Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
CHAPTER 3 THE FUTURE LABOR MARKET FOR BIOMEDICAL AND BEHAVIORAL RESEARCH PERSONNEL OVERVIEW Labor market projections for this report were expanded to include demand from industry, government, and other nonacademic sources. Labor demand projections reflect increased attrition due to the aging of the work force. The mode! has also been refined in its reflection of the various segments of the market for biomedical and behavioral scientists. Finally, the committee attempted to extend its projections to the year 2000. For biomedical scientists, the projected current Ph.D. production is inadequate to meet growing demand. Employment growth is driven by industry, which may become the largest employer by 1995. Jobs in R&D increase more rapidly than in non-A&D. Because of demand growth and higher attrition, Ph.D. production must increase. For behavioral scientists, the labor market should remain in approximate balance. Growing demand for clinical psychologists may draw behavioral Ph.D.s out of R&D. The market for nonclinical psychologists is projected to be in approximate balance in the l99Os, and the model indicates a stable market for other behavioral scientists--anthropologists, sociologists, audiologists, and speech pathologists. The labor market for physician/scientists is more difficult to project. Recent studies indicate that the demand for clinical investigators has been overstated in the past. However, a number of factors suggest that the demand for well trained physician/scientists will increase in the future. Extended projections to the year 2000 reflect similar trends. Demand for biomedical scientists could be twice the current level of biomedical Ph.D. production. The market for behavioral scientists is in better balance with supply unless growth is higher than expected. The demand for physician/scientists could grow substantially, particularly if significant numbers of M.D.s are drawn into basic research in the biomedical sciences. PRIOR COMMITTEE PROJECTIONS AND CURRENT METHODOLOGY The legislation mandating this study requires the committee to assess "the nation's overall need for biomedical and behavioral research personnel.' Past committees have defined this "need" in labor market terms--that is, how many biomedical and behavioral researchers will be "needed" in the future to fill academic demand? Job openings were determined by growth in the number of academic positions and faculty attrition. These projections were developed for the near term; the 1985 committee report included projections to 1990. This report expands the earlier analysis in several ways: In almost every biomedical and behavioral field, the major source of historical and projected employment growth is in nonacademic sectors, iSection 489 of P.L. 99-158. 51
primarily private industry. This report expands the labor market analysis to include industry, government, hospital, and other nonacademic sources of labor demand for biomedical and behavioral scientists.2 2. This analysis develops separate projections for the labor market in general and for that subsection of the labor market associated with scientists whose primary work activity is R&D or the management of R&D. 3. Given concern over the graying of the work force, the current analysis includes a demographic/economic model for estimating scientist attrition due to death, retirement, and net occupational movement. Attrition in this model is a function of the age and experience structure of the scientist work force. 4. ~ The analysis brings labor supply into the labor market assessment. This was not done in earlier reports. 5. Given that the median time to complete a biomedical Ph.D. has grown from seven years in the late 1970s to eight years in 1987, the 1997 biomedical scientist labor market will be influenced by student decisions and NRSA policy in 1989. In the behavioral sciences, median time to Ph.D. has increased from approximately S.5 years to 10.5 years during the same period. These time lags argue for a longer horizon of analysis. Consequently, the current study projects labor market variables to the year 2000. THE MODEL Figure 3-1 is a schematic drawing of the labor market assessment model used in this report.4 The stock of scientists in time period t is characterized by biological age (years since birth) and career age (years since degree). Historical data provide estimates of the deaths and retirements by biological age; these scientists are removed from the stock. Those who do not retire or die can also leave the field for other employment; this too is 2Although the NRSA program is concerned only with research personnel, it is necessary to take account of the total demand for biomedical and behavioral scientists to ensure that an adequate supply is available. Rapid employment growth in the industrial sector could create hiring difficulties in the academic sector even if academic sector employment is stable or declining. 3Despite the more secular nature of the model, it is unlikely that shifting demographics will substantially affect the analysis. The National Center for Education Statistics' (NCES) demographic mode! projections indicate that M.D.s and first-professional degrees will "increase slightly or remain stable" through 1998. NCES also projects college enrollments to decline from 12.56 million in 1988 to 12.17 million in 1998 (midcase). These points indicate that demographics will not have major influences on supply of Ph.D.s or demand for faculty during the projection period. See NCES, Projection of Education Statistics to 1997- 199S, Washing-ton, D.C.: U.S. Government Printing Office, 1988. 4For a more complete discussion of the labor market assessment model, see Joe G. Baker "Biomedical/Behavioral Cohort Model: A Technical Paper," in Volume III of this report. 52
