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 TWO APPROACHES TO THE ESTIMATION OF NATIONAL NEED The vitality of the U.S. heals research effort depends upon the availability of scientists who are personally com- mitted to research, have mastered the theories and tech- niques of science, and can communicate their findings and assimilate new knowledge. In He biomedical and behav- ioral sciences, the nation's need for these research scientists is tied closely to problems of human health as well as op- portunities for employment. Previous National Research Council study committees have focused largely on education and employment in the biomedical and behavioral sciences and the role of the Na- tional Research Service Awards (NRSA) program in main- taining an adequate supply of well-trained scientists to maintain stability and efficiency in that system (IOM, 1985. Given that approach, a substantial data base was developed over the years that permitted study committees to monitor trends in enrollment, degrees, employment pat- terns, and funding in certain of the fields. Analytic models of the training system were developed Trough surveys and special studies (see, for example, NRC, 19784. Supply- demand models were generated in the basic biomedical, be- havioral, and clinical sciences to gauge the nation's need for scientists to maintain the balance in the education-em- ployment system (see, for example, NRC, 1975, 1981, 19894. With this report we mark a departure from the activities of previous NRC study committees. In recognition of the dynamic forces that create the demand for these highly skilled investigators, we present information on what we believe to be the key contextual variables that influence the size and quality of the research work force in the biomedi- cal and behavioral sciences: the priority given by the nation to health research; recent major advances in the fundamen- tal knowledge base, which works both to attract young people to scientific careers and to challenge Rem to master 17 the latest scientific and technical developments; and changes in the demographic composition of the biomedical and be- havioral science work force and recent employment experi- ences of its members. Our focus is the development of the science career and He role of He NRSA program in facili- tating career development. On the basis of our review of available information, we conclude Hat He demand for skilled research personnel in He biomedical and behavioral sciences continues to be strong. In He sections that follow, we identify what we believe to be the primary forces influencing the nation's need for skilled scientists in the next several years. HEALTH RESEARCH AS A NATIONAL PRIORITY The remarkable productivity of biomedical and behav- ioral research in the United States is largely a result of na- tional patterns of investment in research and development (R&D). In He past decade alone, U.S. support for heals research and development nearly tripled, growing from a total investment of $9.6 billion in 1982 to $28.1 billion in 1992 (NIH, 1993). This significant growth occurred when other research sectors experienced less dramatic support, often influenced by shifting priorities in federal budgets. The Defense Deparanent's support for R&D, for example, grew rapidly in He early 1980s but peaked in 1986 and then declined. This change is reflected in He relative standing of the life sciences, engineering, and He physical sciences since 1980 (Figure 2-1~. The primary sponsors of heals research in He United States are government agencies, industry (primarily the pharmaceutical industry), and private nonprofit organiza- tions (including foundations, voluntary heals agencies, and medical research organizations). (See Appendix Table F-1~.
MEETING THE NATION'S NEEDS FOR BIOMEDICAL AND BEHAVIORAL SCIENTISTS Basic Research 6.0 5.0 4.0 t 3.0 2.0 1.0 0.0 6.0 5.0 4.0 3.0 2.0 1.0 Physical sciences _ Environmental sciences Engineering ~ , _ , . . ~ Mathematics & _* l 1 1 1 1 1 1 1 1 ! 1 1 1 1980 1982 1984 1986 1988 1990 1992 Applied Research Englneerlng _ . , , . _ . Phvsical scie 0.0 1 980 1 982 1 984 1 986 1 988 1 990 1 992 FIGURE 2-1 Federal obligations for research, by field (in bil lions of constant 1987 dollars). SOURCE: National Science Board, 1994. Of He estimated $28 billion invested in health R&D in 1992, industry sponsored 45 percent, federal government 20 agencies sponsored about 41 percent, and the nonprofit sec tor sponsored 4 percent. 6 Health research has also grown as a share of total na- ~ 12 tional R&D, from about 13 percent in 1980 to 18 percent in ,, 1992. As Figure 2-2 reveals, health R&D has grown as a ~ ~ fraction of total federal investment since 1984. However, health R&D has grown more rapidly as a share of all other national sources of support for research and development. In summary, although government budgets for health-re lated R&D have grown steadily over the past decade, other sectors have absorbed a greater share of the sponsorship of health research. 