between years, with turnover being almost 70% higher in FY2001 compared with FY2004 (8.3% versus 4.9%). This variability was particularly evident for assistant professors, among whom turnover rates were almost double in FY2001 compared with FY2004. Turnover was more constant among full professors. There were no significant gender differences in turnover, with annual turnover rates being 7.1% among women compared with 6.2% among men (χ2 = 0.5, P = .47).
There was substantial variability over time within individual departments (data not shown). For example, turnover rates in one department ranged from 4.3% in one year to 21.7% in another year.
Table 3 shows the cost for replacing each type of faculty in two of the largest clinical departments, medicine and surgery. It was least expensive to replace a generalist, at $115.554, whereas replacing a subspecialist cost $286,503. Costs for replacing a surgeon were more than half a million dollars, or $587,125. The total costs associated with turnover of these 17 faculty members in the departments of medicine and surgery were almost $7 million, with the average cost for replacing a single faculty member being more than $400,000.
If these costs were extrapolated to all 14 clinical departments by multiplying the average cost for replacing each type of faculty by the turnover in each group, the total cost to all clinical departments of faculty turnover during the five-year period would be more than $45 million.
This study shows that the financial impact of turnover among faculty in clinical departments in one AMC is very high, even when lost hospital and research revenues are not included. The average annual cost of faculty turnover in the departments of medicine and surgery was more than $400,000. In considering the costs to the entire college, it must be recognized that each department has a distinct salary structure and will be differentially influenced by faculty turnover. Nevertheless, insofar as the costs calculated for these departments are representative of all clinical departments, the costs of turnover for the entire college of medicine during the five-year period would be greater than $45 million. Even a 10% reduction in annual faculty turnover rates (from 6.4% to 5.76%) would have resulted in estimated annual savings of more than $900,000 in our institution.
This study also shows that, at our institution, faculty turnover varies over time and by rank, with the highest rates seen among assistant professors. Turnover at our institution was lower than the 9% reported by Waldman et al,14 but it was comparable, albeit from a later time period, with that of medical schools nationwide from 1995 to 1999, as reported in the Association of American Medical Colleges’ most recent turnover report.19
Our estimates of the cost of faculty turnover in clinical departments are somewhat higher than those obtained in other published studies (Table 4), but estimates from across the literature are quite varied. It is likely that much of the variability between schools may be attributable to the inclusion of different variables in the calculations. Bickel et al20 recommended that measures of faculty turnover costs include advertising, search firm, and/or administrative costs; interview travel expenses; faculty and search committee time for interviewing; overtime for other staff to fill in during the hiring process; orientation and training time; loss of productivity; and the clinical income costs of lost patients, referrals, and grants. Few studies have managed to include all these factors. Further, it is difficult to assess the impact of included variables on cost estimates because the variables are often poorly defined. For example, Waldman et al21 included hiring, training, working, and termination costs, but what is included in each of these categories is not spelled out. We have provided clear definitions of the factors included in each of our calculations so that this study can be replicated at other institutions. To our knowledge, this analysis is the first to include some, albeit not all, lost clinical income costs in its calculations, which may account in part for our higher estimates.
Other factors likely contribute to variability in estimates of costs of faculty turnover. First, market factors influence faculty salaries, but the estimates in Table 4 have not been regionally adjusted. In addition, because faculty rank influences turnover rates, variability between institutions in length of tenure could affect cost estimates. Providing turnover rates by rank, as we have done, will allow future studies to assess how estimates might change depending on faculty demographics. The relative proportion of surgeons and subspecialists to generalists that leave an institution will also influence turnover costs, and this proportion may vary by year as well as by institution. Finally, the small number of positions that underlie cost estimates in all of the studies listed in Table 4 likely contributes to the observed variability between schools. Our own estimates were based on 17 completed replacement searches in two departments. Although including costs of replacement searches in all clinical departments would have been more precise, this was infeasible because appropriate historical data were not available for most departments.