Scientists in time t Characterized by 1. Biological Age 2. Career Age Career Age Net Outmig tl dependent upon _L "Surviving" Scientists in time t+1 Scientists in time t+1 Characterized by 1. Biological Age 2. Career Age Figure 3-1. Labor market assessment mode Demand for Scientists 1. R&D Spending 2. Enrollments 3. Other Deaths 8 Retirement dependent upon Biological Age New Hires by 1. Biological Age 2. Career Age - 1. rat Required. Scientists Stock t+1 1. Academic 2. Industry 3. Government 4. Hospitals 5. Other r NOW Hires= Required ~Surviving. - Supply of Scientists ( 1. Readiness `` 2. Recruitment 3. Retention _ ~
assumed to be a function of career age.5 Estimates of migration presented here are net (outmigration less inmigration from other fields) and are estimated from historical data. The surviving scientist stock is available for employment in period t+l. The required scientist stock in period t+l is estimated from submodels that link demand for scientists to the demand for the goods and services that scientists produce (e.g., conducting R&D or training graduate students). The submodels were constructed from demand equations that were estimated from historical data. For example, non-A&D biomedical industrial employment was estimated to be a function of real constant dollar output of the pharmaceutical industry; R&D biomedical industrial employment was estimated to be a function of real constant dollar, private sector, health-related R&D spending. These demand submodels vary by discipline (biomedical, behavioral, clinical), sector (academic, industrial, government, etc.),-and activity (R&D, non-R&D). The difference between the surviving scientist stock and the required scientist stock in period t+l represents the number-of job openings that must be filled by new entrants. These job openings are compared to supply to compute "vacancy ratios," (i.e.,~the number of job openings per new Ph.D.~. Increases or decreases in future vacancy ratios from historical ratios give one a sense of changes in the projected demand/supply balance in the scientist labor market. This same basic analysis is replicated for the R&D subsector by comparing R&D job openings to postdoctoral "graduates." Obviously, the number of job openings is dependent upon ~ wide range of other variables, including health R&D expenditures, general economic growth, wage rates, and other labor market factors. Likewise, decisions to enroll in graduate school are based in part on current student support, starting salaries, expected future earnings, earnings available in alternative careers, and other factors. In the simplified approach used here, however, we have implicitly assumed that these excluded variables are constant; we are primarily concerned with major shifts in the demand and supply of scientists through time. While the resulting analysis takes little account of the responses of institutions and individuals to changing labor market conditions, the projections contained in this chapter do serve to highlight the consequences of current trends and suggest courses of corrective action.6 Where possible, we do discuss potential labor market adjustments to changing conditions. THE FUTURE LABOR MARKET FOR BIOMEDICAL SCIENTISTS The demand for biomedical scientists is derived from demand for the goods and services they produce, primarily biomedical research and instruction. In order to estimate the future need for biomedical scientists, we must estimate the level of demand for these goods and services, e.g., the level of health-related R&D to be conducted and the levels of undergraduate and graduate enrollment. Massive new public health efforts in areas such as AIDS research could increase the demand for health research personnel significantly. 5This is based upon the assumption that there is a "career pattern" of outmobility based upon years since Ph.D. (e.g., the probability that one would leave his/her degree field for other employment is small immediately after receipt of the degree; but in later years scientists may move into management and administration). The reader is referred to the current labor market section for estimates of these outmigration rates. 6An econometric approach to projecting the biomedical scientist labor market that included prices and feedback mechanisms produced results consistent with this chapter. See Joe G. Baker, "The Ph.D. Supply Crisis: A Look at the Biomedical Sciences," paper given at the Western Economics Association Meetings, Lake Tahoe, Nevada, July 21, 1989. 54
