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.
APPENDIX A: Commissioner! Papers "Scientific and Engineering Personnel: Lessons and Policy Directions," Eli Ginzberg (Columbia University), 25 "Modeling the Supply of Scientists and Engineers: An Assessment of the DauffenBach- Fior~to Work," Michael S. McPherson (Williams College), 43 "What Can Demand and Manpower Requirements Models Tell Us About the Impact of Defense Spending on the Labor Market for Scientists and Engineers?" W. Lee Hansen (University of Wisconsin-Madison), 53 23
SCIENTIFIC AND ENGINEERING PERSONNEL: LESSONS AND POLICY DIRECTIONS* Eli Ginzberg Columbia University Introduction Only a small and steadily declining proportion of the U.S. population can recall f~rst-hand the extent to which our participation in World War IT altered the directions of our national life and expenences. Nowhere was the sea change greater than in the new roles of research, research-based activity, and scientific and engineering personnel in the performance of priority work in all sectors of our national life--in government, in non- profit institutions (universities), and in the pnv ate sector. To put the matter simply and sharply: in the prosperous 1920s--the New Era, when the optimists believed that the business cycle had been permanently elim~nated--there was little interest and even less concern about the supply of college-educated personnel, including the numbers of scientists and engineers who were in the labor force or being trained to enter the labor force. A small number of U.S. companies--such as duPont, General Electric, Westinghouse, RCA, Kodak and AT&T--operated industnal laboratories, but they did not experience any particular difficulties in recruiting the numbers of young graduates for whom they had openings. In fact, some of them pursued discriminatory hiring policies that made it difficult or impossible for qualified women, Jews, or Blacks to secure positions. The 10-year devastating Depression of the 1930s led to an even more constrained employment outlook for all who were seeking work, including scientists and engineers. The entrance of the United States into World War ~ in December 1941 and during the year or so preceding, when the country had begun to mobilize, altered the manpower scene radically. Recently, ~ found the following paragraph in a speech that ~ delivered to the American Society of Planning Officials in the spring of 1942: In peacetime, the best of our high school seniors fall to go on to college for economic reasons. Only one in four of our best students continue with their studies. President Conant of Harvard has been crying in the wilderness these last months for a large-scale subsidized program to enable the best of our high school students to go on to college and thereby contribute not only to the war effort but to the long-n~n unprovement of our democracy. Early in World War II, the War and Navy departments enrolled into the reserves limited numbers of draft-eligible college students, particularly pre-medical and science *My long-term associate, Anna Dutka, played a significant role in preparing the filial version of this paper, for which I am much in her debt. lEli Ginzberg, "The Coming Crisis in Manpower with Special Reference to the English Experience," in Proceedings of the National Conference on Planning, Chicago: American Society of Planning Officials, May 25-27, 1942, p. 89. 25
majors, to enable them to continue their undergraduate studies. But as the services' needs for additional uniformed personnel increased, most of these recruits were called to active duty. This was the first of what later came to be frequent instances of federal government policies intended to increase the flow of college and university-educated personnel to meet current or prospective national needs. For more than 40 years, considerations of national defense have dominated the shaping of federal manpower policies in the arena of higher education. Only two other considerations have had much weight: (~) the desirability on equity grounds of removing income and other discriminatory barriers from the paths of young people capable of pursuing higher education and (2) the national need for enlarged supplies of trained persons to assure the growth and competitiveness of U.S. industry. The purpose of this retrospective assessment is to look closely at important federal interventions in the area of scientific and engineering personnel over the last 40 years in order to ex~act the more important lessons that have relevance to the formulation of current and prospective policies. As one who has had a continuing, if not always intimate, relationship to these earlier interventions, it appears that many of the issues that keep resurfacing and that engage the attention of policymakers are new and challenging only to the beholders. Looked at with the benefit of hindsight, they often turn out to be old issues in new dress. A Retrospective of Critical Incidents This venture at retrospective assessment will focus on a limited number of "cntical incidents" affecting the formulation of science and engineering personnel policy from the waning days of World War IT to the most recent concerns (1985-86), as reflected in the agendas of the Congress, the professional societies, and the National Academy of Sciences. The G! Bid of Rights Over 15 niillion young Americans, most of them between the ages of IS and 26, were called to active duty during the course of World War IT. Many of them had" to interrupt their education to don a uniform; arid in many instances, three, four or even five years passed before they were demobilized. In recognition of their senice to their country, Congress passed the Servicemen's Readjustment Act of 1944 (GI Bill of Rights), which provided wide-ranging benefits to 7.8 million World War II veterans and enabled more than 2.2 million of them to attend colleges and universities. The federal government provided tuition and maintenance support from a minimum of one to a maximum of four years, depending on the veteran's length of service in the Anned Forces.2 While the primary interest of Congress in passing the bill was to express the nation's appreciation to those who had borne the brunt of the battle, its supporters, particularly those in the sciences, recognized that enabling large numbers of veterans to return to school to complete their educational preparation for a career would be beneficial to the nation, since it would enlarge its pool of pained persons. Scientific Talent aruiNew Horizons for Research The war permanently altered the place of research on the nation's agenda, for it demonstrated that our national security would henceforth depend on our technological 2Eli Ginzberg and A~sociaes, Patterns of Performance, New York: Columbia University Press, 1959, pp. 167 ff. 26
supenorit,. President Roosevelt had taken the big gamble to support the production of the atomic bomb, and the gamble succeeded with He speedy end of the war with Japan in early August 1945. The principal scientific advisers of the President, with Vannevar Bush in the lead, stressed the much enlarged role that the federal government must play in the future financing of research arid higher education.3 Without the participation of many European scientists who had been forced by Hitler to emigrate in the 1930s _ ~ , _ ~ ~1 _ 1 ~ ~ ~ . ~ ~ a ~ ~ ~ ~ ;, the successful anu~;~ure o~ u~e oorno would not nave ocen posslole. in tne future, we would have to look more to our own institutions and our own population to provide the intellectual leadership required for the discovery of new scientific ideas and their successful application. Appendix 4 to Bush's report to the President consisted of a "Report of the Committee on Discovery and Development of Scientific Talent," chaired by Henry Allen Moe, which made the following critical continents and recommendations: In answer to President Roosevelt's question to Dr. Bush, "Can an effective program be prepared for discovering and developing scientific talent in American youth so that the continuing future of scientific research in this country can be assured on a level comparable to what has been done during the war?" Moe's reply was: "In our judgment, the answer to the question is in all respects in the affirmative.... The intelligence of the citizenry is a national resource which transcends in importance all ether n~tir~nn1 . _ - . ~ ~ . . . resources. l o be effective, that intelligence must be trained.... Our plans, simply, are plans--as regards science and engineering--to train for the national welfare the highest ability of the youth of the nation without regard to where it was born and raised, and without regard to the size of the family income." The Bush Report asked for a level of $122.5 million of federal funding five years into the future, with $29 million earmarked for the Division of Scientific Personnel and Education. Bush noted in passing that "the most important single factor in scientific and technical work is the quality of the personnel employed." It would be an exaggeration, however, to contend that the scientific establishment's can for federal action was responded to by all other leadership groups. In 1947, the Twentieth Century Fund concluded that "we have more than enough manpower ... to fulfill our requirements under every conceivable circumstance."4 But that view did not survive for long. It was undermined by the events of 1950, which found the United States once again engaged on He battlefield, this time in Korea. The Korean War Although we entered hostilities through a decision taken by President Truman without debate and approval of the Congress, the country moved slowly to gear up for combat. General Eisenhower, among others, was greatly disheartened by what he considered the inadequate response of the Washington leadership to the new national emergency. It soon tunnel out that we could not pursue active hostilities in Korea successfully, even with constrained goals, unless we were wining to redirect resources, physical and human, from 3Vannevar Bush, Science--The Endless Frontier: A Report to the President, Washington, D.C.: U.S. Government Printing Office, July 1945. 4Twendeth Century Fund, Amenca's Needs and Resources. New York, 1947. 27
civilian to military projects. We simply could not keep the civilian economy at fun throttle and win on the battlefield Not all of the demands by the military were weD~onceived or justif~eci. ~ recall that in my capacity as consultant to the Assistant Secretary of the Army for Manpower, ~ reviewed with the Chief of Engineers an urgent request for 400 additional engineers. It turned out that what he really needed were recruits with a high school diploma who, with 90 days' specialized Gaining, would be able to repair certain priority equipment. The war also illuminated another facet of the engineering problem--the relative inflexibility of segments of the supply of engineers. At one point, after General Motors had finally been persuaded to cut back on the production of civilian automobiles in favor of expanding its military output, the corporation offered the Army 50{) "surplus" engineers whom it could spare for the duration. After examining the background and competencies of We members of this group, the Army decided that it would reject the offer because most of the engineers had been so highly specialized for such a long time on a narrow sector of automotive manufacturing that most of them could not be placed on priority military assignments. The Small College-Age Cohort of the 1950s The stalemate on the battlefield in Korea came to be accepted by all the parties and an armistice was signed in March 1953. With Eisenhower in the White House, the stage was set for a four-year period of renewed economic advance. But Korea left its mark. It was clear that the United States could no longer assume that the "Cold War" would remain cold and that it was, therefore, compeDed to invest more in basic research and in maintaining enlarged defense forces at the same time that it sought to establish and strengthen its alliances with fluency nations. _. . . ~ . . - -i ne sc~ent~c ano eng~neenng personnel outlook was complicated by a number of contradictory factors that influenced the flows into and out of the college and university pool, including the graduation of the last remaining veterans; the decI~n~ng numbers in the college-age group (reflecting Me lowered birth rate during the Depression); and the larger proportion of the age group entering college. The situation was further complicated on the demand side by the fact that many U.S. colorations had decided to establish and/or expand their research and development operations, thereby adding a signiD~c ant increment to the total requirements for scientists and engineers. Added to this expanded civilian demand were direct and indirect demands stimulated by much enlarged federal funding for basic research and for a wide variety of defense-related projects. Finally, liberal funding from the federal government led to Me expansion of major research universities. The personnel figures were volatile. The number of engineering graduates increased from 31,000 In 1948 to a peak of 52,000 In 1950, only to decline to 40,000 in 1951 and 30,000 in 1952. In 1951, there was an estimated demand for 80,000 new engineers, but by We following year the demand had declined to 40,000.5 The demand data were no more than employers' reported intentions to hire--if We numbers were available--at a salary that the corporations were willing to pay. Apparently, when the numbers fell far short--as In 19SI, when the gap amounted to 40,000--employers got on with much reduced numbers. It is worth noting that the so-called "shortage" in 1951 amounted to about 10 percent of Me total number of engineers. With the end of the fighting in Korea, the lightness ~ We engineering market moderated. 5Nadonal Manpower CounciL A Policy for Sczentif c and Professional Manpower, New York: Columbia University Press, 1953, pp. 162 if. 28
The Response to Sputnik The launching of the first satellite into space by the USSR in 1957 had a major unsettling effect on the self-confidence of the American people. Although the U.S. defense position had been weakened in the early 1950s, when the Russians developed the capability of manufacturing nuclear bombs, we continued to fee! reasonably secure in our technological leadership. After all, the Russians had foreshortened considerably their period of developing a nuclear technology by exploiting the secrets they had stolen from us and our allies. Their success with Sputnik, however, represented a challenge to our scientific leadership that led to much introspection and criticism among our leaders--scientific, political, and educational. In 1958, Congress provided a partial answer in the fob of the National Defense Education Act IDEAS. For the first time, the federal government assumed a major across-the-board obligation to strengthen the nation's manpower pool, particularly scientists, engineers and other specialists (such as area and language experts) whose work was deemed essential to the national defense. While President Eisenhower was reluctant to expand the scope of the federal government's activities into the arena of higher education, he signed the bin because of its potential to strengthen the nation's defenses. In addition to liberal funding for college and graduate students in designated scientific fields, the bill also aimed at enlarging Me pool of college-eligibles by making funds available for strengthening high school instruction in the sciences and mathematics. It should be noted parenthetically that while the federal government had made funding available to selected groups of students through selected federal agencies, most of the earlier support via the Armed Services, the National Science Foundation and the National Institutes of Health had been directed at graduate and postgraduate students. NDEA was a much larger and broader type of federal intervention, covering many more students, mostly at the undergraduate level. In its 1953 report, the National Manpower Council (a non-governmental body established at Columbia University under funding from the Ford Foundation, of which ~ was the Director of Research) estimated that only about half of the coHege-age population with an AGCT score of 120 or above entered college and that only about one-third went on to graduate.6 Continuing federal support in the decade after the passage of NDEA, as well as the substantial efforts of many state governments to expand their systems of higher education, led to the substantial elimination of the financial barriers that had earlier blocked high school graduates with good aptitudes from entering and graduating from college. The Revision of the Immigration Act in 1965 Twenty years after the conclusion of World War IT, Congress finally acted to revise our immigration laws to bring them more into consonance with He changing role of the United States in world affairs. We moved away from preferential quotas for the countries of Western and Northern Europe and enabled immigrants from Asia to gain admission on a large scale for the first time. The new law--by permitting up to 10 percent of total immigration to consist of professionals, scientists, and artists of exceptional ability--made it easier for professionally trained persons to enter the United States and subsequently to regularize their status as permanent residents and to acquire citizenship. In the long debate leading up to the revision of the act, it was pointed out repeatedly that the United States had been the beneficiary of the forced eITugration to this country of many professionals whom Hitler and Mussolini had persecuted; the contribution of many displaced European scientists to the success of the atom bomb was widely recognized. 6lbid., p`. 82. 29