18 Patterns of Federal Support The principal sponsor of government research in the health sciences in the United States is the National Insti- tutes of Health (NIH). It is estimated that NIH will provide over 90 percent of the $10.9 billion of federal budget for health research and development in 1994. (See Appendix Table Fob. Research highlights from the fiscal 1994 health budget for the federal government reveal that the National Cancer Institute has the largest share of R&D funding within NIH ($2.08 billion in 19943, closely followed by the Na- tional Heart, Lung, and Blood Institute ($1.3 billion). These two institutes account for about one-third of the total NIH R&D budget. Although the administration proposed a 3.2 percent increase over fiscal 1993 levels for each of the 20 institutes and centers within NIH over fiscal 1993 levels, the U.S. Congress responded by increasing fiscal 1994 lev- els by over 5 percent. However, this rate of growth be- tween fiscal years 1993 and 1994 represents a lower rate than that observed in the late 1980s which averaged about 8 percent a year. Components of the NIH budget in 1994 receiving the largest increases included: · the Human Genome Center (23 percent), · the National Library of Medicine (16 percent), · the Fogarty International Center (10 percent), · the National Institute for Allergy and Infectious Dis- eases (8 percent), and · the Division of Research Resources (6 percent). In summary, federal support for health research remains strong although the rate of growth has slowed somewhat in recent years (AAAS, 1992 and 1993~. Coupled with cur- rent economic considerations (described below), it is uncer- tain, however, whether anticipated growth for health re- search will match that observed in the 1980s. Trends in Industrial Support In 1987 the pharmaceutical industry provided almost $5.4 billion and the biotechnology industry provided $1.4 4 - ~ ~ 980 ~ 984 ~ 988 ~ 992 I ~ Federal ~ Non-Federal FIGURE 2-2 Funding of health R&D as a percentage of total R&D, by source. SOURCE: National Science Board, 1994.
APPROACHES TO THE ESTIMATION OF NATIONAL NEED billion for health R&D (OTA, 1988). The Pharmaceutical Manufacturers Association reported that the combined R&D expenditures of its member firms exceeded He total NIH budget in 1989. For several years industry has been viewed as the most rapidly growing sector of health R&D, but re- cent reports suggest that the grown of industrial invest- ments in R&D may be leveling off, a result of corporate restructuring, slowdowns in productivity and economic growth, and other general economic conditions that inhibit industrial spending in activities that yield long-term payoffs rather than short-term gains (GUIRR, 1992~. Other Factors In considering the future of heals research in He United States during the next decade, it is important to realize Hat research progress may be in conflict with or limited by eco- nomic possibilities. These limiting factors include the im- pact of the budget deficit on federal spending, the economy, the unknown effects of health care reform, and the regula- tory costs of research investments. Impact of the Budget Deficit Federal agencies throughout government are experienc- ing the costs and uncertain future associated with the mount- ing federal debt. As a result, efforts are underway through- out the Congress, the Office of Management and Budget, and elsewhere to cut unnecessary government spending and to reduce costs. In recent years there has been little annual increase in the research budgets for NIH after adjustment for inflation, apart from certain areas designated as priori- ties or major initiatives (such as the Human Genome Project). The Economy Industrial investments in health R&D were a major source of the growth experienced in this sector in the 1980s and compensated in part for reductions in federal growth patterns. However, slower economic growth and the lack of short-term market payoffs from prior research invest- ments have raised questions as to whether significant in- creases in private sector funds allocated for health R&D can be sustained in the 1990s. Unknown Effects of Health Care Reform National interest in health care reform has raised many questions about the extent to which fundamental changes in physician-patient relations, co-payment arrangements, and the financing of medical and hospital care will affect the conduct and support of research. The emerging shift in 19 emphasis from diagnosis and treatment of disease to the promotion of good health and the prevention of physical and mental disorders, for example, may deeply influence future research priorities in the health sector. Regulatory Costs of Research Ir~vestmen~ts The conduct of research has become part of a broader social and institutional context, and various social objec- tives or interests have generated a range of regulatory re- quirements and oversight mechanisms that carry additional costs. Issues such as the protection of human and animal research subjects, the protection of He environment (includ- ing the handling and disposal of toxic substances and the siting of hazardous facilities), and the need to ensure re- search integrity and fiscal accountability have raised new questions about the full costs of the research enterprise and the extent of the impact of hidden infrastructure costs on future research budgets. ADVANCES IN RESEARCH Another factor that plays a significant role in influencing the nation's need for biomedical and behavioral scientists is the growth of science. If we start, for example, from the work of the biologists and physicists who first gave us the structure of DNA 40 years ago, we can trace how that very fundamental discovery opened the field of molecular biol- ogy. The introduction of recombinant DNA technology was another milestone that gave us the tools to understand the gene. In addition to being able to sequence genes rapidly, we can clone them and over-express proteins. The tools of physics and computer engineering have allowed the devel- opment of instrumentation Hat enables us to determine He structures of the macromolecules rapidly and accurately. This, in turn, has given rise to the field of structural biol- ogy. The point is that basic science has given us the field of biotechnology, which now allows us to explore biology and medicine in more effective and exciting ways. It is this kind of serendipity Hat has caused science and medicine to advance. The remarkable accomplishments of the biomedical and behavioral sciences are evident in the answers that are emerging to a host of disease problems-such as cancer, heart disease, acquired immune deficiency syndrome (AIDS) and to the revolutionary advances in every branch of medicine. The success of fundamental research in the past few years can only be expected to accelerate the pace of discovery in He near future. In the chapters that follow we identify the recent ad- vances in the biomedical and behavioral sciences which we