Private sector research, especially in the emerging biotechnology industry, has the potential to command an increasing portion of limited biomedical scientific resources. Demand for teaching faculty is dependent upon the enrollment levels in graduate programs as well as on university R&D. As the biomedical scientist cadre matures, attrition from death and retirement will increase the demand for replacement in these areas. Growth in Employment In order to provide projections of future demand for biomedical scientists, the committee developed the following scenarios, which are summarized in Table 3-1: Low Case: The low-case scenario corresponds to a conservative retrenchment from current trends: cuts are imposed on growth rates of federal health R&D, and the private sector retreats from current growth levels. In this scenario, both federal and private health-related R&D slow to approximately half their current rates of growth. Graduate and undergraduate enrollment in biomedical sciences decline by 1 percent annually. 2. Mid Case: The mid-case scenario is largely "status quo" with regards to both federal and private health- related R&D funding. Enrollments are assumed to be stable at 1987 levels. High Case: The high-case scenario is based on large increases in health- related R&D funding in such areas as AIDS and alcohol and drug addiction. Private sector efforts in biotechnology and other areas exceed historical levels. Enrollments are assumed to grow by 1 percent annually. TABLE 3-1. Projection Model Assumptions, Biomedical Sciences Real Health R&D Expenditures Historical Assumptions for Future _ Annual Growth Rate Annual Growth Rate 10 Year 5 Year 1 Year Low Mid Federal 2.7°% 5.4% 7.3% 1.5% Private 9~3% 10.3% 9.6% 5.0% Other 6.2X 8.8% -1.1% 2.0% Enrollment in -0.3% -0.2% -0.3°% -1.0% Universities New Ph.D.s 1.4% 0.0% 2.7% -1.0% 2.~X 4.0°% 9.0% 13.0% 3.0°K 4.0% 0.0% 1.0% 0.0% 1.0% SOURCE: Volume II, Tables B1, B3, and Be. Projection assumptions developed by the committee. Projections of the required stock of scientists (Figure 3-1) were developed using these assumptions in the labor demand submodels. ~ Figure 3-2 details the results of these projections. Regardless of the scenario chosen, the same basic trend is evident: industrial employment of biomedical scientists dominates-growth in the future labor market, and in 15 a:
all but the low case it becomes the largest employment sector by c the year 1995. Historical growth !, in biomedical employment averaged 4.S percent annually 3 for the 1973-1987 period; the c projected growth rates for the low, mid, and high cases are 1.8 c percent, 3.6 percent, and 5.2 v percent, respectively. Biomedical scientists engaged primarily in R&D or the management of R&D will grow , faster than the biomedical work e force in general. This is because private industry, in which a high percent of biomedical scientists are engaged in R&D, is projected to grow faster than the other employment sectors. Attrition Job openings are also created by death, retirement, and outmobility. Given the graying of the scientific work force, these openings are expected to increase in the future.7 Figure 3-3 portrays the estimated number of openings from death, retirement, and net outmobility for historical and projected periods (mid-case scenario). Openings due to death and retirement are projected to increase from approximately 1,200 per year in the latter 1980s to 1,650 per year for the 1987-1995 period. Net worker mobility, which created approximately 650 job openings annually in the 1980s, is projected to increase to 950 per year for the 1987-1995 period. For the 1987- 1995 period, therefore, approximately 2,600 new scientists will be required annually just to maintain current employment levels. This 12 ~ 10 - 8l 64 4] -2] O C] LOW Came ; l' Mid Cam High Case Academic Industn~ Govenunent Over Tote Employment Sector SOURCE: Appendix Table A-18. Figure 3-2. Projected annual growth rates for biomedical scientists, 1987- 1995. 3000 - a OF ~5 ;- e] Net Mobdi~ C3 Dka~IRebre. 1973-78 1978-83 SOURCE: Appendix Table Am. 1983 -87 1990-95 Figure 3-3. Historical and projected biomedical scientist attrition, 1973- 1995. 7In 1987, approximately 10,800 employed biomedical scientists were aged 55 or older. Under the mid-case scenario, this will grow to approximately 15,500 in 1995. See Joe G. Baker, "Biomedical/Behavioral Cohort Model: A Technical Paper," in Volume III of this report. 56