The long Genoa of largely uninterrupted economic expansion--a quarter of a century since 940--unquestionably contributed to the willingness of the Congress to revise the immigration statute, in particular for individuals whose education and experience made it likely that they would add to the nation's pool of competence and stimulate rather than detract from its future economic development. The revised rules and regulations also set the stage for a much expanded inflow of students from abroad, particularly graduate students, to undertake advanced study. While many would eventually return to their native country to help speed its development, new opportunities were created whereby those who preferred to remain In the United States would be able to do so, even if they might have to leave arid re-enter to comply with the law. Cutbacks in Space and D~ense The period 1968-1972 represented the first substantial shock to the expansionary environment that had dominated the employment prospects of scientists and engineers since the onset of mobilization in 1940, three decades earlier. The completion of the successful "moon shot" in 1969 and Me decrease c - the defense budget led to large-scale reductions in the employment of many engineers and scientists, especially among aerospace contractors on the West Coast but also in plants located elsewhere, including those on Route 128 in the Boston area. Many were caught by surprise, including employers, universities, Congress, the professional societies, and still others. Although the federal government, with the passage of the Manpower Development and Training Act in 1962, had begun to fund training and retraining programs for unemployed persons, these programs were aimed at assisting individuals with lifted education and skills. As chainnan of the National Manpower Advisory Committee (NMAC) from the inauguration of the effort, ~ can state unequivocally that we never addressed the unemployment of engineers and scientists until the severe cutbacks at the end of the decade.7 That was the first time that the problem had surfaced; and as it grew, we found ourselves poorly positioned to respond. The training programs that were in place did not fit the needs of unemployed professionals. The National Academy of Engineering established a special committee to explore the issues and to make recommendations and ~ was asked, in my capacity as chairman of the NMAC, to accept membership. We recruited a senior manpower analyst, Seymour Wolfbein, to develop the relevant statistics and to help us formulate possible lines of remedial action. By the time our report was ready, the worst of the recession and its accompanying unemployment had passed.8 The late 1960s-early 1970s represented a confluence of three negative forces: the peaking of the expansion in the student body; Me significant reductions In defense and space research and development; and an economic recession. Any one or two of the above would have put some part of the scientific and eng~neenng personnel under stress, but the s~mult~ei~cy of ah three occurrences led to a serious reversal In the hitherto expansionary trends. While earlier shifts in the financing of space and defense projects and the weakening in We civilian economy had resulted In selected cutbacks, this was the first time so large a segment of the market for trained personnel--literally tens of thousands of scientists and engineers--was affected. Further, the concentration of so many aerospace activities on Me West Coast--and the unwillingness of most of the displaced professionals to relocate--intensif1ed He emblems of designing and implementing remedial programs. 7National Manpower Advisory Committee, Manpower Advicefor Government: Letters to the Secretaries of Labor and of Health, Education and Welfare, Washington, D.C.: U. S. Department of Labor, 1972. National Academy of Engineering, Committee on Engineenng Manpower Policy, Engineering and Scientific Manpower: Recommendations for the Seventies, Washington, D.C.: National Academy of Science's Printing and Publishing Office, 1973. 30
One important by-product of the NAE explorations was a deepened perspective on the limitation of professional retraining. We were informed repeatedly, by those in a position to know and act, that it made little sense to retrain a mechanical engineer who had spent eight or ten years on a narrow job assignment. He no longer had the knowledge base and the intellectual flexibility to justify the investment of sizable funds that would be required to retrain him, for instance, for electncal engineering. The Committee accepted this explanation. The industry's lack of broad investment in keeping its engineering work force up to date squared with the evidence of many engineers' having become superannuated after 10 years of aerospace employment. Fortunately, the economy expanded in the early 1970s--in fact, our highest rates of real growth in the entire post- WorId War rid era occurred in 1971-72--and what had earlier appeared to be a difficult, almost an insoluble problem disappeared while remedial actions were still being explored. Weakening of the Academic Labor Market in the 1970s The vast increase in the number of college and graduate students in the 1960s, combined with greatly enlarged public and private expenditures for research and development and further stimulated by worId-wide economic growth in which research-based industries were in the lead, set the stage for the much enlarged and sustained demand for scientists and engineers. This occurred in all sectors--especialTy in the academic world, where large numbers of scientists were needed for both the classrooms and the laboratories. Allan Cartter had raised the issue about faulty projections for ever-increasing numbers of faculty during the explosive 1960s, warning that the deeply entrenched view of continuing shortages of professional personnel failed to take into account the prospective leveling off of college and graduate student enrollments, as well as more constrained governmental funding for research and development. However, Cartter's was a minority view. In late 1963, the Subcommittee on Science, Research and Development of the House of Representatives held hearings on "Government and Science," which elicited agreement from most of the nai~on's scientific leadership that we needed to enlarge our pool of qualified researchers. This was the view of Jerome Wiesner, the President's science advisor, as well as of the Vice President of Research for duPont, the advocates of a strong space program including Dr. Werner van Braun, and many others in academic and public life. Only Nobel Laureate Harold Urey took a different view: in a written communication he noted, "I have a belief that we are training as many scientists as can reasonably be expected from the crop of students coming from our high schools and colleges. What we need is better scientists than we have--not more of them."9 Since 1967, evidence was accumulating that jobs in science were more difficult to obtain. By 1970, the American hnstitute of Physics recognized that physics was confronted with a changed labor market. But even in 1970 (a catastrophic year in the employment history of physicists), the market was not weak across the board, but rather had turned down for particle and nuclear physics, while the demand for specialists in acoustics and optics remained strong. At the end of the 1970s, the results of an American Physical Societr study of the present and projected employment of physicists pointed out that a high proportion of the best-trained young physicists who had failed to obtain a tenured academic position were unsettled about the ways in which their careers had developed.~° 9U.S. Congress, Committee on Science and Astronautics, Subcommittee on Science, Research and Development, The National Science Foundation: A General Review of its First 15 Years: Hearings, 88th Congress, 1st Session, October 15-16, 18, 22, 24, 29; November 5, 19-20, 1963, no. 8, Appendix A, Washington, D.C.: U.S. Government Printing Office, 1964, p. 427. 1°American Physical Society, Physics Manpower Panel, The Transition in Physics Doctoral Employment, 1960-90, New York, August 1979. 31
There was considerable turmoil as well among young mathematicians, many of whom were also forced by a much tightened academic environment to forego a professorial career and seek out alternative employment opportunities. The Ad hoc Committee on Resources for the Mathematical Sciences found four general reasons why "mathematics" seems to have been the field hardest hit by the general post-1968 trends: . Research in the mathematical sciences is concentrated almost entirely in universities and colleges; hence, it is very strongly affected by any general weakening of the support of academic research. Much (but not ally mathematical research has long-tenn payoffs; thus, the field will be strongly affected by federal policy shifts which emphasize mission relevance or immediate applicability to technologies. The long periods of time involved in developing many important mathematical tools make it unlikely that the commercial sector will support large fractions of the research; therefore, relatively little help will be found from industry when there is a weakening of federal support for fundamental research in the field. Mathematical scientists require relatively little in the way of facilities, equipment or technical staff to conduct their research; hence, their needs are less visible and often seem postponable. The Energy Crisis and the Computer Revolution But all was not gloomy in the 1970s and early 1980s. In 1973, OPEC took control of the international of] market as a consequence of which the search for new oil accelerated, a search further intensified by the second oil price increase that OPEC instituted in the late 1970s. As a consequence of this roiling of the oil market, a vastly increased demand for petroleum engineers, geologists, and related specialists was met not by only attracting scientists and engineers from related fields, but also by stimulating the expansion of key college and university departments that were in a position to respond More important in terms of the numbers involved was the growth in "computer sciences," including the aDied fields of electrical engineering and applied mathematics. The steadily growing demand on the part of industry for larger numbers of computer specialists, as well as the efforts of many institutions of higher learning to establish and expand departments In this new area. created a host of new opportunities for talented Young scientists and engineers. -or- ---lo The demands created by the oil crisis were over by the early 1980s with surpluses of specialists replacing the earlier shortages as drilling for new oil was severely reduced and as the seven large U.S. oil companies sought to shrink their swollen work forces in line win the new, unfavorable market realities. On the engineering front, aside from recurrent modest surpluses or shortages as the numbers in the educational pipeline led or lagged fluctuations in new hiring, a long-tenn difficulty arose--namely, the recn~itment/retendon of qualified faculty members. The nub of these facula problems was centered in electrical engineering and computer sciences. Most universities found themselves increasingly outbid by industry for recent doctorates whom they would have wanted to add to their faculW. While the most acute imbalances were centered In recruiting assistant professors, many engineering schools found them- seives increasingly exposed also in the higher ranks. The issue of facula shortages is not ~ National Research Council, Commission on Physical Sciences, Mathematics and Resources, Renewing U.S. Mathematics: Report of the Ad hoc Commit tee on Resources for the Mathematical Sciences. Washington, D.C.: National Academy Press, 1984, pp. 34 ff. 32 ,
amenable either to simple definition or solution. The Engineenng Faculty Shortage Project, a 1984 joint study by the American Society for Engineenng Education and the Amer~car~ ~ ~ ~ ~ ~ ~ . ~ . . ~ ~ ~ . ~ ~ . Association of engineering Societies, concluded that there was a faculty shortage or about 20-25 percent. This conclusion was predicated on the evaluations by the deans of engineering schools and further supported by the deteriorating ratio over the decade between facula and students. While facula salanes have increased and outside consulting provides some additional source of income to faculty members, the ratio of academic to industrial salaries, particularly in electrical engineering and computer sciences, is still inadequate to provide an adequate supply of faculty personnel. Furthermore, many of the topmost talented undergraduates, confronting multiple job opportunities, are not pursuing the Ph.D; and among those who do, only a minority are interested In an academic career. The total output of Ph.D.s in engineering is well below the 1972 level and there has been a precipitous decline in the ratio of native Ph.D.s to the total. The pool of young faculty members is increasingly composed of foreign nationals who, whatever their competence, often have limited communication skills essential for effective teaching. Finally, a decreasing proportion of engineering schools, according to the Accreditation Board for Engineering and Technology, are now receiving a full six-year accreditation because of curriculum deficiencies, faculty shortages, and inadequate facilities and equipment.~2~3 A number of leading corporations, individually and collectively, decided to make special gifts of money and equipment to leading universities to enable them to offer higher salves and other benefits to their staff. In the absence of such help, the corporations recognized that their future recruitment of competent employees would be at risk. Even this special response, however, failed to encourage large-scale increases in the number of native-born Americans willing to pursue their studies to the completion of a doctorate and to acceptance of an academic appointment. Foreign-born students account for a steadily increasing proportion of all doctoral students in engineering and computer science. In fact, in 1980, only one in 12 bachelor's degrees in engineering, but more Man one in four of all master's degrees and more than one in three of all doctorates were earned by foreign nationals.~4 By 1983, more than one-half of all postdoctorals in both mathematics (excluding computer sciences) and physical sciences were granted to foreign students and about one in three in the biological sciences.~5 These ratios make clear why foreign national students dominate the entry ranks of faculty in science and engineering departments. A Decade of D~ense Buildup (1976-1986J With the winding down of the Vietnam War, Congress reduced the defense budget by about one-third in real terms between 1970 and 1976. President Carter succeeded in reversing this trend, but it was during President Reagan's adm~nis~ations that the defense budget rose not only to its 1970 level but increased to a level about 20 percent higher. National defense purchases grew from a low of $157.5 billion in 1976 to $171.2 in 1980 National Research Council, Office of Scieniih~c and Engineenng Personnel, Labor-Market Cond~i`onsfor Engineers: Is There a Shortage? Proceedings of a Symposium, Washington, D.C.: National Academy Press, 1984. 130ffice of Technology Assessment, Demographic Trends and the Scientific and Engineering Work Force--A Technical Memorandum (OTA-TM-SET-3Sy, Washington, D.C.: U.S. Government Printing Office, December 1985. 14Edith Fran Cooper, United States' Supply and Demand of Scientists for Defense Research and Technology (part D, Washington, D.C.: Congressional Research Service, Library of Congress, 1981, p. 9. 1sBetty M. Vetter, The Technological Marketplace--Supply and Demand for Scientists and Engineers (3rd ed.), Washington, D.C.: Scientific Manpower Commission, May 1985, p. 23. 33