MEETING THE NATION'S NEEDS FOR BIOMEDICAL AND BEHAVIORAL SCIENTISTS believe hold promise for attracting bright students to think creatively about the furler development of science and its application to the heals needs of the nation. Certain of these advances have been driven by significant investment in problems of national concern such as substance abuse or violence. However, over advances represent the next steps that will occur Trough the accretion of knowledge or the development of new technologies that make exploration of new problems possible. MARKETPLACE REQUIREMENTS The first report of the National Research Council on the subject of national needs for biomedical and behavioral re- search personnel was issued in June 1975, only 4 months after the Council accepted the task proposed under the NRSA Act of 1974. Because of time constraints, He first study committee devoted its initial report to a description of the organization of the study, an outline of He issues in- volved, and a presentation of the limited data available at that time. Each subsequent study committee updated or enlarged the scope of topics addressed in prior reports and induded some new issues. In organizing He first study, the 1975 NRC committee divided the biomedical and behavioral fields into four ar- eas: (1) basic biomedical sciences, (2) behavioral sciences, (3) clinical sciences, and (4) health services research.2 A panel of experts was formed to assist the committee in each area, and an additional panel was created to guide the data collection and analyses. It was recognized very early in that study Hat He legis- lative request to specify the nationts personnel needs in the fields of biomedical and behavioral research would be im- peded by He difficult problems of definition and classifica- tion. An attempt was made, therefore, in the first report to define each of the four broad areas in terms of He disciplin- ary fields included in them. These initial definitions were revised in subsequent reports, but the problems of taxonomy and determining need at the disciplinary level continue to be among He most intractable ones facing every commit- tee. The major problem, as pointed out in the 1975 report, is that the boundaries between disciplines are difficult to draw. This problem is compounded by the adaptability of biomedical and behavioral scientists and their capacity for mobility within and across fields. This is especially true for transfers from more fundamental to more applied fields and for transfers facilitated by postdoctoral training. Lastly, there is the difficulty of predicting major scientific develop- ments and their potential impact on personnel requirements. In view of these considerations, the recommendations of previous NRC committees have been directed almost exclu- sively to broad areas rather than to disciplinary subgroups. The 1975 study committee commented on the weak 20 nesses inherent in all available manpower models when it first began to assess methods of projecting the labor market. That committee observed Hat supply depended largely on such intangibles as students' perceptions of the prestige, prospects, and value of various careers. Demand, in turn, responded to another set of factors future economic con- ditions, levels of federal funding, and evolving research pri- orities-all of which could change abruptly and unpredict- ably. Nonetheless, during the 1970s He study committees used supply and demand models to assess the relationships be- tween the production of Ph.D. scientists, He need for re- search personnel in universities, and He commitment of funds from various sources to R&D in the basic biomedical and behavioral sciences. However, in 1981, another study committee reported that significant problems existed in making projections from available data. First, the latest data were not always available for use in He model. Sec- ond, changes within specific disciplinary subgroups were not measurable (only the number of people in broadly de- f~ned categories was known), thus limiting the utility of the analysis to estimate "national need". Third, assumptions about job mobility were flawed: it had been assumed in the past Hat turnover in He pool of career scientists in U.S. colleges and universities was principally due to retirement and death. More recent assessments indicated this was not the case, that there was a substantial amount of switching between academic and nonacademic positions. During He mid-1980s, succeeding committees contin- ued to use supply and demand models, introducing certain improvements. In particular, the 1989 study committee enhanced earlier analyses of the labor market for biomedical and behavioral scientists in several ways: · expanded the labor market analysis to include indus- ~y, government, hospital, and other nonacademic sources of labor demand; · developed separate projections for the labor market in general and for scientists working in R&D or management of R&D; · included a demographic-economic model for estimat- ing scientist attrition due to dead, retirement, and net occu- pational movement; · brought labor supply into He labor market assessment; and · projected labor market variables to the year 2000. Another improvement over the years has involved He disaggregation of disciplinary fields for purposes of analy- sis, specifically in the behavioral sciences. In the earliest reports, study committees presented the labor market out- look for Ph.D.s in the behavioral sciences as a whole, wi~- out distinguishing between He subfields of psychology, so