Although it might be expected that turnover among junior faculty is less expensive than turnover among senior faculty, Hobbs et al22 found that the salaries required to lure assistant professors of finance (like academic medicine, a highly competitive labor market) and those of full professors close to retirement were almost equal. In any case, because turnover is highest among assistant professors, the bulk of turnover costs to the institution pertain to junior faculty. Further, the loss of faculty members early in their careers is more expensive because the institution’s investment in that faculty member may not have been recouped. Joiner 23 has estimated that, depending on the start-up package, level of institutional support, and other factors, it takes two to four years for the revenues generated by the faculty member to exceed his or her costs. The large costs associated with turnover among junior faculty provide an important context for discussions of faculty development programs and retention packages. This analysis suggests that investing more in both the development and retention of highly productive faculty members could be cost-effective.
One potential limitation of our study was that we included in the turnover rate all clinical faculty members who left for whatever reason. Van Der Merwe and Miller4 suggest that because turnover studies are undertaken to affect management decisions, only avoidable turnover should be included in the analysis. Avoidable reasons for turnover include low pay, poor working conditions, and problems with leadership at the home institution—areas where management can intervene.4 Unavoidable reasons for turnover are those such as a spouse-imposed relocation to another geographic area, a career change, or a decision to stay home to care for family members.24 Nagowski25 suggests that associate professors are the most critical group to consider in terms of avoidable turnover, as junior faculty may leave because they do not attain tenure and senior faculty members can be expected to retire. Future analyses of turnover costs should try to distinguish avoidable from unavoidable turnover and should calculate costs of turnover specifically among those faculty whom the institution would like to retain.
This study undoubtedly underestimates the true costs of faculty turnover because a number of very real but more intangible costs have not been included. For example, during the final months on the job, a faculty member is measurably less productive, but data were not available either for these costs or for the lost productivity during the time the new faculty member is becoming fully functional on the job.20 Other intangible costs include efforts expended for incomplete or unsuccessful searches, reduced morale among remaining faculty members, and disruption of research/practice/teaching activities during the period of change. In addition, costs of retention packages were not included in our calculations, because those data were not readily available.
On the revenue side, these estimated costs represent only a fraction of the potential cost to the AMC. Research funding is an important source of revenue for the institution, as discussed by Joiner et al,26 as well as a source of prestige. Thus, turnover among faculty members who have research funding would increase the revenues lost from this source. Further, faculty turnover is costly for the other components of the AMC, which can be thought of as an “ecosystem.”27,28 Indirect losses of revenue due to lost patient referrals and subsequent inpatient and outpatient visits (which were not included in our calculations) obviously affect the hospital and the practice plan. For example, if a critical transplantation surgeon leaves, the clinical productivity of the entire multidepartmental team would be compromised, with financial losses across the AMC enterprise that persist until a replacement is recruited and hired. Additionally, there are costs for hiring replacement faculty to the practice plan and to the hospital, which share these expenses with the college of medicine.
Finally, faculty turnover has nonfinancial costs for all components of the AMC ecosystem, because faculty turnover can have a long-term impact in terms of patients lost29 and reduced teaching and research opportunities. Also, it can affect the reputation of the AMC. Ideally, future research will consider all of the costs discussed in this paper, including lost hospital and research revenue.
In conclusion, our analysis shows that turnover among faculty at one AMC varies by rank and specialization. There are some positive aspects of employee turnover, such as the fact that more junior replacement workers can be hired at lower salaries and reduced benefits, and they bring new, fresh ideas into organizations.1 Nevertheless, the costs of this turnover, even using incomplete estimates, are impressive, being more than $400,000 annually for two of the largest clinical departments. The magnitude of these costs would seem to warrant substantial efforts to foster faculty success and retention.
The authors thank Lloyd Shinn and Bruce Saul, for assistance with data management and analysis, and Keith Joiner, for helpful comments on a previous version of the manuscript.