figure will rise to approximately 3,400 by the year 2000. Given that about 4,000 Ph.D.s are awarded currently, growth in employment can occur only if the number of Ph.D.s increases. Future Demand and Supply The sum of openings due to attrition and growth is the total number of "new hires" required annually (Figure 3-1~. Figure 3-4 compares the projected labor demand (job openings) to labor supply (new biomedical Ph.D.s). It shows that the number of positions per new biomedical Ph.D. has grown during the 1980s over the 1970s. Assuming that degree production changes in proportion to enrollments (-1, 0, and 1 percent growth for the three scenarios), both the mid-case and high-case scenarios will produce even greater imbalances between job openings and new biomedical Ph.D.s. In summary, the committee projects a future labor market for biomedical scientists that is characterized by increasing imbalance between demand and supply. Because death and retirement are expected to double in the next decade, and because most of the openings created by retirement will be in academe, this presages a time of unprecedented competition between academic institutions and the emerging biotechnology industry. This situation can be avoided by some combination of the following factors: Decreased growth: Current biomedical Ph.D. production could support future demand if future growth is in the neighborhood of 2 percent annually. However, this is a considerable decrease from the 4.S percent historical growth rate in biomedical Ph.D. employment. a . a on a o - A: oils 1 CPhD s l 7eQ0 - ~ 6000 - 5000 - 4000 - l ~ ~ 3000 - 2000 - 1000 - ~ ~ , , I , , , o Biomedical Scientists oils C//~ r/f ire ~ 1 Ph.D.s 1973- 1979 . 1979- 1987 SOURCE: Appendix Table A-l9. . . . oils - ~A_Y~ DAM ma_ ~A~W9YW~ RZ . BY . 1~ 1 IA C~1 Ph.D.s 1 ~ Al 1 990- 1 995 EN Ph.D.s ~ Grown g' Net Mobility Death/Retired Figure 3-4. Average annual biomedical job openings and new Ph.D.s, 1973-1995. 57
o o o o o o Increased production: Biomedical Ph.D. production could support faster growth if the number of degrees increases from the current level of 4,000 per year to the 5,000-6,000 range. Given current trends in first-year graduate enrollments in biomedical sciences, however, there is likely to be no major increase in Ph.D. production until the mid-199Os.8 Decreasing graduate school attrition anal time to degree: The high rates of graduate student attrition and lengthy time to degree increase the sluggishness and decrease the productivity of graduate schools in the production of new scientists. Little is known about the causes of these problems; therefore it is difficult to offer solutions other than more research If time to the doctorate and attrition could be reduced, however, the benefits would be more immediate than increasing enrollments. Increases! retention of women: Female scientists have much higher dropout rates than men (see Chapter 2~. If programs were developed to increase full- time science participation by females who hold Ph.D.s, effective labor supply would increase. Recruitment of outside Ph.D.s: Inmobility of Ph.D.s from other fields could increase. Recruiting new entrants from other fields is viable so long as there are people in other fields to draw on. However, most studies of the science and engineering labor market in the 1990s project a similar imbalance between new Ph.D.s and job openings in other fields. Therefore, it may become increasingly difficult to draw entrants from other fields. Also, these entrants may not be as productive as scientists with degrees in biomedical sciences, and there may be other costs associated with hiring them, such as training costs or increased supervision requirements. Very little is known about these tradeoffs, however. Decreased attrition: Patterns of outmobility and retirement could change; perhaps incentives could be developed to affect this behavior such as rewarding delayed retirement. Recruitment of foreign Ph.D.s: Foreign nationals with permanent visas could help meet U.S. science need. THE FUTURE LABOR MARKET FOR BEHAVIORAL SCIENTISTS The behavioral sciences are made up of a group of disciplines that includes clinical psychology, nonclinical psychology, anthropology, sociology, and speech pathology (these latter three are grouped together as "other behavioral sciences"~. Clinical psychologists are mainly involved in patient care; the behavioral science disciplines that are it&D-oriented are nonclinical psychology and other behavioral sciences. Waif one assumes that biomedical Ph.D. output will change in parallel with lagged first- year biomedical graduate enrollment change, one can estimate Ph.D. labor supply through 1995 based upon current enrollment, although historically such a mode! yields a poor fit. Using this assumption does not substantially change the results, however: the lagged enrollment model shows Ph.D. output in the biomedical sciences increasing very slowly, from 4,000 in 1987 to only 4,100 in 1995. See Joe G. Baker, "The Ph.D. Supply Crisis: A Look at the Biomedical Sciences." 58