APPROACHES TO THE ESTIMATION OF NATIONAL NEED ciology, anthropology, and speech and hearing science. In 1978 the study committee, realizing that the analysis of the labor market was hindered by treating the behavioral sci- ences as a single entity, separated the data into clinical (clinical psychology, counseling and guidance, and school psychology) and nonclinical fields. This disaggregation en- abled the identification of divergent market trends within the behavioral sciences. Status Report on the Work of the Panel on Estimation Procedures The Panel on Estimation Procedures was asked by the present committee to analyze models of supply and demand used by previous NRC study committees. The panel de- cided to analyze the model used by the 1989 committees and presented their results to the committee in September 1993.4 While the Panel on Estimation Procedures plans to prepare a separate report summarizing its deliberations, a brief overview of their work follows. 1989 NRC Model The 1989 NRC model includes a demographic model that projects the numbers and characteristics (primarily age) of the supply of scientists in the work force and a curve- fitting procedure that forecasts the demand for scientists by sector (NRC, 1989~. The panel concluded that all compo- nents were found wanting. The demographic projection model confuses age and cohort effects; it could only project accurately in a steady state, when no projections are needed. It has no mechanism for projecting new entrants into the labor market. The demand model restates earlier ad hoc committee models and assumes, falsely, that ratios of stu- dents to teachers and of scientists to research dollars are fixed and do not change as economic and technical condi- tions change. Finally, the treatment of equilibrium implic- itly (and incorrectly) assumes that today's excess demand or supply has no effect on tomorrow's market. The panel recommended to the committee that the 1989 NRC commit- tee model "should not be the base the committee uses in developing its recommendations".5 Instead, the panel rec- ommended the exploration of alternative approaches to esti- mating supply and forecasting demand. New Techniques for Esfimafing National Needs The panel explored the feasibility of developing projec- tions of supply through demographic techniques. (Prelimi- nary work using this technique may be found in the next two chapters of this report.) These techniques begin by listing the characteristics of a given population (e.g., age, sector of employment, and employment status) and projects 21 changes in the population based on the life history of mem- bers of the population. These multistate life tables (Keyfitz, 1985) are useful for answering the questions, What will be the characteristics of the labor force in 5 years? How long do workers remain in a particular job? The answer is generated through a series of statistical calculations by making assumptions about both the rates of transition of individuals from state to state (employed to retired, for example) and the rates of new entrants to the system. The panel expects to continue its work in this area in 1994 and will prepare a final report containing recom- mendations for the further development of forecasting ac- tivities along these lines.6 The panel is skeptical, furthermore, about the possibility of generating useful forecasts of demand and gave two rea- sons for doubting that useful long run forecasts of demand could be made. First, the conceptual basis of current fore- casting models is questionable. Demand forecasts in the tradition of previous reports assume the existence of some fixed function relating scientists needed to students and re- search dollars. They assume, in the simplest case, that stu- dent-faculty ratios are fixed and that the number of dollars required to support one scientist is fixed. This is, on its face, silly. Student-faculty ratios depend on market condi- tions and historically have varied considerably. The rela- tionship between research dollars and Ph.D. scientists em- ployed depends on the cost and capability of machinery and the relative costs of technicians, postdoctoral appointees, and Ph.D.s, all of which are absent from these models. More complex models assume complicated (and completely ad hoc) relationships between students and faculty and sci- enlists and dollars spent on research; they continue to as- sume that these relationships are fixed and independent of economic and technological conditions. The second reason the panel is skeptical about the possi- bility of generating useful long-term demand forecasts is simpler. To be of use, the forecasts must predict accurately a long period in the future. Those who enter graduate school today will be beginning Heir independent research careers 6 to 8 (or more) years from now. The predictions of fore- casting models become imprecise when projected 5 to 10 years. Models Hat previous committees developed are a case in point. The panel believes it is both feasible and useful to develop short-term indicators of demand. Monitoring these will pro- vide information about current employment conditions. This information is of limited use to policymakers because today' s decisions about fellowships and traineeships prima- rily affect He scientific labor market a decade from now. Nonetheless, it is useful for two reasons. First, it may sug- ~est ways in which current policy should be changed. For example, if a field has an extremely tight labor market, a shift from post- to pre-doctoral support may be appropriate.