1 Cavanaugh SJ. Nursing turnover: Literature and methodological critique. Adv Nurs. 1989;14:587–596.
2 Hom PW, Griffeth RW. Employee Turnover. Cincinnati, Ohio: South-Western College Publishing; 1995.
3 Irvine DM, Evans MG. Job satisfaction and turnover among nurses: Integrating research findings across studies. Nurs Res. 1996;44:246–253.
4 Van Der Merwe R, Miller S. The measurement of labour turnover: A critical appraisal and a suggested new approach. Hum Relat. 1971;14:233–253.
5 Hall TE. How to estimate employee turnover costs. Personnel. 1981;58:43–52.
6 Racz S. Finding the right talent through sourcing and recruiting. Strategic Finance. 2000;82:38–44.
7 Phillips JD. The price tag on turnover. Pers J. December 1990;69:58–61.
8 Bahrami B. Factors affecting faculty retirement decisions. Soc Sci J. 2001;38:297–305.
9 Dynlacht JR, Dewhirst MW, Jall EJ, Rosenstein BS, Zeman EM. Toward a consensus on radiobiology teaching to radiation oncology residents. Radiat Res. 2002;157:599–606.
10 Hall JG. The challenge of developing career pathways for senior academic pediatricians. Pediatr Res. 2005;57:914–919.
11 Kohrs FP, Mainous AG 3rd. Retention of family medicine faculty development fellows in academic medicine. Fam Med. 1999;30:23–27.
12 Wenger DCK. Conducting a cost-benefit analysis of faculty development programs: Its time has come. Acad Phys Sci. May/June 2003;1:6–7.
13 Scott K. Physician retention plans help reduce costs and optimize revenues. Healthc Financ Manage. January 1998;52:75–78.
14 Waldman JD, Kelly F, Arora S, Smith HL. The shocking cost of turnover in health care. Health Care Manage Rev. 2004;29:2–7.
15 Billingsley BS. Special education teacher retention and attrition: A critical analysis of the research literature. J Spec Educ. 2004;38:39–55.
16 Bluedorn AC. A taxonomy of turnover. Acad Manage Rev. 1978;3:647–651.
17 Jones CB. Staff nurse turnover costs: Part II, measurements and results. J Nurs Adm. April 1990;20:27–32.
18 McConnell CR. Analysis and control of employee turnover. Health Care Manag (Frederick). 2007;26:84–94.
19 Yamagata H. Trends in faculty attrition at U.S. medical schools, 1980–1999. AAMC Analysis in Brief. 2002;2(2).
20 Bickel J, Wara D, Atkinson BF, et al. Increasing women’s leadership in academic medicine: Report of the AAMC project implementation committee. Acad Med. 2002;77:1043–1061.
21 Waldman JD, Yourstone SA, Smith HL. Learning curves in health care. Health Care Manage Rev. 2003;28:41–54.
22 Hobbs BK, Weeks HS, Finch JH. Estimating the mark-to-market premium required to fill vacant business school faculty lines: The case of finance. J Educ Bus. 2005;80:253–258.
23 Joiner KA. A strategy for allocating central funds to support new faculty recruitment. Acad Med. 2005;80:218–224.
24 Abelson MA. Examination of avoidable and unavoidable turnover. J Appl Psychol. 1987;72:382–386.
25 Nagowski MP. Associate professor turnover at America’s public and private institutions of higher educaton. Am Econ Spring. 2006;50:69–79.
26 Joiner KA, Hiteman S, Wormsley S, St Germain P. Timing of revenue streams from newly recruited faculty: Implications for faculty retention. Acad Med. 2007;82:1228–1238.
27 Kaslow NJ, Mascaro NA. Mentoring interns and postdoctoral residents in academic health sciences center. J Clin Psychol Med Settings. 2007;14:191–196.
28 Olsen L, Aisner D, McGinnis JM. The Learning Healthcare System: Workshop Summary (IOM Roundtable on Evidence-Based Medicine). Washington, DC: National Academies Press; 2007.
29 Berger JE, Boyle RL Jr. How to avoid the high costs of physician turnover. Med Group Manage J. November–December 1992;39:80,82–84,86 passim.