and then moved upward at a more rapid rate, reaching $236 in 198Se In 1973 and 1974, the percentage declines in national defense expenditures were respectively 6.8 and 4.5; in the first four budget years under the condor of President Reagan, the year to year percentage increases amounted to 7.5, 7.0, 6.3, and 7.~.~6 The second half of the 1970s has seen the economy in a strong recovery, followed by a deep recession in the early 198Os, strong recovery beginning in 1983, and a reduced rate of grown in the last two years (1985-861. The fact that except for Me recession years 1979-80 and 1981-82 the defense budget was expanding rapidly at a tone when the civilian economy was also expanding would lead one to expect that the demand-supply relations affecting scientists and engineers might have become quite stained. The second of! crisis of 1979, which was followed by a feverish expansion in the demand for petroleum engineers, geologists, and other professional personnel required for expanded of! . . .. ... ~ . . . . . exploration and alternative energy sources, could have been expected to place additions strain on the scientific-eng~neenng pool. There is little question that the scientific-engineering market tightened at least selectively in response to the upward tilt in both the civilian and the defense sectors. The Deutsch, Shea & Evans High Technology Recruitment Index (1974 base of 100) stool! at ~ 14 in 1977 and moved to a high of 144 in 1979, remained high until the depression year of 1982, and moved back up to 133 in 1984, only to decline to ~ 12 in 1985 and strengthen at the beginning of 1986. The Job Offer Index for bachelor degree candidates by curriculum, with 1980 as 100, records that 1981 saw a strong demand for most engineering and science graduates, but a distinct weakening in the following years right up to 1986, with aeronautical engineers and computer scientists the only two exceptions. If the focus is shifted to the changes in the employment status of doctoral S/Es involved in DoD work between the mid 1970s and the mid 1980s, this is what one finds: total employment in the four fields--mathematics, computer science, physics and eng~neenng--~ncreased from 18,963 to 22,533, a gain of 3570 or slightly under 20 percent In each of these two time periods (1973-75 and 1983-85), there was considerable movement between the defense and nondefense sectors: in the 1983-85 period there was a net inflow of 740 doctoral personnel into DoD-type work but an outflow of 537 engineers into non-DoD work. The explanation for the relative ease with which the nation was able to accommodate the steep buildup of defense in the 1980s without experiencing any serious S/E shortages must be sought In the following: (~) the dominance of such personnel in the nondefense sector, which provided a pool to draw on; (2) the softness of the nondefense sector of the economy in 1981-82 and the aftermath of caution in investment and hiring decisions; and (3) We large output of new engineers from bachelor degree programs. In 1981-82, with a total pool of approximately I.2 million engineers, defense requirements came to about 140,000; in the case of scientists, defense work accounted for only 3 percent of the total supply. Although the proportion of scientists and engineers in defense work is relatively much greater Man in nondefense work--on the order of 5:~--the nondefense sector continues to dominate the U.S. economy. ~ 1976 the number of engineers graduating with a bachelor's degree was slightly below 3S,000; in 1985 it was just under 7S,000, an increase of over 100 percent In the decade. A sizable increase, from 16,500 to 22,500, also occurred in master's degree recipients, a gain of just under two-~fths. The output of eng~neenng doctorates declined slightly from the 3,000 level In 1976, but by 1985 had increaser} to 3383. 16Electronics Industries Association, The Military Electronics Market: Perspectives on Future Opportunitzes--The 21st Annual EIA Ten-Year Forecast, Washington, D.C.: The Requirements Committee, 1985. 34
For the reasons identified above, the U.S. was able to cope with the additional S/E demands engendered by the sustained defense boom without major difficulty. With the defense budget as a percentage of GNP likely to level off in the remainder of the 198Os, if it does not decline, there is little risk of any serious imbalances in the engineering and scientific manpower market In the near and middle term. Continuing Concerns: Supply and Utilization Women and Minorines In the decade 1963-73, major governmental and non-governmental initiatives were undertaken to remove barriers from the paths of women and minorities seeking to pursue educational and career objectives in science and engineering. There were strong responses to diverse federal initiatives by business, the academic community, foundations, and the professions, as well as by women and members of m~nori~ groups. More recently, legal and institutional pressures aimed at lowering discriminatory barriers have weakened, but the earlier momentum has enabled many more women and some m~nonties to make sizable progress in pursuing professional careers. There have been striking Increases in the number and proportion of women entering and completing college, as wed as increases in the numbers earning masters' or doctorates, including those entering a course of study in engineering or one of the natural sciences. True, men continue to outnumber women in the hard sciences and in engineering by a factor of three, four or live, but these differences have narrowed considerably over the last 20 years. For instance, in 1965 women accounted for around one percent of engineering school students, but in 1985, they represented over 15 percent of the enhance class. With the exception of Asian-Americans, the increase of Snooty group members, particularly Blacks and Hispanics, into engineering and the natural sciences shows only modest gains. The explanations for the slow progress seem to be embedded in a galaxy of reinforcing negative factors, including weak family supports; weak basic schooling; weak science and mathematics departments in segregated colleges; lack of role models; and attractive alternative career opportunities for talented minority students. While the foregoing ad hoc explanations may go far in accounting for the continuing large-scare underrepresentation of m~nonty students in science and engineering, a parapet challenge is to expia~n the marked overrepresentabon of Asian Americans in these f~elds.~7 The following factors underpin the reason that the public and private sectors should continue to be concerned with structural transformations in the future supply of women and m~nonties into science and engineenng. Women now represent the majority of all college students; but despite the strikingly higher proportion enrolled for an engineering degree, they still account for only about 15 percent of the class, and they continue to be substantially underrepresented in the physical sciences at all degree 1evelse The long-term persistence of a relatively stable ratio among college men as between science and non- science majors, roughly 30 percent vs. 70 percent, underscores the desirability of encouraging more women to pursue a career in engineering or the sciences. Several reinforcing factors support this view. The demographic trends point to a substantial declirle in the college-age group, particularly the white, college-age group that 17G ail E. Thomas, The Access and Success of Blacks and Hispanics in U.S. Graduate and Professional Education, a working paper for Me Office of Scientific and Engineenng Personnel, National Research Council, Washington, D.C.: National Academy Press, 1986. Also, Office of Technology Assessment, Demographic Trends, op. cit., pp. 114-127 and Appendix B. 3~;
represents the major source of future candidates for baccalaureate and higher degrees. Further, enrollments in engineering schools peaked in 198 1-82 at slightly above I 15,000, decided since then to ~ 12,000 in 1985, and in light of the above demographic ~ends, are likely to decline further. Other than a radical change in our immigration policies aimed at admitting larger numbers of students and graduates in science and engineering, women represent the only large potential pool that could yield over the next years significant additions to the nations supply. Scientific and Engineering Personnel and the Competitiveness of the U.S. Economy When the U.S. was eclipsed by the USSR in the space race in 1957, analysts looking for an explanation made much of the fact that the Russians had both a much larger absolute number of engineers and a higher ratio of engineers to their total work force than did the United States. Some concluded, therefore, that the U.S. might lose out in the struggle with the USSR unless it moved quickly and strongly to increase its output of scientists and engineers. Much the same argument has resurfaced in recent years as Americans have explored the reasons for the superior performance of the Japanese economy. Many analysts have pointed to the much larger relative number and proportion of engineers in the Japanese, as compared with the American, economy. Since the Russians were unable to repeat their triumph in space in other technological spheres, the supposed vuinerabili~ of the U.S. stemming from an insufficient number of engineering personnel lost credence. Tune also revealed that many Russian engineers were so narrowly trained that many were more technicians than professionals. Simple numerical comparisons of the proportions of engineers in their respective work forces can also be misleading in assessing the relative economic strength of the United States and Japan. A more sophisticated analysis suggests that numbers aside. Japan has important advantages In the continuing attention that its top management has pain to process engineering, in the heavy investments that it continues to make in the education of its engineering work forces, and in the excellent relations that it has encouraged between engineering supervisors and the technician work force. Additional factors need to be considered, including the relatively much larger defense and defense-related activities in which We United States engages. Marty observers have noted that even after allowing for positive spiD-over effects on the civilian economy from our large defense sector, we have a smaller proportion of our total specialized personnel resources focused on output for the civilian market The heavy concentration of SIE personnel in the 700 federal research laboratories and We sizable amount of all federal dollars that these laboratories spend on basic research-- approximately I/3--have rmsed questions from time to time concerning the impact of this sizable federal effort on Me competitiveness of the U.S. economy. The questions have been two: Is too high a proportion of our nation's total investment in basic research being devoted to military goals? Secondly, do the laboratories make effective use of their S/E personnel? The answer to the first must be sought in the budget proposals of successive a~minis~ations and the subsequent actions by the Congress. Over the years, the question as to whether the federal government makes effective use of its S/E personnel has been repeatedly raised, most recently (1982) by the White House Science Council's Review Pane} of federal laboratories, chaired by David Packard. The Pane} found that the laboratories had difficulty recruiting and retaining mathematicians, electrical engineers, and computer specialists and recommended more flexibility in salaries for "superstars," in the belief that such action might offer some relief. Broad-banding of civil service grades to give supervisors greater flexibility in setting initial salary offers and subsequent in-grade Labor-Market Conditionsfor Engineers, op. cit., pp. 123 ff. 36
increases was also recommended as a way to permit supervisors to link salary to performance, rather than to seniority.~9 A recent newspaper report indicates that the President favors the above action, and the administration has begun to implement it. 20 The Government-University-Industry Research Roundtable also suggested (1986) that since federal laboratories are generally perceived to be of vanable quality, reassessment of the scope and mission of some of these may be warranted, particularly where the programs could be carried out equally well in universities. In early 1986, the National Academy of Science set about to explore the issue whether, and to what extent, much enlarged appropriations for defense, including the Strategic Defense Initiative program, may have a dislocating, retarding effect on the capacity of our civilian research and development fins to strengthen their competitive position. This is not the first time that the subject of diversion of our research and development resources from the civilian sector due to sudden and large increases in defense activities has been raised. in the early 1960s, the Vice President of Research for duPont (quoted earlier) had called attention to this intersectoral competition and pointed out that sudden increases in defense spending had the inevitable result of drawing resources from civilian firms and concurrently raising the costs of civilian research and development. But the fact that informed persons have repeatedly expressed concern about the consequences of sectoral distortion, following upon large increases in defense spending, cannot be accepted at face value. The opposing argument emphasizes that a high level of, and ever more steeply rising, defense expenditures send a signal to the colleges and universities that there will be more good jobs for scientists and engineers, which in turn will lead to an expansion in the supply. But an interval of four to five years occurs between the sending of the signal and the outflows of new streams of engineers with a bachelor's or master's degree. Our feast- famine approach to defense appropriations introduces substantial instability into the flows of engineering and scientific personnel to both the civilian and defense sectors. The aerospace companies with new large contracts often raise salaries to attract additional personnel, most often creating problems for nondefense firms and the universities. But if and when defense expenditures are reduced, there are likely to be untoward reductions in the new supply some years down the road. While the numbers and quality of engineering and scientific personnel available to private sector industries on occasion may place the U.S. economy at a comparative disadvantage, there is little evidence to conclude that this handicap is the core of our difficulties in maintaining our ~ntemai~onal competitive position. There is, instead, mounting evidence that many large U.S. companies have been poorly organized and poorly managed to respond quickly and effectively to what is rapidly becoming a world economy. These systemic weaknesses have also reduced the effectiveness with which many U.S. companies utilize their scientific and engineering personnel. If this explanation has meet, then the challenge that we face may be less connected with the size of the pool than with the more effective utilization of the available supply.22 Considerable efforts have been expended over the years to develop and improve estimates of the future supply and demand for engineers. A recent critical review of these . . . 19D avid Packard, "The Loss of Government Scientific and Engineering Talent," Issues in Science and Technology, Spring 1986, pp. 126-131. 'Reagan Asks Civil Service Pay Changes," The Washington Post, April 30, 1986, pages Al, A7. 21Government-Universi~-IndustIy Research Roundrable (sponsored by the NAS, the NAE and IOM), What Research Strategies Best Serve the National Interest in a Period of Budgetary Stress? Report of a Conference. Washington, D.C.: February 2~27, 1986. 22Eli Ginzberg and George Vojta, Beyond Human Scale: The Large Corporation at Risk, New York: Basic Books, 1985. 37