The future labor market for clinical psychologists is projected to be characterized by increasing imbalance of demand over supply. Given that most of this growth is practice-related, the direct consequences for behavioral science research personnel are slight. However, clinical practice will offer an attractive employment opportunity for behavioral scientists in general, and could be expected to draw scientists from R&D activities. Nonclinical Psychology Nonclinical psychologists make up approximately 30 percent of the behavioral scientist work force. Unlike clinical psychologists, their work consists largely of teaching and research. Over 80 percent of nonclinical psychologists work in private industry or academic settings; the remaining 20 percent are distributed among government, hospitals, and other employment sectors. The key assumptions employed for projecting future levels of demand for nonclinical psychologists appear in Table 3-2 and are summarized here: o Graduate enrollment growth in the behavioral sciences is modest or negative under three scenarios: low case (-1.0 percent annually), mid case (0.0 percent), and high case (1.0 percent). TABLE 3- 2. Pro ject i on Mode l Assumpt i ons, Behavi ore l Sci ences H i star i ca l Assumpt i ons f or Future Annua l Growth Rate Annua l Growth Rate Variable 10 Year 5 Year 1 Year Low Mid High ~- Graduate Enrol lrnent i n Universi ties Industrial Employment of Noncl inical Psychologists Nonacademi c employment of Other Behavi ore l Scientists R&D 3.5% -2.2% NonR&D 14.4% 15.7X New Ph.D.s -0-7°/O -1.1% 1~1% 0.2% : -0.3X -1.0% 0.0% 1.~% 6.6% 3.7% 3.2% 2.0% 3.0% 4.0% 38.~% 3.0X 5.0% 7.0% -2.0% 7.0% 10.0X 15.0% -1~8% ~1aO% 0~0% 1~0% SOURCE: Volume II, Tables C1, C10-1, and TabLe A-10 from Joe G. Baker, "Biomedical Cohort Model: A Technical Paper" in Volume III of this report. Projection assumptions developed by the committee. 9In 1987 about 20 percent of employed clinical psychologists had degrees in other fields. However, future movement into the field may be slowed by changes in state certification requirements. 59
o Demand growth for nonclinical psychologists in industry continues to grow under the low case (2.0 percent per year), mid case (3.0 percent), and high case (4.0 percent). This compares to an historical growth rate of 3.1 percent per year for the 1985-1987 period. O Growth in demand for nonclinical psychologists in the government sector is assumed to be 1 percent annually in each scenario. Figure 3-5 displays the resulting projections. Total growth in the employment of nonclinical psychologists averaged 3.1 percent annually for the 1973-1987 period; projected employment growth is expected to range from-0.4 percent to 1.9 percent. In all cases nonclinical psychologists engaged primarily in R&D or the management of R&D are projected to remain at a fairly constant proportion (approximately 27 percent) of the total behavioral scientist work force. Figure 3-6 shows the historical and projected trends in the number of annual openings from death, retirement, and outmobility. Total annual attrition increased from approximately 660 openings in the early 1970s (230 death and retirement and 430 net outmobility) to approximately 1,000 in the late 1980s (430 de-ash and retirement and 570 net outmobility). This level of annual attrition is projected to increase only slightly, to approximately 1,050, for the 1990-1995 period under the mid-case scenario. As shown in Figure 3-7, the number of job openings per new nonclinical psychology Ph.D. has been between 0.8 to 0.9 for the 1973-1987 period. Assuming degree production changes in parallel with enrollments (-1, 0, and +1 percent annual change for the three scenarios), there will be little change from these vacancy ratios in the 1990s. Even the 8 - in o ~ 6 ~ 4 v ~ 2 c, L. _ 0 en I: SOURCE: Appendix Table A-20. [] Low Case 81 Mid Case El High Case ~ L -2- , , i> Academic Industrial Goverrunent Employment Sector Other Total Figure 3-5. Annual growth rates for nonclinical psychologists, 1987-1995. 60
high-case scenario projects approximately one job opening per new nonclinical psychology Ph.D. Taken together, Figures 3- 5 through 3-7 show that the labor market for nonclinical psychologists moved from excess supply in the 1970s to approximate balance in the 1980s. This balance is projected to continue into the 1990s, with approximately 80 percent of all job openings coming from attrition. Projections for nonclinical psychology R&D employment indicate a softening of the job market, with fewer R&D job openings per postdoctorate through 1995 than 2000 - 1500 - 1000 - o - 3 500- rat Owls Cl /// PhD.s 1200 - 1000 =, 800 o ~ 600 400 OF ~ 200 ¢ ' Net Mobility e] DeaWRetire. AL '//f/> /~/ O- _ 1 l ~ // _ 1973-78 1978-83 1983-87 1990-95 SOURCE: Appendix Table A-9. Figure 3-6. Historical and projected average annual attrition for nonclinical psychologists, 1987-1995. Nonclinical Psychologists Ides Openings ~ l 1 PhD.s __ Logs ~ ~ ~ _ 1 l ~ I l ~ ~ 1973- 1979 . 1979- 1987 . 1990- 1995 SOURCE: Appendix Table A-21. E3 Ph.D.s @' Growth Net Mobility [] Dea~JRetire. Figure 3-7. Annual average job openings and new Ph.D.s in nonclinical psychology; 1973-1995. 61