MEEl~lNG THE NATION' S NEEDS FOR BIOMEDICAL AND BEHAVIORAL SCIENTISTS Second, current market conditions are an important influence on young people's decisions to become research scientists. Summary The Panel on Estimation Procedures has conducted a se- ries of preliminary assessments of available models of sup- ply and demand and has concluded that they are of no util- ity to this present effort to establish He nation's overall need for biomedical and behavioral research personnel. The committee has accepted this conclusion and suspended use of mathematical models of supply and demand with this study. The panel has concluded Mat new approaches are needed to project the supply of these researchers and to estimate demand. The panel has suggested the use of multistate life tables for assessing changes in the composition of the labor force over time. The committee has accepted this sugges- tion. The panel has also offered sample short-term indica- tors of demand, some of which are included in this report.7 The committee believes that the panel has made a sig- nif~cant contribution to the process of establishing overall need for biomedical and behavioral scientists by demon- s~ating the potential value of techniques that monitor changes in He supply of scientists and the value of short- term indicators of demand over previous mathematical ap- proaches to estimating supply and demand adopted by the NRC. Although the product of Heir work, found in the next two chapters, must be viewed as preliminary, it already shows promise as a policy tool in human resource studies. NOTES 1. See, also, Appendix A for a brief overview of the key features of previous NRC reports in this area. 2. Legislative reform in the 1970s resulted in the inclusion of nursing research personnel around 1978. Oral health research personnel (included by that name in this present study) represented an outgrowth of the clinical sciences. Research Gaining needs in that area were addressed separately as "dental research training" in the late 1970s. 3. The Panel reviewed earlier models developed by previous NRC committees but concluded that their attention to "academic employment opportunities" restricted their utility to present market concerns, and re- moved them from serious consideration for further use by the National Research Council. 4. M. Rothschild, report to the Committee on National Needs for Bio- medical and Behavioral Research Personnel, September 11, 1993. 22 5. M. Rothschild, ibid. Rothschild continues: "Tinkering with the model while retaining its basic structure will not make it a useful tool." 6. Since future Ph.D.s are produced only by those prior Ph.D.s who entered academia, projections of the supply of future Ph.D.s must separate these two components of doctoral supply. Simple projections of future supply based on the current entire doctoral population are misleading. 7. The committee restricted the application of these new techniques to the market outlook in the basic biomedical and behavioral sciences. This was due to the fact that data requirements made detailed analyses possible only in those areas. To the extent data were available for a similar set of variables, they were included in other chapters. However, the multistate life table method was restricted to work in chapters 3 and 4 owing to the lack of information of sufficient detail to permit the application of that method in the other areas. REFERENCES American Association for the Advancement of Science 1992 Congressional Action on Research and Development in the I;Y1993 Budget. Washington,D.C.: AAAS. 1993 Congressional Action on Research and Development in the FY 1994 Budget. Washington, D.C.: AAAS. Government- University-Industry Research Roundtable 1992 Fateful Choices: The Future of the U.S. Academic Research Enterprise. Washington, D.C.: GUIRR. Keyfitz, N. 1985 Applied Mathematical Demography. 2nd Ed. New York: Springer-Verlag. Institute of Medicine 1985 Personnel Needs and Training for Biomedical and Behavior- al Research. Washington, D.C.: National Academy Press. National Institutes of Health 1993 NIHDatabook 1993. Publication No. 93-1261. September. Bethesda, MD: National Institutes of Health. National Research Council 1975 Personnel Needs and Training for Biomedical and Behavior al Research. Washington, D.C.: National Academy Press. 1978 Personnel Needs and Training for Biomedical and Behavior al Research. Washington, D.C.: National Academy Press. 1981 Personnel Needs and Training for Biomedical and Behavior al Research. Washington, D.C.: National Academy Press. 1989 Biomedical and Behavioral Research Scientists: Their Train ing and Supply. 3 Volumes. Washington, D.C.: National Academy Press. National Science Board 1994 Science and Engineering Indicators - 1993. Washington, D.C.: National Science Foundation. Office of Technology Assessment 1988 New Developments in Biotechnology. Washington, D.C.: Office of Technology Assessment.