efforts by Professor W. Lee Hansen concluded that the extant models simply do not possess the requisite power, given the many weaknesses in the data bases, to yield reliable forecasts. Lessons On the basis of the foregoing brief consideration of 10 cntical incidents on the science-engineer~ng horizon that surfaced over the last four decades, we are now better positioned to extract the "lessons" that can usefully inform present and future policies relating to the supply and utilization of scientific and engineering personnel. Since such personnel issues are directly and closely related to the effectiveness of our defense and to the competitiveness of our economy, they clearly warrant continuing consideration. . The U.S. has been strikingly successful in the post-WorId War IT period in removing financial barriers in the path of qualified young people to pursue a college or higher degree. Moreover, all three sectors--the federal government, state governments, and philanthropy--have cooperated to expand and improve the infrastructure of higher education, in order to accommodate the much enlarged inflows of students. The federal government and selected sectors of the private economy recognized the importance of vastly expanding their investments in research and development as a result of which this country became a leader in a great many industries including aircraft, nuclear power, computers, electronics, space, telecommlm~cadons, biotechnology, and many others. Once it became clear to the American~eople in the early 1950s that the international situation pointed to a long "cold war," the nation decided that it had to make a large and continuing commitment to defense and defense-related activities, no matter what the cost in diverted resources. In retrospect, one can argue that except for the perturbations resulting from years of active hostilities in Korea and Vietnam, which strained the civilian economy, the resources required by defense have been secured more by an expansion in the supply and operating the economy closer to full employment than in deflections from the nondefense sectors. The substantial appropriations of the Congress for defense have played a major role in expanding and maintaining a much enlarged infrastructure of scientific research and higher education, far larger than would have been likely in the absence of the "cold war." · The fluctuations in defense expenditures had greater or lesser consequences on particular organizations, groups of specialists, and locations that were directly affected by the start-up or completion/cancellation of a major defense pro- gram. But for the most part, the two major nondefense sectors--the higher educational establishment and the civilian economy--were able to adjust reasonably wed to these fluctuations, We early 1970s alone excepted. In that period, He defense cutbacks coincided with an economic recession and the end of a boom in college entrants, which created a softness in the academic market with long-tenn dislocations among doctoral candidates and graduates who had looked forward to an academic career but faced instead a broadscale hiring freeze. The explanations for the remarkable adjustment potential of the U.S. economy over the last four decades must be sought and found in the following: (~) the substantial and sustained period of economic growth, which provided the 38
. necessary resources to expand the educational and Raining infrastructure and which created a great number of new alternative jobs and careers for scientists and engineers; (2) the availability of an untapped talent pool, which could be attracted to the rapidly expanding professions once the financial barriers to the pursuit of a higher education were lowered or removed; (3) the willingness of trained personnel, as well as those in training, to follow the signals of the market; and (4) the willingness of employers to modify their hiring criteria when they encountered tightness in their conventional sources of supply, as regards both the gender and race of prospective employees and their prior experience. With He advantage of hindsight, it becomes clear that the elasticity in the supply of students who are able, willing, and eager to pursue higher education-- together with the long-term flexibility in the U.S. labor market, which reflected the behavior of both employers and employees--made it possible for the U.S. to expand by orders of magnitude its involvement in R&D activities in both the defense and nondefense sectors. It was clearly not a process without some "transaction costs," witness the unsettlements of the SIE labor market in the early 1970s, but for the most part these costs were not excessive. The question that must still be explored is how these lessons might inform policy in the years ahead. Policy Directions On the basis of our analysis of the critical incidents and the lessons extracted therefrom, we are now better positioned to call attention to directions that should inform our future policy affecting the supply and utilization of scientific and engineering personnel. Two points are worth exploring at the outset. First, we cannot follow the same policies and practices of the earlier decades and anticipate on balance the same favorable results. The reason is simple: many prior existing conditions have been permanently altered, as for instance, the proportion of the college age-group that enters higher . .. ~ . .. .. .. . . . , , . , , . . ~ Hi. ~ education. At the same time, it would be an error to assume that tile lessons or the past cannot be used to provide constructive guidance for the future. The real challenge to the polices analyst is to select the "right" lessons and to apply them with due consideration for the changes In the parameters and infrastructure that have occurred and that will inevitably continue to change, though only time win reveal In what direction and with what speed. To structure the following discussion of policy directions, we will take up issues connected first with supply and then with the utilization of scientific and engineering personnel and conclude with some broader considerations bearing on policies affecting the educational and research infrastructure. With respect to the future supply, it is important to note the following: . It will prove much more difficult in the decade or two that lie ahead for the United States to expand its supply among the native population. The best yield would come from encouraging more women to select science or engineering as a college major, and later on to pursue a career in these sectors; but under the best of circumstances, it would be wrong to assume that significant increases are likely in the near or even middle term. With widespread and continuous encouragement from the larger society, the educational system, and employers, increases are Rely; but the rate of change will, at best, be relatively slow. - 39
· When it comes to the Black and Hispanic minorities who will represent an increasing proportion of college-age youth in the decades that lie ahead, the challenge is much greater, since these groups are senously underrepresented among high school graduates capable of pursuing higher education and, more particularly, higher education in science or engineering. Hence, any substantially enlarged flow will require a longer time penod than a decade, sufficient time to see a substantial improvement in the economic status of minority families, improved schools in ghetto areas, and other basic adjustments that require long lead times. The best prospects for any significant increase in the size of the pool could come about if larger numbers of college students were encouraged to improve their mathematical skills, after which they would have the tools to pursue majors in science, engineenng, computer science, or technology. But the prospects of such a reform would depend upon the existence and maintenance of a much enlarged demand for such scientific personnel as reflected in higher beginning salaries as well as more attractive career opportunities. Neither of the above preconditions appears at this time to be on the honzon. There are two further ways in which the pool could be expanded: the present retirement age could be increased and He immigration laws could be amended to provide a higher pnont-Y rating for persons with the desired skills. What the above helps to make clear is that if a combination of presently unforeseen circumstances requiring a substantial increase in the number of scientists and engineers were to anse, considerable elasticity remains to accommodate such an increase over a reasonable number of years, primarily by encouraging more women college students to opt for such training; by encouraging some part of the large non-physical science groups to shift to science; by promoting delayed retirements; and by modifying our immigration regulations. Over a long time period, the Yield from better prepared Black and Hisnanic cohorts should also be expandable. _~ ~ . - ¢--r 1he other mayor area tor potential adjustment to a much enlarged demand for scientific and engineering personnel would involve improvements in the utilization of the existing supply which, it must be remembered, in the case of engineering is roughly 15 times the number of annual new entrants. The following paragraphs suggest some of the important areas where significant gains In utilization should prove possible: . . · A more systematic effort by large employers to provide more disservice and external opportunities for con~anu~ng education of their S/E staff. A review by large employers aimed at preventing excessive "Iock-in" of S/E personnel on highly specialized work to a point where they lose their capacity to shift to other work. More attention to the appropriate division of work responsibilities, as between scientists and engineers on the one hand and technicians on the over, to be sure that the lath have the responsibility for most routine assignments. Larger capital investments by employers, particularly in the rapidly advancing computer technologies, which win enable scientists and engineers to increase Heir productivity substantially. Proved direction and leadership of large scientific and eng~neenng groups in both the defense and nondefense sectors. A considerable body of evidence points to potentially large gains In utilization that would follow upon improved organizational and manage ial practices. The foregoing does not pretend to exhaust the many existing opportunities that have long been present and that continue to exist for significant gains in utilization . The fact that 40
many of them have not been addressed, or addressed only half-heartedly, points to the absence of any acute long-term shortages that senior management has recognized to the point where it is willing to devote effort and resources to their elimination. Since the cycle of preparation for entering upon a career in science and engineering is elongated and since many, especially scientists, look forward to pursuing a life-time career in their chosen field, the following improvements in the educational and research infrastructure could have positive results: · Since the federal government plays such a dominant role in the financing of most basic research, as well as most research and development that is defense- related, it should seek more than heretofore to avoid wide fluctuations in its funding cycles. Large and sharp fluctuations in government spending impose heavy costs by encouraging many people to respond to the new signals by sudden shifts in their educational plans and their professional work. Since scientists and engineers are trained by colleges and universities, it is essential that all the parties involved in the financing and direction of higher education--the federal government, the states, and business--str~ve to strengthen and stabilize the academic base to ensure that requisite faculty and other resources such as equipment are available to provide a flow of graduates properly prepared to pursue careers as successful professionals. Since the problems involved in assuring an adequate national supply of well- educated scientists and engineers and their effective utilization have come center stage not once but repeatedly since the end of WW IT, and since Me successive debates concerning these issues have repeatedly suffered from inadequate data, weak models and fault analyses, the knowledge base required for sound policy formulation urgently needs to be strengthened. Accordingly, the National Academies of Engineering and Sciences should take the lead to obtain the resources required for an ongoing and enlarged research effort that could improve the decision-making process with respect to S/E personnel both inside and outside the federal establishment. . . 41
MODELING THE SUPPLY OF SCIENTISTS AND ENGINEERS: AN ASSESSMENT OF THE DAUFFENBACH-FIORITO WORK Michael S. McPherson Williams College Introduction This paper reviews the present state of our knowledge in modeling the supply of scientists and engineers. It focuses especially on what we know about the capacity of the supply system to respond to shifting defense requirements. Such shifts might either be temporary, requiring rapid and flexible response, or longer term, requiring more lasting adjustments in levels of supply. This paper doesn't discuss the likely magnitude of such shifts, or the impact of spending shifts on demand for personnel--those are demand-side questions. The question here is "If significant shifts occur, how well equipped are we to assess the consequences?" Although this paper, like the DauffenBach and Fiorito work it focuses on, comments on ah science and engineering fields and on personnel with advanced degrees as well as entry-level degrees, its main concerns are with engineers and with bachelor's level supply. This is in keeping with the overall concerns of this project and is also where there is greatest concern about the adequacy of dle supply system. There are several reasons for giving most attention to the work of Robert DauffenBach and Jack Fiorito (D-F hereafter). Theirs is the most comprehensive and analytically challenging among recent projection models of the supply side of technical labor markets. They are comprehensive both in coverage of fields and degree levels and, more important, in sources of supply. In particular, their work includes serious attempts to come to grips with "occupational mobility" as a source of supply. Since in recent years less than half the additions to engineering employment have been persons with fresh engineering degrees, mobility is obviously a key problem to analyze. D-F's work is important also because it is a central component in the influential NSF mode! of science and engineering labor markets. Some general Signets on the aims of work in this area should be noted. The D-F work (as well as other supply projections) focuses on the supply of personnel to particular occupations, narrowly or broadly defined. They do not attempt to examine relative supplies to different kinds of employers within an occupation (e.g., government, industry, and academics; or military and non-military). This work, then, doesn't address one impor~t range of utilization issues--in particular, it doesn't address the important question of how personnel might be reallocated between military and industrial uses in response to changing defense requirements. Such reallocations could have important effects on industrial productivity. Lee D-F work and similar models also 1leglect issues about qz~ali~ of personnel, at least in their formal modeling. The models measure stocks and flows of numbers of personnel and make no attempt to distinguish, for example, the quality or productivity of top-ranked and lower ranked graduates, or of entrants from other occupations and newly degreed personnel within the occupation. -rid ~ D-F's informal discussions include some perceptive comments on these matters, as we shall see, but they are not embodied in their models. These are generic limits to what quantitative supply models do; some limits specific to the D-F efforts will be noted below. 43
General Features of the D-F Models Modeling Components Figure ~ (reproduced from D-F, 1983) provides a useful overview of the D-F modeling framework. Starting from the stock of science and engineering personnel in a given year (disaggregated by field and education level), movements into and out of that stock are projected for the following year. Additions to the stock--additions to the supply of The basic strategy in each component Is to estunate relahons~ps based on past experience with that component and to use those relationships to project future values of the component. personnel--are analyzed through three model components ~ A JO (~) New entrants to a field are newly degreed personnel (the degree need not be in the employment field). In their 1980 and 1983 models for the National Science Foundation (NSF), the production of new entrants was analyzed through four subcomponent models: projecting the number of persons obtaining degrees, their distribution among curncula, their choice to enter the labor market, and their choice of occupation. Two of these steps-- the choice of curriculum and the choice of occupation--were made responsive to labor market conditions fas projected by the Bureau of Labor Statistics (BES)l, so that students were projected to be more likely to enter large or growing fields. In a more recent projection of engineering degree attainment for the Engineenng Manpower Commission (EMC), D-F adopt a similar approach, relating levels of degree production in venous engineering fields to demographic variables and the level of demand for engineering employment (as projected by BEST. (2) Occupational mobility flows measure the entry to a science and engineering profession by persons previously employed in another science and engineering profession or in another part of the labor force. Mobility includes persons reentering a field after a time spent out of it. In their 1980 effort for NSF, D-F developed a mode! that projected occupational mobility on the basis of the age and education of a field's members and the size of the labor force in that field. This mode] did not make mobility depend on labor market conditions. That dependency is introduced in the 1983 model, but other features of the 1980 model are dropped, for reasons discussed below. (3) Immigration to the U.S. science and engineering work force is handled through a simple mode! that links rates of immigration to a field to the rate of employment growth as projected by BES. This component is not further discussed here. StructuralFemures It is important to underline some of the key structural features introduced by D-F in welding these components into a model of supply: (~) Their models make the supply of personnel responsive to demand, thus incorporating a key feature of real-worId labor markets. The reverse feedback, from supply to demand, is, however, suppressed in Heir models to keep them manageable. This raises some technical problems in Heir estimation procedures, since employment is treated as an exogenous measure of demand, rather than as the endogenous outcome of supply-demarld equilibration. But Heir approach is, in this respect, a reasonable compromise in keeping He project manageable. [These are "intermediate non" projections. Their shon-run "pipeline" projections in the same study are not discussed here. 44