during the 1973-1987 period, assuming that postdoctorate production remains near its 1987 levels. In summary, the future labor market for nonclinical psychologists will be characterized by approximate balance unless (1) future rates of growth in employment exceed the high-case scenario rate of 1.9 percent annually; (2) the award of doctoral degrees declines substantially (by approximately 3.0 percent per year) from its current level; and/or (3) the clinical psychology market captures a larger share of nonclinical psychologists. The potential for demand (job openings) in excess of supply (new nonclinical psychology Ph.D.s) is even greater in the late l990s (see "Prospects for the Long Term," below). Other Behavioral Scientists Other behavioral scientists include Ph.D.s in anthropology, sociology, audiology, and speech pathology. In 1987 other behavioral scientists made up about one in five behavioral scientists. The "other" group is dominated by academic employment: almost 85 percent of total employment is in academic work, with the remaining 15 percent scattered across the other employment sectors. Given the predominance of academic work, future employment will be closely linked to academic demand. The assumptions used to project future employment, shown in Table 3-2, are as follows: o o Annual growth in graduate enrollment remains modest in all three scenarios: low case (-1.0 percent per year), medium case (0.0 percent), and high case (1.0 percent). As a result, academic employment of "others" grows very little. Employment of "others" in nonacademic employment sectors grows more rapidly. The low-, medium-, and high-case growth rates are 7, 10, and 15 percent annually for non-A&D scientists, respectively, and 3, 5, and 7 percent annually for scientists engaged in R&D or the management of R&D. Figure 3-8 displays the resulting projections. Total growth in the employment of other behavioral scientists averaged 4.8 percent annually for the 1973-1987 period. Projected growth of employment ranges from -0.3 percent to 3.4 percent. In all but the low case, employment in R&D or the management of R&D grows more slowly than non-A&D employment. Attrition is projected to grow to approximately 700 per year for the 1987- 1995 period (300 death and retirement, 400 net mobility) from about 650 in the late 1980s. The number of job openings available to each new "other" Ph.D., which had been approximately 0.9 for the 1973-1987 period, would increase only under the high-case scenario. R&D vacancy ratios through 1995 show little change from current values. In summary, the future labor market for other behavioral scientists will be characterized by relative stability unless (1) the future rate of growth in employment exceeds 2.0 percent annually; and/or (2) de0gree production decreases from its current level of 950 annually to less than 700 annually.] If time to the doctorate in the behavioral 10This assumes that degree production changes proportionately with enrollments. First- year graduate enrollments in other behavioral sciences have fallen, from 4,470 in 1980 to 3,995 in 1987. Although the relationship between first-year enrollments and lagged Ph.D. production is poor, these declines could translate into lower Ph.D. output during the 1990s. If this occurs, a portion of this supply shortfall could be absorbed by increasing faculty productivity: in 1975, 5.4 B.A./B.S. degrees in other behavioral sciences were awarded per faculty member; in 1987 this statistic had fallen to 1.6. 62
sciences continues to grow, degree production could decrease. A combination of these events could also produce a market imbalance. In the R&D sector, movement of postdoctoral students into the labor market at the current level of approximately 120 annually would satisfy long-term R&D demand. The committee's projections of greater growth in demand for biomedical than for behavioral scientists disagrees with the findings of a recent study by Bowen and Sosa. They examined the prospects for arts and sciences faculty and concluded that impending shortages would be greater in the social sciences and humanities than in the natural sciences. One reason for this disagreement is the fact that this report focuses on research personnel and thus excludes the fastest-growing area of the behavioral sciences--clinical psychology. Another is that they make assumptions about changes in student values that our committee did not malted - E a: - o. ~8 - e 16 14 12 10 8 6 4 2 O -2 4 i _ ~1 . ~ ~ Low Case gel Mid Case ED Him Case _~1 Academic All Other Total Employment Sector SOURCE: Appendix Table A-22. Figure 3-~. Projected annual growth rates for other behavioral scientists, 1987-1995. THE FUTURE LABOR MARKET FOR PHYSICIAN/SCIENTISTS Demand for Clinical Investigators Estimating the demand for clinical investigators would be a difficult task in the best of circumstances. Terms such as "scientist," "investigator," and "research" are poorly defined and frequently misapplied. Past reports and projections by this committee have taken a similar misstep by virtue of equating "clinical faculty" and "clinical investigators." More recent evidence indicates that the demand for clinical investigators may be lower than previously estimated. William G. Bowen and Julie Ann Sosa, Prospects for Faculty in the Arts and Sciences. A Study of Factors Affecting Demand and Supply, 1987-2012, Princeton, NJ: Princeton University Press, 1989. bother relevant differences are that Bowen arid Sosa focus on faculty and do not explicitly take account of research. They also use a proportional assumption concerning faculty and enrollment that we consider to be unrealistic: in their baseline case, they find a tighter market as the number of college-aged students stops declining and then begins to Increase. 63