non-S&E field (out of scope) movers stayers occupational mobility pattern _, retention in_ Lit attrition _ (out of scope) S & E field mobility inflow , l new S&E _ additions stock to stock ~ grants foreign national students Figure 1. Schematic representation of the S/E labor supply system. Occupational envy ral~r~orce entry New Entrants curriculum choice tot~grees ~ (2) D-F do not model the process by which supply accommodates to demand. When shortages occur in realiny, Me effect is to raise wages, stimulate films' recruiting efforts, and so on. As Hansen (1984, p. 94) notes in commenting on D-F, this sort of adjustment "is not made explicit in the model." (In economics jargon, this is a "reduced fonn" rather than a "structural" model.) Explicit modeling of these adjustment processes would lend more confidence Hat the projections capture the forces at work and might give a more adequate fee} for the dynamics of adjusunent. For example, as Hansen notes (1984, p. 95), an explicit mode} of the adjustment process might reveal a tendency for supply to overshoot or undershoot in response to shifting demand [compare Freeman (1976) on cobweb adjustment In engineenng]. (3) D-F Beat the components of their mode! as independent or as sequentially determined. For example, In the 1983 NSF model, the level of new entrants is de~nnined 45
prior to the determination of mobility flows: new entrants can influence mobility, but not vice versa This suppression of interaction among mode} components (like the Reagent of demand as exogenous) simplifies the operation of the mode! but may miss significant phenomena. These limitations are not pointed out in a spot of finding fault. No manageable model can be fully general, and the compromises D-F have made are sensible ones. They are also admirably explicit about the choices they have made. Still, these limits do bear on the interpretation and uses of their results. The New Entrants Component A key feature in assessing the usefulness of the D-F mode! for projecting the consequences of changes in military spending is the linkage of supply to demand. In the NSF version of the D-F model, that linkage appears in two subcomponents of the new entrants component (as well as elsewhere in the model): (~) the choice of curriculum by students and (2) the choice of occupation by labor-market entrants. Funicular choice is linked to the level of demand by occupation, with a lag (so that, for example, the distribution of bachelor's Melees by major field depends on the distnbution of employment four years earlier). Choice of occupation by degree recipients is linked, however, to the rate of change in employment. It's not clear why one choice is linked to the level and the other to the rate of change in employment. In conceivable circumstances, this could produce odd results: a large field that was not growing would attract many majors who would then avoid the occupation. _. . . . . . the estimated relationships in the equations underlying 'tine projections suggest fairly strong relationships between labor-market variables and student choices--stronger for graduate level than undergraduate choices. In the NSF projection results, however, these links don't show up very strongly. We can, for example, compare two demand scenarios considered by D-F in which employment of aeronautical engineers differs by about 25% (or 23,715 people) in the final projected year (19871. New entrants in 1987 in the two scenarios differ quite modestly (2270 vs. 2029), only about Il%. No doubt a major reason for this is the lag In effect on curricular choice. The projection is only for six years; and for bachelor's students, there is a four-year lag assumed in the effect on choice of major. It is difficult to judge whether the dynamic adjustments implied in these relationships would look sensible over a longer projection period, when the lag effects could work themselves out more fully. One way to find out would be to run longer term simulations with the D-F model. The recent D-F work on engineering for EMC examines only the degree attainment and curricular choice subcomponents of a new entrants' model, but it considers those over a longer time frame. This interesting study links rates of degree production by field of engineering and level of degree to demographic trends (number of ~ 8-year-olds) and lagged employment levels. . This work shows quite strong responsiveness or student Interest to Increases In Demand for engineers, especially at the bachelor's level. Thus, BES projects a rise in engineering employment of about 50% between 1980 arid 1995, arid D-F project this would cause an increase in bachelor's degree production between 1981 and 1995 from 63,000 to Il3,000--nearly an 80% increase. D-F speak of this as an estimate of Me "need" for new B.S. engineers, but this may be somewhat misleading. It actually projects the extent to which student entry to engineering would be induced by job growth at the rate BES projects. This Knight over- or under-shoot actual needs. Thus in 1981, new bachelors were about 5% of overall engineering employment. In 1995, on these projections, they would be about 6.3% of 46
overall employment. If job openings as a fraction of employment remained constant over this penoct, then these projections would imply that new bachelors would be fining a larger fraction of job openings in ~ 995 than in the recent past. If needs were roughly being met In the early 198Os, this might be taken as a sign of oversupply at the later date; alternatively, if one sees the recent past as a time of shortage, this implies that supplies would be somewhat more adequate in 1995. Both the NSF and the EMC work mode! only the student choice aspect of new entrants. As D-F emphasize, institutional constraints on educating the number of students who may seek degrees in engineenug or science are not modeled. But currency, many engineering schools report faculty shortages and face venous difficulties in expanding engineering enrollments. It would clearly be desirable, In principle, to project the capacity of colleges and universities to supply places in engineering (there do not seem to be capacity problems in most science fields). This is a difficult job, since it would require modeling the behavior of individual non-profit institutions in deciding on the size of their engineering enrollments, as well as modeling the potential entry of new engineering schools. A needed subcomponent of such a mode} would be a model of the academic market for engineering Ph.D.s, since their availability is an important constraint on expanding engineering schools. The importance of this problem is noted in informal discussion in D-F (1984) and DauffenBach (1984), but it has not been incorporated into their models. Occupational Mobility As noted above, the two versions of the supply mode! D-F developed for NSF (1980, 1983) beat occupational mobility quite differently. The earlier effort focused on Me heterogeneity of the pool of potential movers: not all personnel are equally likely to make a switch. The mode} they developed tried to account for historical patterns of movement as a basis for making Projections and found that the most important factors influencing ~ . ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ · ~ ~ ~ ~ · ~ movement were the age or the worker (older workers are less moulle) and his or her education (more educated workers are less mobile). Their mode! also captured the greater likelihood of movement between fields that were closer In the skins they drew on. However, a key drawback of the early NSF monies was that it did not make mobility rates sensitive to market conditions. Since it is clear empirically that changes In movement patterns among experienced workers are closely linked to shifts in demand, this was a very important omission. Unfortunately, D-F's attempt in their 1983 study to add market condition variables to their earlier mode! failed: they were not able to capture this market sensitivity emp~caDy In a way Mat led to empirically plausible results. D-F (under considerable time pressure) responded by ~n~oducing a wholly different analysis of mobility. Their new analysis in effect shifted the focus from the sources of , , supply of mobile workers to the size of the demand gap that needed to be filled For each field they were modeling, D-F in effect estimated from historical data the correlation between the hiring of mobile workers and (a measure of) the degree of excess demand for workers in that field and used this estimate to project future reliance on mobile workers. This analysis (as the authors recognize) comes very close to assuming that all otherwise unfilled demand will be satisfied by mobile workers (compare Hansen, 1984, p. 96~.2 This approach has the virtues of capturing the close empirical relation between He state of employment demand arid the degree of reliance on mobile workers, as well as 2The link is made tighter by the facts (a) that the authors treat demand as exogenous (so that all of the correlation between supply and demand is treated as supply adjustment) and (b) that the authors' demand measures don't include any allowance for unfilled vacancies. 47
calling attention to the importance of the phenomenon of Seld mobility, which has been widely neglected in work on supply. It's clear, however, that the approach embodied in the recent D-F work for NSF leaves a great deal to be Resow. Unlike their earlier work, this mode! sheds no light on the composition of the mobile work force. Had D-F succeeded in adding mobility to their earlier model, they could have learned something about how the composition of the groups mobile to a field shifted with the state of excess demand. We refight hypothesize, for example, that when excess demand is moderate, most mobility is from workers in closely allied fields, or from workers Pained in the field but working elsewhere; as excess demand increases, workers may be drawn from further afield, with attendant implications for training costs and quality. The D-F mode! sheds no light on this question. The earlier work also includes an explicit mode} of the pools from which mobile workers are drawn. The long-run implications of reliance on mobility plainly depend heavily on the rate at which these pools are being replenished. This question is not raised in the recent D-F work. Analysis of occupational mobility and its implications is clearly of great unportance. In recent years, more than half of the new hires in engineering as a whole have beers in- mobile workers. The Office of Technology Assessment reports that similarly high rates held throughout the decade of the 1960s (1985, p. 971. The recent D-F work suggests what is almost certainly true, that this sort of mobility has proved and will continue to prove a very effective means of solving short-run labor shortages in technical fields. In fact, in periods too short to train new workers, mobility is essentially the only way to respond to unanticipated demand increases. (It also provides a valuable alternative for workers In fields that experience short-run demand declines.) The longer-run implications of reliance on mobility to respond to growing demand may, however, be quite different. There is clearly Me possibility of "using up" stocks of qualified mobile personnel if high in-mobility rates to a field or set of fields are sustained over time. Thus, in the early 1970s a sluggish market for engineers sent many into other occupations, and they have been available to meet some of the rising demand of recent years. That stock may eventually be depleted. On the other hand, physicists, mathematicians and even social scientists are also among the in-mobile to engmeenng, and the stocks of such personnel tend to be replenished over time. We know very lithe empirically about the relationships among these stocks and flows. D-F have good informal discussions of these issues in their work (DauffenBach, 1984; D-F, 1983), but Weir recent empirical and modeling work does not illuminate them. One further perspective on the mobility issue may be worth bringing to bear here. ~ the short run, reliance on mobility means drawing down available stocks of workers in other fields. But if mobility is a stable part of He long-run supply of personnel to a field, it may be useful to view "occupational mobility" and "direct training" as alternative technologies for producing new workers. For example, midleve! management in a technical firm may be produced either directly, as when the firm hires a freshly-minted M.B.A., or through occupational mobility, as when the firm promotes a bachelor's-level engineer. Bachelor's-leve! engineering jobs can similarly be filled either directly by new eng~neenng graduates or indirectly by persons with general science degrees who are hired into engineering jobs after several years' experience in scientific work in industry. Which technology is superior for "producing" workers in a given case depends on many factors. Among them are the precise character of the job requirements, He relative costs (including opportunity costs) of alternative training routes, and the risks involved in acquiring more general versus more specialized training. It's possible to see the current situation as one in which universities perceive the costs of expanding engineering enro~nents and degree production as high and are, therefore, in effect forcing industry to rely on the "alternative technology" of hiring experienced workers trained in other fields. Whether this is a sensible long-run response requires investigation. 48
Conclusion: Uses and Limits of the D-F Modeling Efforts My conclusions address two questions. First, how well do the D-F models handle the specific questions about the capacity of the supply system to respond to shifts in military spending raised at the outset? Second, what are the broader strengths and limitations of the D-F framework? Short-run surges in military requirements for science and engineering personnel raise quite different issues from longer-run secular increases in military requirements. Regarding He short-run questions, the D-F work gives a sensible qualitative picture of how the supply system would respond. There would be little short-nln supply response from new entrants. Their models assume lags in the response of curricular choice to demand shifts (four years for bachelor's and Ph.D.s; two years for master's), which rule out much quick response. There could be in their mode! more response from shifting occupational choices of new graduates, but this is fairly limited too. It's more limited, of course, the broader the set of fields for which demand nses. The D-F mode! implies that the short-run response to higher demand will come largely through occupational mobility and that response will be rapid and s~ong. This seems clearly right as a general matter, and it underlines the short-run flexibility that mobiliny lends to the science and engineering supply system. Unfortunately, the D-F ~ ~ ~ · ~ ~ ~ ~ ~ ~ ~ . ~ ~ model aoesn t ten us much about where these mobile personnel come from; nor, as noted earlier, is it designed to say anything about the reallocation of personnel within a field between defense and nondefense uses. Finally, in regard to Be short run, the dynamics of the mode! are not rich enough to give reliable answers to questions about whether the supply response will overshoot longer-run requirements. Suppose, for example, that engineering employment rose abruptly for two years and then dropped back to an earlier level as a result, say, of shifting Congressional attitudes toward defense R&D. The D-F model would predict a surge in engineering bachelor's degrees in the second year of the slowdown, followed by a drop in degree production two years later. There would be a sharp increase in in-mobility when demand expanded, followed by a very sharp drop in in-mobility when the drop in demand met up with the surge in supply of new entrants. One shouldn't put much confidence in this sort of projection, since the mode! was really not designed to pack quick responses to such fluctuations. Turning to the long run, the basic concern is with the implications of long-run growth in military requirements for science and engineering personnel, especially if accompanied by strong growth in industrial needs for such workers. Putting together the D-F work for NSF (1983) and for EMC (1984), one would reach the following assessment: degree production in the affected fields--or rather decisions by students to seek such degrees--would be strongly encouraged by rising demands. The EMC projections suggest that the response might be strong enough in the long run actually to increase relative supply to fields in sustained high demand. However, possible institutional constraints on the production of such degrees are not modeled in the D-F framework and might lead actual degree production to fan well short of their projection. If new degree production falls short of demand, their mode! projects that occupational mobility will make up the gap. But as a long-run projection--in contrast to the short-run response discussed above--this may not be a convincing or reassunug response (as D-F clearly recognized. The mode! does not tell us where the mobile people are coming from, or whether long term reliance on these sources will deplete available "reserve stocks" of qualified personnel. Nor do D-F have any means in their mode} to capture the process by which (we may suppose) employers move from hiring more qualified to progressively less qualified personnel as excess demand pressures increase or persist. These are issues which future supply models must find ways to address. Let me turn finally to the general strengths and weaknesses of the D-F modeling 49