The American Association of Medical Colleges (AAMC) Faculty Roster indicates that 69.6 percent of academic physicians are engaged in research activity, but several studies have concluded that this significantly overstates the number of medical school physician faculty involved in research.~3 Levey, et al., analyzed surveys of faculty in research-intensive clinical departments (internal medicine) and arrived at a figure of 34 percent, later revised (using stricter criteria) to 20 percent. Sherman's reexamination of the same data yielded a median R&D effort of 25 percent. Institute of Medicine data, based on seven-day diaries of 3,400 faculty members in 20 schools of medicine and more than 100 teaching hospitals taken in the spring of 1975, indicated that 20 percent were involved in research, either alone or in combination with teaching and patient care; this survey did not include R&D administration as a specific reported item. It seems reasonable, based on these studies, to assume that 20-25 percent of all clinical faculty members would be properly classified as clinical investigators. This places the demand for physician/scientists in medical schools at between 10,000 and 12,000; the total demand would be something on the order of 16,000 to 1S,000 (10,000 to 12,000 medical school, 434 NIH, and 5,651 other). Similar numbers can be arrived at by extrapolating from the NIH-documented population of clinical investigators. Their numbers have increased by 50 percent in the last 17 years, from 4,300 in 1970 to 6,400 in 1987, or roughly 2.4 percent per year. This figure would represent a lower boundary on the number of academically-based research M.D.s. Assuming that the ratio of M.D. grantees to total M.D. investigators is the same as that for Ph.D. investigators--15,589 grantees out of 22,751 Ph.D.s in academic it&D--then by analogy there would be 9,330 M.D.s engaged in academic R&D. Because patient-related research is poorly supported by NIH, the figure is probably somewhat higher, plausibly in the 10,000 to 12,000 range suggested above. New research efforts requiring M.D. training would increase these demand figures to some unknown extent. New, well-funded initiatives in such areas as health services research, outcome assessment, and epidemiology, in both academic and industrial settings, plus renewed interest in occupational and environmental health, would require additional M.D. participation in areas where medical training has traditionally given little emphasis. All of this suggests that the time has come to take a new look at the clinical investigator demand model, particularly the demand equations, and to develop a more accurate model on a more rational basis. For example, almost 50 percent of medical school revenues now come from patient care sources; this is hardly an indicator of demand for clinical investigators. Medical student enrollment has leveled off, but the mix of faculty effort may be more directly influenced by house staff and fellow enrollments. The clinical scientist employment reported by NIH appears low--where, for example, are Veterans Administration (VA), Food and Drug Administration (FDA), and other federally-employed investigators reported? To reflect how the demand for clinical investigators is created, and how decisions to employ them are made would require a model in which demand is expressed as an explicit function of such variables as NIH intramural and extramural budgets, FDA budget, pharmaceutical industry R&D expenditures, and state research appropriations to medical schools. ~3See G. S. Levey, et al., "Postdoctoral Research Training of Full-Time Faculty in Academic Departments of Medicine," Annals of Internal Medicine, vol. 109, no. 5 (September 1988), pp. 414-418; Charles R. Sherman, "The NIH Role in Training of Individual Physician Faculty: A Supplementary Analysis," NIH memorandum dated March 30, 1989, and Institute of Medicine, Medicare and Medicaid Reimbursement Policies Study, Washington, D.C.: National Academy Press, 1976. 64