work. In some ways the strengths and weaknesses are the same. Thus, the hallmark of the D-F work is the effort to provide a comprehensive framework for modeling the supply of science and engineering personnel. The strengths of this ambitious approach are clear: it provides an overview of the supply system, especially of the interconnections among science and engineering fields--and between them and the rest of the labor market--that are created through occupational mobility. At the same time, the ambition of comprehensiveness limits D-F to data that can be assembled on a consistent basis across the range of fields and degree levels and constrains their ability to incorporate Be institutional peculianties that may apply within specific fields. If, for example, one set out to construct a mode! of the supply of Ph.D.s in the life sciences, one could draw on more refined data and take better account of institutional features Intake NIH traineeships and the prevalence of postdoctoral fellowships, which a broadly gauged mode! is forced to ignore. These special features will matter more for some fields and degree levels than others. To note this is not to criticize D-F but simply to observe an inescapable trade-off. The second broad ambition of the D-F mode} is to incorporate feedback from demand to supply. Such feedback effects are clearly important to capture in supply models, and the only drawback of the D-F effort is that it makes one so aware of how much further it is still necessary to go. The D-F model makes us aware of the various points at which supply can respond to excess or deficient demand--through choice of curriculum and of occupation by new entrants and through the mobility decisions of experienced worker--and brings out their differing Importance in shorter and longer runs. But the D-F efforts whet the appetite for more. It would be desirable to have a better dynamic specification of market responses--one that explicitly models adjustment processes, examines the channels through which information about demand conditions is communicated to the supply side of the market (e.g., wages and recruiting efforts), and accounts for the influence of expectations. A more structured, but sills market-sensii~ve, analysis of mobility processes would also be very desirable. D-F underline the critical importance of mobile workers in filling demand gaps. But that analysis must be integrated with a picture of where the mobile workers come from, of which workers from the potential mobility pools tend to move, and of the processes by which pools of potentially mobile workers are replenished as well as depleted None of this is to denigrate the very substantial achievements D-F have to their credit. Their work has provided an improved understanding of how these labor markets respond. Not least among their accomplishments is providing models that direct policy discussions to the right questions and direct future research in promising directions. From the policy perspective, D-F stress, rightly, that the issue in labor markets is not whether supply and demand will be brought into balance, but how, and with what costs, the adjustment will be accomplished. This has directed the policy discussion away from questions about supply-demand "gaps" toward discussion of quality, of adjustment strategies, and of mining and retraining costs. This is where discussion belongs. From the research perspective, D-F's work points to valuable directions for future inquiry. Among these are the need for better dynamic specification and for better understanding of ~nterf~eld movements, which have already been discussed. D-F have also begun the process of recognizing the heterogeneity of the science and engineering labor force, In terms of likely mobility and quality and character of training, which must be incorporated in future work. Finally, D-F's work highlights the need for better understanding of the role of colleges and universities in influencing the supply of science and engineering personnel. It's clear that their decisions about such matters as limiting engineering enrollments, expanding or constricting Ph.D. production, and so on play cndcal roles in the overall workings of the supply system and are poorly understood. 50
References DauffenBach, Robert C., and Jack Fionto. "The Engineenng Degree Conferral Process: Analysis, Monitoring, and Projections," Oklahoma State University and Oak RidgeAssociated Universities Labor and Policy Studies Program, November 1984. "Projections of Supply of Scientists and Engineers To Meet Defense and Nondefense Requirements, 1981-1987," Washington, D.C.: National Science Foundation, April 1983. -----, and Hugh Fold. "A Study of Projected Supply/Demand Imbalances for Scientific and Technical Personnel," Washington, D.C.: National Science Foundation, 1980. "Supply Projections of Scientific and Technical Personnel: Dynamic Response to Changing Employment Requirements," Annual Meeting of the American Association for the Advancement of Science, May 29, 1984. Freeman, Richard. "A Cobweb Mode! of the Supply and Starting Salary of New Engineers," Industrial and Labor Relations Revieu', vol. 30, no. 2, Ianuary 1976. Hansen, W. Lee. "A Review of Four Studies," In Labor-Market Conditions for Engineers: Is There a Shortage? Proceedings of a Symposium, Washington, D.C.: National Academy Press, 1984, pp. 75-98. Office of Technology Assessment. Demographic Trends and the Scientific and peering Work Force--A Technical Memorandum. Washington, D.C. Government Pruning Office, December 1985. 51 Engi- : U.S.
WHAT CAN DEMAND AND MANPOWER REQUIREMENTS MODELS TELL US ABOUT THE IMPACT OF DEFENSE SPENDING ON THE LABOR MARKET FOR SCIENTISTS AND ENGINEERS? w. Lee Hansen University of Wisconsin-Ma~ison Introduction This paper reviews the demand and manpower requirement moclels for scientists and engineers (S/E). Its purpose is to examine the usefulness of these models for assessing the impact of defense spending, particularly the Strategic Defense Initiative (SDI) program, on the nondefense labor market for S/E personnel. Of special interest is the ability of these models to indicate whether, in light of the prospective supply of S/E personnel, enough manpower resources will be available so that defense spending programs can move ahead at their projected rates without curtailing productive activity in the nondefense sectors of the economy. Present and projected levels of defense spending and the advisability of initiating the SDI program have been the subject of enormous discussion and analysis (Congressional Budget Office, 1983, 1984, 1986; Office of Technology Assessment, undated; Aspin, 1984; Penner, 1984; and 1'hurow, 1986). The reason for our interest is concern that the heaw utilization of highly specialized S/E personnel will divert these ~ ~ ~ = CF I .r · _ , _ resources from the nondefense sector of the economy, thereby slowing our rate of technics progress, reducing our capacity to compete effectively in world markets, and limiting our ability to accelerate Me nation's rate of economic growth. Of particular concern is the rapid projected grown of the SDI program, which will be intensive in its utilization of highly specialized SfE personnel. By way of illustration, the SDI program--as a percentage of the Department of Defense's (DoD) Research, Development, Test, and Evaluation (RDT&E) budget--nses from less than 4 percent of the 1984 plan for RDT&E to almost 16 percent by 1989 (CBO, 19841. Thus, it is possible that the SDI program could not only intrude on nondefense activity but also restrict other defense achnty. For these reasons it is essential to have a better understanding of the S/E demands general by the defense and nondefense sectors. To the extent Cat the objectives of these various activities conflict, we have two alternatives. One is to live with these conflicts, letting market and perhaps political forces resolve ~em; the other is to devise policies and take actions designed to reduce, if not eliminate, these conflicts. Whether much can be done through policy measures to reduce them is moot. Setting the Background IdeaDy, a fun assessment of the supply and Remarry for SIE personnel would have been an integral part of planning new defense initiatives. The idea of developing [Preliminary discussion of these models took place at a January 1986 workshop sponsored by OSEP-NRC and summarized in OSEP, 1986. 53 t
manpower impact statements to accompany new programs received considerable discussion in the 1960s with the elevation of manpower concerns to national importance. Producing such statements is a formidable task, however; so much so that the idea of manpower impact statements never took hold. In view of the importance attached to recent proposals for increasing defense spending and implementation of the SDI program, it is surprising that so little analysis has been done.2 Not only is the narrow technical feasibility of the SDI program at issue, but there is also uncertainty about whether the unique constelIai~on of S/E personnel required by this program will be available. Considerable effort has gone into developing a system for exploring these issues, but the system has not yet evolved to the point that anyone can be reasonably confident about resolving them. Thus, it is hoped that the Office of Scientific and Engineenng Personnel study will help to fill the void. Whether the results can influence decisions already made or yet to be made remains unclear. Why should we ask the question posed in the title of this paper? The simplest answer is cur~osi~: how big an effect will the defense and SD! programs exert on He S/E labor market, and how do we go about estimating this effect? The size of the effect is far from obvious, and the method of measuring its size poses art interesting challenge. A more complicated answer is that we need to know the magnitude of the effects on S/E personnel so that we can determine whether output in other sectors of the economy will have to be curtailed. This requires a more diligent effort because we must establish both the nature and the extent of the interdependencies among the sectors. A still more complicated answer, but one of primary interest to most people in Washington, is the policy response to whatever we can learn about the effects of the defense and the SDI programs. Put another way, to what extent can He available policy instruments, or new ones that Night be created, help ensure that the objectives of defense programs are achieved, that neither the SD! nor the non-SD! defense programs are compromised, or that both such programs can go forward unaffected by each other? The natural and rapid gravitation to policy concerns exemplified by the query about what federal policymakers can do raises the intriguing and as yet unanswered} question: What policy instruments are available to deal with direct and in direct effects of defense spending, especially SDI, as they affect the labor market for S/E personnel? A widely prevalent view is that these instruments are quite limited. Indeed, we would have difficulty producing a list of policy instruments that could have any substantial effect on the S/E labor market.3 Were there an extensive list of policy instruments available, we could use its essential elements as a focus for evaluating the venous Remarry models. For example, we might want to determine whether the venous demand models generate ~nformai~on that would feed into and potentially trigger one or more of the available policy instruments. Were this possible, this paper could have a sharp and direct focus that would obviate the need subsequently to translate into a policy context our findings on the demand models. Of course, this assumes that the federal policy instruments can be utilized with reasonable speed and produce their advertised effects. Neither of these assumptions should be accept as fully plausible, however. In the absence of this inventory, we face a quite different task. We must try to identify He elects of interest from the standpoint of the S/E labor market, determining to 2This point has been emphasized by Aspin (1984), and Thurow (1984 and 1986~. 3For an effort to show what federal policy instruments are available to affect graduate education in science and engineering, see Alan Fechter, "I he Effectiveness of Federal Programs for Science and Engineering Graduate Education," a paper presented to the U.S. House of Representatives Committee on Science and Technology, Task Force on Science Policy, July 9, 1985. 54
what extent the various demand models can illuminate how the S/E labor market operates. The results should still be of interest to decisionmakers--namely, those individuals and Cans who find themselves participating in or affected by the defense program. The labor market information produced will not only reveal what is happening but may also produce responses that will alter Me condition of the S/E labor market and eliminate unbalances Hat might otherwise be a matter of concern. If this is the case, then the best of all possible outcomes may be achieved. In a sense, we have been asking, "Who is the audience for projections of employment?" The policymakers have too few levers to press. Employers must respond to current market pressures regardless of what the projections indicate. Prospective workers may be ill-advised to place much faith in projections of requirements because past ones have usually been considerably off-target. In any case, monitoring current and prospective labor-market conditions is something that the private sector already does and, hence, one could argue that the demand models are of relatively little use to most people in the S/E sectors. This does not, of course, preclude our interest In the topic. Establishing the Criteria for Evaluating the Demand Models Rather than plunging immediately into a detailed analysis of the venous demand models that might be employed to examine the impact of the SD! program on S/E personnel, it is important to try to characterize what kinds of infonnation that we want about the operation of the S/E labor market and then to contrast this with what the different models provide. We draw upon the knowledge and experience of labormarket analysts and their efforts to identify key infonnation to analyze labor markets (OSEP, 1984; COPAFS, 1985). Our list of infonnation is presented below, phrased in the form of questions. We act as if we are back in 1984, when the SDI program was first proposed; this helps to avoid complications that arise because the SDI program is already under way. I. What is the planned and what is He likely pace of anneal growth in the SD! program and overall defense spending over each of the next five years and through the five years beyond that? We recognize that SD! is a major developmental effort that win extend over a long period. However, the pace of development each year depends cnticaDy upon progress achieved through the previous year. The only way of dealing with this is to understand the evolution of the program from year to year. This requires not only longer-term projections of, say, five and even ten years but also annual updates to track new developments. 2. What is the likely variance in the pace of He program's development? The technological difficulties appear to be enormous, with the result that delays in accomplishing certain critical tasks are likely to slow the development of the entire program. The uncertainties appear to be far more substantial than those encountered in most sectors for which demand projections are made. As a result, we wart to know the likely annual range in expenditures and utilization of S/E personnel arising solely because of these uncertainties. 3. What kinds of knowledge and skills wiD be required of S/E personnel to assure the technical progress necessary to keep the SD! and other defense programs on schedule? What type of scientific and engineering knowledge will be needed and in what sequence over time? For example, the need for basic research may be heavier in the earlier years of the program's development, with developmental-type activities coming later. Since these activities will undoubtedly require different knowledge and skills, the impact on labor markets win surely change over timee 55