The Supply of Clinical Investigators There is no reason to assume that the number of M.D.s available for clinical investigation will increase because of an oversupply of physicians who find patient care less satisfying for reasons of economics,~competition, and/or personal preference. For example, the demand for M.D.s in the health care system's bureaucracy is not trivial, and many physicians may decide to become health care executives or take other career paths that have little or nothing to do with their training as ~D.s. Also, in spite of their best efforts, some of them may not have the intellectual wherewithal, imagination, and drive required to compete in the world of clinical investigation. For all of these reasons, it is not logical to assume that research is more attractive than patient care, nor is there any basis to assume a reduction in the research attrition rate. The current supply model arbitrarily sets a fraction of the total physician manpower supply as the supply of clinical investigators--a procedure that is simplistic to an unsatisfactory degree. Supply also bears some relationship to the students' responses in the AAMC surveys regarding their future intentions, to the proportions that select subspecialty research training, and most importantly, to the opportunities for training and careers in clinical investigation. A more sophisticated supply model, one that has both relevance and some basis in available data, would express supply as an explicit function of such variables as the numbers of medical students, house staff, and fellows; medical student interest in clinical investigation; and training positions in clinical investigation. It is equally important to recognize how various disciplines change and open up whole new areas that cannot be predicted by these demand and supply functions. Examples include the impact of new diseases such as AIDS, the growing concern over alcohol and drug abuse in our society, and the growing geriatric population with all of their medical and social needs. Further, the demand for outcome assessment of health care from the standpoint of efficiency will also create a demand for more and more physician/scientists who have the tools to address these problems. PROSPECTS FOR THE LONGER TERM: 1995 TO THE YEAR 2000 The preceding analysis of the labor markets for biomedical and behavioral scientists focused upon the period 1987-1995. As was discussed in the text, the policy options for this period are somewhat limited, given that the average time to complete the Ph.D. is in excess of eight years--that is, the "policy window" for the 1995 labor market is quite small. What of the longer term--that is, to the turn of the century? Although the projections become more uncertain, the influence of policy as a corrective mechanism becomes stronger. The following discussion of a "status quo" future is based upon the assumption that the key variables--mainly enrollments, degree production, and employment growth--continue on the paths established during the late 1980s. Biomedical Scientists If recent trends in employment growth (4.1 percent annual rate) and attrition continue, job openings for biomedical scientists will exceed 8,000 per year by the year 2000. Replacement of scientists lost to attrition alone would require approximately 2,900 scientists. This is more than double the current level of Ph.D. production in the biomedical sciences, about 4,000 per year. Even if one assumes that growth falls to only 2 percent annually, job openings in the year 2000 would require 5,000 new Ph.D.s. 65
Behavioral Scientists Unlike the market for biomedical scientists, the year 2000 labor market for behavioral scientists appears to be more in line with current levels of degree production. There exists the potential for imbalances in this market if even modest rates of employment growth are achieved during the l990s; this is certainly a topic that needs to be monitored in the future. Extending recent trends (2.0 percent annual growth) in nonclinical psychology employment growth through the year 2000 results in annual job openings of approximately 1,900 in that year. Current Ph.D. output of almost 1,400 would clearly be inadequate to meet this level of demand. However, it is the committee's opinion that current rates of employment growth will not be sustained; current levels of Ph.D. output should meet projected employment growth of 1 percent annually during the l990s. Employment growth in other behavioral sciences has been virtually nonexistent during the late l980s.~4 Assuming a 1-percent growth rate results in approximately 1,000 job openings in the year 2000, this is in line with 1987 production of approximately 900 Ph.D.s. Physician/Scientists The current number of employee physician/scientists is estimated by the committee to be in the range of 15,000 to 20,000. Given historical trends in public and private biomedical R&D funding, it is probable that demand for physician/scientists will grow substantially in the l990s.~5 However, it is difficult to project accurately the long-term labor market for physician/scientists because of the data deficiencies cited above. In addition, if demand for basic biomedical scientists continues to grow (as the committee projects), it is possible that physician/scientists will be asked to supply some of the basic science research effort in the biomedical field. The committee's recommendation to examine the issue of post-M.D. basic science training is consistent with this scenario (see Chapters 4 and 5~. reestimates from the Survey of Doctorate Recipients for total employment of other behavioral scientists was 12,478 in 1983 and 12,736 in 1987. 15Real funding for health-related R&D has been doubling approximately every 12 years since 1960. See Table B-7 in Volume II of this report. 66