4. What level of occupational detail describing manpower requirements is needed for the various actors in the program--the Grins that win be hiring S/E personnel and also the new and existing S/E personnel who are working or Freight work on the SDI program? Will it be sufficient to produce estimates of manpower requirements based on the traditional occupational classification system that is related to the kinds of collegiate training people obtain? Or will we need much finer classifications of the kind shown by the 3- and 4-digit occupational codes in order to highlight the increasingly specialized nature of manpower demand? Or do we need some entirely different classification system? How do we answer this question? 5. To what extent can these models reflect prevailing elasticities of substitution among different types of S/E personnel? We know there is considerable flexibility in what many S/E personnel can do. At the same time the technologies involved in developing SDI may be highly specific and thus limit substitution of one type of S/E for another. By utilizing a broad classification system for S/E personnel, these substitutions can be ignored. And yet the critical labor market problems are likely to arise because of shortfalls of particular types of specialized personnel. Thus, we need to know how easily employers can shift workers across classification lines and also the extent to which they can shift workers among different job functions, such as research, development, management of research, and the like. Again, this may require a different system for classifying S/E personnel. To the extent that easy substitutability exists. the likelihood of specific S/E labor bottlenecks is reduced. . 6. To what extent can these models encompass changes in the elasticity of substitution between labor and capital resources? As a result of recent advances in computer technology, for example, it is plausible to believe that capital can be substituted for labor more easily and quickly in SDI than in other parts of He defense and nondefense economy. If this is the case, requirements for S/E personnel may rise at a slower rate than ar~i~cipated. How can these changes be incorporated into demand models? And what estimates of elasticides emerge? 7. Can these models capture the extent of substitution between new and recent entrants into the S/E labor market? If experienced workers and expenenced scientists are in limited supply, then perhaps two new degree recipients can be utilized to do what a more senior person would do if available. What do we know about these possibilities? 8. How can we be certain that He demand models capture the effect OF changes in License spending, particularly spending on the SD! program? Most of the models are built on average relationships from which marginal impacts are inferred. Yet the essence of He SD! program is its uniqueness and the fact that it will require a constellation of S/E personnel that may differ appreciably from the present stock of S/E personnel. Unless the particular nature of these marginal impacts can be identified, the results of the models will be off target in pinpointing the very problems that they are designed to help uncover. Some may object to these criteria because they impose severe standards on existing demand models. Indeed, because of Heir may limitations, these models may not receive high marks when evaluated against these critena. The only way to remedy this is to make efforts to enrich these models, to develop alternative models, and to find new approaches so that more effective assessments can be made of the impact of SDI and other defense spending on SIE labor markets. 56
An Inventory of Demand and Requirements Models What types of models or approaches are available? It is useful to list and then to review each approach to gain an appreciation for its potential effectiveness in providing estimates of the impact of the SD} and other defense programs. Macroeconomic Approaches We have a number of approaches that are all closely related but yet differ in significant ways. Each approach is summarized briefly here: I.Bureau of Labor Statistics (BESJ: Employment Projections. Every five years or so, BES produces employment projections on a lO-year horizon for 550 detailed occupations in each of 378 industries. These projections are generated by combining its labor force projection model, aggregate economic projections denved from the Wharton Econometncs macroeconomic model, its own industry demand model, its own industry employment model, and its own occupational employment model, which relies on the BES industry- occupation matrix. The results of these studies are published in the Monthly Labor Review arid also in various BES Bullet~ns.4 2. Data Resources Incorporated (DR]J: Interindustrv Forecasting Model. , O This private firm produces employment forecastsS for 163 occupation categories in each of 82 industries. These forecasts are generated by DRI's Occupation by Industry Model, which combines the results of the BES occupation by industry data, and by DRI's 400-sector employment forecasts, derived from its Intenndusmy Model, aD of which are based on its Macro Model. These results are proprietary and Bus not generally accessible. 3. Data Resources Incorporated: Defense Interindustry Forecasting System (DIES). This system, developed for the Department of Defense, produces employment forecasts for 163 occupational categories in each of 81 industries. A' Amp ' ' ' ~ · ~ . . . . 1ne L) moce1 combines ~lve-year projected defense outlays and already authorized expenditures for 50 budget accounts, which are then converted into final demand by commodity through the Defense Industrial Share Matrix and integrated into the Standard Industrial Classification industry groups. Combined with the results of the DRI Quarterly Mode} of the U.S. Economy, production for defense and nondefense sectors is estimated and then converted through a dynamic input-output model into estimates of direct and indirect production. Subsequently, these production estimates are converted into industry employment through a series of production equations. The final step is to distribute industry employment across occupations with the help of the Occupational Employment Statistics (OES) matrix developed by BES. These results are not easily accessible.6 4U.S. Department of Labor, Monthly Labor Review, November 1985; Bureau of Labor Statistics, "BLS Economic Growth Model System Used for Projections to 1990," BLS Bulletin 2112, 1982; and Bureau of Labor Statistics, "The National Industry-Occupa~onal Matrix, 1970, 1978, and Projected to 1990," Bulletin 2086, 1981. 5Data Resources Incorporated, The DR1 Interindustry Service: Occupation by Industry Model, Washington, D.C.: February 1983. See also Otto Eckstein, The DRI Model of the U.S. Economy, Englewood Cliffs, N.J.: McGraw-Hill, 1983. 6Institute for Defense Analysis, The D~ense Translator. IDA Record Document D-62, June 1984. 57
4. Department of Defense: Labor Defense Economic Impact Modeling System (LDEIMS). This approach, according to the available documentation, is quite similar to the DIFS model. The only real difference is that the DIES model is based on the more highly aggregated budget data published by DoD. The LDEIMS model, by contrast, is based on quite detailed 5-year projections of expenditures that Congress is expected to approve. These two models are likely to produce quite similar results because of aggregation and the fact that the published and unpublished data do not differ substantially. Again, the results are not easily accessible.7 5. National Science Foundation Model. This model examines the impact of defense and nondefense needs on the science, engineering, and technology labor market, using demand or requirements estimates from the DIES model and supply projections from the DauffenBach/ Fionto/Folk (DFF) Model, and the Stock Flow Mode! of Science and Engineering Labor Supply. Two aspects of this approach deserve mention. First, the oetense and noncetense requirements are estimated over a t~ve-year time honzon. Second, annual projections of supply estimates are developed for 21 occupational groups and distinguish between new entrants, occupationally mobile experienced workers, and foreign immigrants. The results have been published by NSF; the DFF results appear in a series of unpublished papers and reports.8 ~ ~ ~ ~ r · , i. , ~ ,~ ,. , 6. institute for Economic Analvsis {IEA i: Dynamic Inout-Outout Model. This mode} was developed by Wassily Leontief, Faye Duchin, and their associates at New York University for the purpose of estimating the employment effects of automation for 53 different occupations In 85 different industries over a long-run time honzon--e.g., to the year 2000. No explicit attention is given to the defense and nondefense sectors but, in principle, there is no reason why this model could not be adapted to estimate defense employment impacts, something that Leontief has done In earlier work.9 Microeconomic Approaches These approaches are more difficult to descnbe, largely because we have few examples that are linked to the defense sector. Nonetheless, several different approaches have been employed, and they are described briefly below. I. Production Funchon Model. This approach In one of its various forms has been applied to particular industries to measure such things as productivity increases and elasi~cii:es of substitution; it can also produce employment forecasts. This family of models is United in its ability to differentiate among various types of labor; typically, this mode} focuses on one and perhaps two categories of labor, such as "the more and Be less educated" or "the more or less skilled." These models are less useful for prediction than for explaining what happened in the past Because these models require fairly extensive dme-senes data, they 7Deparunent of Defense, Defense Economic Impact Modeling System--DEIMS: A New Concept in Economic Forecasting for Dense Expenditures, Office of the Secretary of Defense, July 1982; Department of Defense, D~ense Use of Skilled Labor: An Introducion to LDElMS, undated. See also Department of Defense, Defense Purchases: An Introduction to DEIMS. undated Diagonal Science Foundation, Projected Response of the Science, Engineering, and Technical Labor Market to Defense and Nondefense Needs: 1982~7, (NSF, 84-304y, Washington, D.C.: U.S. Government Printing Office, 1984. 9Inshtllte for Economic Analysis (Leontief-Duchin), The Impacts of Automation on Employment. 1963- 2000, New York: New York University, New York, April 1984. 58
are often difficult to estimate. It should be noted that production function equations constitute part of several macro models, most notably the BES and DRI models.~° 2. Recursive Model. This approach has been popularized by Richard Freeman and is typic ally applied to a single occupational group, with the purpose of not only explaining the past but also forecasting the future. Numerous applications have been made to highly trained occupational groups, including engineers, and college faculty members. Essential features of this approach are lags in the production of new entrants who respond to changing wage levels. As with the production function approach, extensive time-genes data are required. Survey Estimates of Future Demand By Sector A standard technique used over the years entails surveying strategically placed people in an Occupation or industry for Heir best estimates of the level of future demand or requirements for specific types of personnel in the short run. In this case, respondents might be asked to estimate the impact of increased defense spending and the SDI program on employment requirements. i. Engineering Manpower Commission. Periodically since the Korean war, the Engineering Manpower Commission has initiated surveys of individual engineers and also employers to ascertain expected employment changes over the next year or several years. The purpose of its most recent surveys is to provide information for short-term planning purposes. The assessments from both employers and employees of expected demand conditions over the next year make this an especially interesting approach. Typically, however, responses are heavily affected by current conditions and do not do a particularly good job of identifying the magnitudes of actual demand changes. 2. Naizonal Science Foundation Survey S - ies. These annual surveys initiated in the early 198Os ask large fines to indicate the recent, current, and prospective status of the labor market for ~ types of scientists, 15 types of engineers, and ~ categories of technicians. Respondents can be grouped by industry and, within that, by whether they are in defense- rela~ work. This pennits the tabulation of results showing We relative shortage condition for each occupational group, along with projected hiring and an assessment of shortage conditions for the following yearns 3. American Electronics Association.~4 Another example of a more focused effort, though not explicitly on the defense sector, is the survey by the American Electronics Association on annual hiring plans by the electronics industry for the next five years. Essentially, 19Richard B. Freeman, "A Cobweb Mo~1 of the Supply and Starting Salary of New Engineers," Ink and Labor Relations Review, vol. 30, no. 2, January 1976. 1tW. Lee Hansen, et al., "Forecasting the Market for New Ph.D. Economists," American Economic Review, vol. 40, no. 1, March 1980. 12Engineering Manpower Commission, The Demand for Engineers: 1982, New York: American Association of Engineering Societies, Inc., 1983. 13National Science Foundation, 1985 NSF Science and Engineering Labor Market Study, Washington, D.C.: Market Facts, Inc., April 1986. 14Pat Hill Hubbard, "Technical Employment Projections, 1983-1987: A Summary," in Labor-Market Conditions for Engineers: Is There A Shortage? Proceedings of a Symposium, Washington, D.C.: National Research Council, 1984, pp. 11-28. 59
respondents are asked to estimate changes in employment for several different categories of engineers and other technical professionals, their perceptions of the economy and the particular labor markets, and methods of accommodating to shortfalls of particular types of personnel. Ad hoc Models Because the cost of developing the macro approaches is so high and because of the generally unsatisfactory nature of the micro and survey approaches, other approaches have been devised that set out In quite pragmatic ways to estimate in some systematic fashion the future demand for particular types of personnel. One such approach described below attempts to estimate scientific and technical personnel requirements for the research, development, and engineering activities connected with the SD} Innovative Science and Technology Office (ISTO). manpower expert in the Washington, D.C., area. The developer of this approach is Ivars Gutmanis, a - Prepared under contract with the L'epartrnent of Defense, the Sterling Hobe Corporation Models develops annual estimates of scientific and technical personnel requirements for the ISTO component of SDI for~the penod 1986-1990. The approach is quite straightforward and is described as an emp~ncal methodology. The circumscnbed nature of this study, which can be viewed as a pilot approach to estimating personnel requirements for other aspects of SDI, did not warrant developing a more elaborate model. A summary of the complex methodology used follows. To develop estimates of the personnel required to carry out research, development, design, and engineering activities that would be undertaken by ISTO, the study examines the experience of research organizations that were already performing similar activities. From the data for these organizations, it is possible to calculate a set of coefficients showing the average number of employees per unit of operating expenditures necessary to staff a research operation. After matching this information with the ISTO categories and the appropriate levels of operating expenditures, the total personnel requirements for each ISTO area are calculated. The professional component is then estimated for each area, and this is disaggregated into different occupational groups based on data obtained from the research organizations. These estimates are then developed for each year to reflect the buildup of ISTO activity. The approach seems like a plausible one, but it is necessarily crude. How accurate this approach will prove to be cannot yet be ascertained. Informed Judgments by Knowledgeable Experts Despite the fonnal and less fonnal approaches outlined above, we frequently find long-6me experts who possess the institutional background and know the data so well that they can provide qualitative assessments of the effects of complex changes and do so with reasonable speed and accuracy. The judgments reached by such individuals are not easy to replicate and, hence, can be no more than judgments. This approach is the antithesis of that employed by He mode! builders, who in extreme cases know little orno~mg about He world their models attempt to describe. As examples of knowledgeable experts in this field, one cannot help but think of people such as Harold Goldstein and Harold Wool. Undoubtedly, He names of others should be added to this list. Sterling Hobe Corporation, Scientific and Technical Personnel Requirements Related to Activities of Innovative Science anal Technology Office Strategic Defense Initiative Organization Washington, D.C., January 1986. 60
Evaluating the Various Models Because the venous models and approaches differ so considerably, it wall be easier to separate them in two groups. Accordingly, we first examine the macro approaches and then turn to the remaining approaches. Macro Models We list in Table ~ each of He macro approaches and then indicate how they stack up aghast the venous evaluation criteria outlined earlier. First, the capacity to produce annual projections of the impact of defense and other spending on S/E personnel exists for DRT, for DIPS band LDElMS, and for NSF. Each is limited, however, to a five-year time horizon because of linkages to the DoD budget, which covers only the next five years. The likely accuracy and timeliness of these projections is limited by lags in the data but even more important by the use of essentially fixed coefficients. The actual relationships are quite likely to change in response to cyclical conditions, among other factors, and as a result the accuracy of these projections is suspect. Only the BES and LEA indicate that they do not produce short-run projections; they restrict themselves to a 10-year or longer time horizon. To sum up, only two of the six models--DIFS and LDElMS--can provide much help In illuminating the short-run impact of the expansion of defense spending. The uncertainties connected with SD} and other defense programs appear to be recognized but are largely ignored by these models. Perhaps more important, the actual path of development may be affected by delays In essential technical developments, matenal shortages, testing difficulties, and labor bottlenecks. The only way to deal with these uncertainties is to indicate some range in the levels of projected manpower demand. None of the approaches pay attention to the kinds of knowledge and skills required except insofar as they are captured by occupational designations. Nor is it clear from these models how the mix of personnel by occupational category may change as spending programs evolve. Hence, the range of uncertainty is large. Second, a severe limitation of the venous approaches is that they stick with the traditional occupational categories. These categories are not descnptive of the kinds of knowledge and skills required of S/E personnel. The usual categories of engineers (civil, mechanical, electrical, etc.) reflect, to a large extent, the collegiate degree programs from which these people emerge rather than the categories of engineering skills used by employers in their search for both new and already expenenced personnel. As an example, one national recruiting firm that specializes in placing engineers utilizes a 55-item position code, a 37-item listing of areas of competence, and a 10-item function code--permitting identification of both what employers seek and what individual job seekers can do. Without such detailed information, it would be difficult, if not impossible, for the firm to make appropriate job matches. If this amount of detail is essential to S/E labor markets as they actuary operate, one cannot help but wonder about the utility of the usual macro approaches for estimating He impact of defense spending on S/E personnel. The abundant substitution possibilities that exist are given little or no attention by these models. For example, labor substitution across occupational lines is completely ignored. Moreover, substitution between labor and capital is typically hidden behind adjustments made in the ~nput-output matrix, capital-output ratios, and productivity assumptions. Because these adjustments reflect the informer! judgments of the projection team, based on a wide array of information, and because these adjustments are not explicit, students of occupational projections experience difficult knowing whether actual projected figures capture adjustments on the demand side that are of interest. In other words, it would be more informative to have projections made with and without the various 61
- o - ct Id a Ct .c is ED S Cal .= I: . Cal SO do. o U. - Ct ~ · ~ Cal SO o Ct ·~ · ~ m Cal - o o Cal Cal me CQ V, .~ ~ O O ~ ~ ~ o ~ o ~ o o C. o o o C ~ O C ·DO , '~' ~0 C ~ ~ ~ O C~ Z ~ o C) .= .= · _ C~0 C~ OO OO ZV ZC) 7 ~ ~ -;a o 4o ~D ~ ,~ ~, ~ 7 ~ ~ ~ O O ~ ~ ~ O ~ O ~ 0 0~ 0 0 0 0 ~ _ c ~ ]4 ~ ~=a 3 ~§g ~ ~ ~ ~ ~. ~ ~ ~ 62 C~ C~ o s~ - C~ o ~, e~ o · ~ o
adjustments so that changes on the demand side could be isolated. This is particularly important if some of these demand side adjustments are responses to labor-market conditions, such as sudden supply-side shifts. Nor is any attention given to substitution between older and younger S/E personnel. One solution is to simulate different situations and to incorporate into them a range of substitution possibilities so that the sensitivity of the result can be established more precisely. This complicates presentation of the results because it is ordinanly necessary to allow for several different sets of assumptions. Finally, no effort is made to distinguish between average and marginal effects. Because of what we already know about the sensitivity of estimates to even small differences, it seems essential to take account of the unique character of new programs because they are so likely to diverge from the average character of existing programs. In a session this point sums up all of the above points. In sublunary, these models are no doubt useful first efforts, but they do not take us very far in understanding how the SIE labor markets operate. It is not even clear in what sense they reflect requirements for S/E personnel. The basic problem with these models is that their complexity makes it difficult to elucidate the assumption underlying them. Moreover, these models also entail a host of judgments that are difficult to detail.~6 For example, in using the input-output table to project industry employment, adjustments are made for anticipated technological change and attendant labor-capital substitutions. It is difficult to know exactly how these adjustments are made. It is even more difficult to know whether they are truly exogenous adjustments dictated by the on-going pace of technology, or whether instead they reflect, at least in part, responses to future changes in labor-market conditions and alterations in the relative prices of labor and capital. The basis for adjustments In the ~ndust~y-occupai~on matrix to reflect prospective changes in utilization of different types of skills is also unclear: do these adjustments reflect the impact of exogenous factors, or are they too in part endogenous? Another problem is the failure to develop a more explicit modeling of both the demand and the supply sides of the labor market. The term "requirements" suggests a demand-side orientation. Yet projected requirements are taken to reflect what actual employment will be, given the assumptions underlying the projections. This interpretation is substantiated by subsequent comparisons made by BES between its projected requirements for some year arid actual employment for that year, with discrepancies being characterized as errors (Carey, et al., 19821. ~ fact, the venous ad hoc adjustments in the ~nput-output coefficients and in the industry-occupation matrix very likely reflect the implicit introduction of supply-side considerations so that the projections represent something closer to forecasts. If that is the case, they should be described as such. A superior approach niight be to generate projections of requirements based on unchanged input-output coefficients and an unchanged industry-occupation matnx. By then introducing a separate supply model, it would be possible to generate results that, when combined with the requirements models, would produce something that we Freight describe as reflecting "requirements-supply balance." Adjustments in input-output coefficients and Me ~ndustry-occupation matrix would then become endogenous and help to reconcile differences in Drosnective requirements and supply. Whether such a balance would flow out of the interaction of the requirements and supply models is not apparent; they might have to be forced to produce such a balance. The NSF model goes further in specifying shortage occupations by contrasting requirements with available supplies Eat take into account ~nteroccupational shifts as well as flows of new entrants. Unfortunately, we don't know to what extent supply adjustments are already embodied in the projections of requirements. Nor is it clear that requirements or supplies are as inflexible as implied by the model. 16 The following paragraphs draw on Hansen, 1984. 63 .
The ultimate test of these models is to determine how accurate they are. As pointed out earlier, with respect to BES projections, they are not very accurate if they are intended to predict future employment levels. But if they are indeed estimates of the demand side of the markets, there is no reason to expect accuracy because supply forces may dominate over demand forces. Yet the common practice is to compare projected requirements with realized employment totals. Somehow this conflict has to be resolved. The Other Models The other models are more difficult to evaluate, largely because of their great diversity and the fact that their use in analyzing the S/E labor market with respect to defense and nondefense impacts has been quite limited. Hence, we provide no table comparable to Table I. The micro models, both production function and recursive, can be applied-quite flexibly but, unlike the macro moclels, can do little to reflect interdependencies among markets for different types of S/E personnel or the defense and nondefense sectors. These models can generate annual projections, reveal the extent of uncertain through measures of variance or through simulations, and identify a very limited range of substitution possibilities. Offsetting these advantages is the fact that the required data needed to implement these models are unlikely to be readily available. Moreover, the macro environment within which projections are made must be imposed based on other research. To sum up, the partial nature of these micro approaches limits their applicability except for quite stable and weD-defined occupational Soups. The survey approach is also quite flexible. The responses to such surveys are no better than the knowledge and judgment of those who respond. To the extent that respondents differ ~ their position within Be grin and as a result do not have access to the same information, it is difficult to know how to interpret survey results. An additional difficulty comes from the tendency of respondents to project ahead from the time they respond, without recognizing that the current environment captures a vanes of seasonal and cyclical conditions as well as an overlay of unique events. The combination of these conditions often prevents respondents from offering an informed and informative assessment of the S/E labor market (Hansen, 1984~. Only one so-called ad hoc mode} is presented here; perhaps there are others that have not come to the author's attention. In arty case, the Sterling Hobe study provides a sharp contrast to anything else we have reviewed. The fact that its scope is so limited, being confined to the ISTO portion of the SDI program, makes generalizing difficult about the wider applicability of this approach--to the entire SD! program, to the entire defense program, and to the nondefense sector. Perhaps the most important contribution of the study is its effort to use marginal rather than average relationships. This shows up in the assumption that the personnel requirements for the ISTO program will approximate those of firms already doing similar type work. This is a far cry from imposing the BES occupation-industry matrix on changes in industry employment to produce occupational requirements. This approach still does not overcome some of the shortcomings mentioned above: the occupational categories are the traditional ones; there are Shi] elements of the fixed coefficients approach embodied in the method; and He approach does not reflect the interplay between SD! and other defense programs, much less interaction with the nondefense sector. Thus, while not a solution, this approach does highlight the impact of SDI. In this sense, it is a building block for a larger effort to estimate the effect of SDI. Conclusion It is clifficult to come away from this review with a sense that we can descnbe with 64
any certainty the magnitude arid perhaps even the directions of labor market effects on S/E personnel resulting from acceleration of the defense and SDI spending program. Whether the demand for S/E personnel will adversely affect the nondefense sector is difficult to say. We do not yet have the knowledge, the data, and the models necessary to help understand this very complex subject However, we do have an opportunity to learn more about these matters. Soon it should be possible to evaluate the BES projections for 1985. Two years from now the 1987 data will be available, making possible a careful retrospective on the NSF study embracing the 1982-1987 penod. And similarly with the other models. Still, the task of trying to reconcile the projections with what happened will not be easy because these models neglect so many elements. Excluded are wage changes, utilization rates, changes in how work is scheduled, overtime, and alterations in hong standards. None of this is meant to disparage the work that is now done. A wide range of existing and new approaches is required to help us comprehend what is happening and why it is happening. We need additional research work at all levels--conceptual, modeling attempts, new data collection, expanded analyses of existing data, and the like. Unfortunately, the answers win not be easy to obtain. References Aspin, L,es. Defense Spending and the Economy. Washington, D.C., Apn} 1984 (unpublished paper). Congressional Budget Office. "Analysis of the Costs of the Administration's Strategic Defense Initiative, 1985-1989," staff working paper, Washington, D.C., May 1984. Defense Spending and the Economy. Washington, D.C.: U.S. Government Printing Office, 1983. Congressional Research Service. The Strategic Defense Initiative: Program Description and Major Issues, Report No. 86-6 SPR, January 7,1986. Council of Professional Associations on Federal Statistics. Scientific and Technical Personne!in the 199Os: An Examination of Issues and Information Needs (Proceedings of a May 9-10, 1985, conference). Washington, D.C., 1985. Fechter, Alan. "The Effectiveness of Federal Programs for Science and Engineering Graduate Education," paper presented to the U.S. House of Representatives Committee on Science and Technology, Task Force on Science Policy, July 9, 1985. General Accounting Office. Specific Technological Assumptions Affecting the Bureau of Labor Statistics' ·995 Employment Projections. Washington, D.C., May 20, 1985. Hansen, W. Lee. "A Review of Four Studies" in ~or-Market Conditions for Engineers: Is There A Shortage? Proceedings of a Symposium. Washington, D.C.: National Academy Press, 1984, pp. 76-98. National Research Council, Office of Scientific and Engineering Personnel. Labor Market Conditions for Engineers: Is There a Shortage? Proceedings of a Symposium. Washington, D.C.: National Academy Press, 1984. -----. Summary of the Round(table Discussion to Review Demand Models. Washington, D.C.: National Research Council, January 23, 1986. Of lice of Technology Assessment, Ballistic Missile D~ense Technologies, unman. Penner, Rudolph G. "Defense Spending and the Economy," paper presented to the U.S. House of Representatives Committee on Armed Services, February 23, 1984. Thurow, Lester C. "The Economic Case Against Star Wars," Technology Review, February/March 1986, pp. ~ I-15. 65
