Recent Trends in the Federal Funding of Research and Development Related to Health and Information Technology
McGeary and Smith
This paper documents recent trends in federal funding of research related to health and information technology (IT). It shows especially the impact of the halt in growth of federal funding for several years in the mid-1990s, which resulted in a substantial shift in the composition of the federal research portfolio. In broad terms, federal investment in biomedical research has expanded relative to federal investment in most fields of the physical sciences and engineering. That shift has raised the question of whether the federal research portfolio has become “imbalanced.” The paper analyzes the shifts in funding of research fields, compares them with nonfederal trends, and discusses the implications of the reductions in funding during the 1990s of many fields in the physical sciences, engineering, and geosciences.
Part II examines funding of core fields, biological and medical science in the case of health and computer science and electrical engineering in the case of IT. Part III looks at federal funding of research and development (R&D) in health and IT in a national context; that is, how federal funding relates to nonfederal funding. Part IV is an analysis of recent trends in the funding of other fields of research, including those that also contribute to progress in health and IT. In the conclusion (Part V), the question of whether recent shifts in funding—toward the biological, medical, and computer sciences and away from most fields in the physical sciences and engineering—have given rise to an unbalanced federal research portfolio is discussed. To some extent, where one stands on the imbalance question depends on where one sits, but several broader approaches to resolving the question are presented.
RECENT TRENDS IN FEDERAL FUNDING OF LIFE SCIENCES AND INFORMATION TECHNOLOGY RESEARCH
Historically, federal funding of R&D associated with advances in health and the development of computer-related products and services has been substantial and important. This is particularly the case for support of fundamental long-range research where industry investment is often missing.1 Federal investment in the life sciences2 and IT3 increased many-fold from the 1950s to the 1990s. In the 1990s, however, efforts to reduce the federal budget deficit and the post-Cold War reductions in the defense budget affected federal R&D funding in terms of both amount—where funding was reduced for several years—and composition—where some research fields expanded greatly relative to others. Budget authority for federal R&D funding peaked in FY 1992 in real terms. By 1996, it had fallen by 9 percent. Real growth in the R&D budget resumed in 1997, but the federal investment in R&D did not equal the 1992 level until the FY 2000 budget. Research was less affected.4 Federal research funding went flat in 1993 and began to increase again in 1997 (see Figure 1).
As we will see in more detail in Part IV, the pattern of funding by field in the area of science and engineering changed substantially during the period from 1993 to 1997. During this period, overall federal funding for research was flat in real terms.5 The Administration and Congress, however, continued to increase the budget of the National Institutes of Health (NIH), which provides more than 80 percent of the federal support for the life sciences, including the biological and medical sciences.6 At the same time, they responded to the end of the Cold
War by cutting the research budgets of the Department of Defense (DOD) and, to a lesser extent, the Department of Energy (DOE). Historically, those two agencies have provided the majority of federal funding for research in electrical engineering, mechanical engineering, materials engineering, physics, and computer science.
These trends in agency funding of research have affected, in turn, research related to life sciences and IT. Funding of life sciences research has increased while funding for most fields in engineering and the physical sciences has fallen.7 Nearly all the departments and agencies that support R&D have increased their funding of research since 1997 (see Figure 2). As we show in Part IV, however, the reallocation of funding among fields of science and engineering that occurred in the mid-1990s has not changed.
Researchers in the biological and medical sciences have been favored by the budget success of NIH, their main source of federal support. However, support of the biological sciences and the medical sciences differs by degree (see Figure 3). Federal obligations for medical sciences increased more than that for biological sciences, especially in the first half of the decade.
The decline in DOD’s budget, including its research budget after 1993, has put heavy downward pressure on the fields that rely on it for most of their
federal funding. DOD funding of nearly all fields of engineering, including electrical engineering, was down substantially from 1993 to 1999. Because other agencies did not step up their support substantially, federal funding of electrical engineering declined after 1993. DOD, however, did increase its funding of particular fields of research, including computer science. Several other agencies sharply increased their investment in computers sciences research, and that field had substantially more federal support in 1999 than in 1990.
Federal obligations for biological research increased by 28 percent from 1990 to 1999 in real terms, from $5.0 billion to $6.5 billion (see Figure 4). That increase was driven by NIH’s steady budget growth in the 1990s of 67 percent. NIH provides more than 80 percent of all federal funding of biological research8 and accounted for all of the net increase in federal funding of the field from 1990 to 1999.
The emphasis on basic research and university performers is high in federally supported biological research. The ratio of basic to applied research funding remained at about 2 to 1 during the period. The share conducted at universities and colleges increased from 58 percent in 1990 to 66 percent in 1999.
Federal funding of medical sciences increased by 65 percent from 1990 to 1999, from $4.1 billion to $6.8 billion (see Figure 5). NIH accounted for 82 percent of federal funding of the medical sciences in 1999,9 up from 75 percent in 1990, and NIH’s steady budget growth during those years has permitted the agency to make such a large increase in funding.
Funds allocated to basic research and university-based research in the medical sciences are less than those allocated to biological research. Funding shifted a bit toward basic research but away from universities during the 1990s. The ratio of basic to applied research increased from 1.2 to 1 in 1990 to 1.4 to 1 in 1999. Universities and colleges received 46 percent of the funding in 1999, down from 51 percent in 1990.
As in biological and medical sciences research, federal support of research in computer science increased substantially during the 1990s (see Figure 6). Obligations increased by 121 percent from 1990–1999, from $685 million to $1.5 billion. In 1990, DOD provided 63 percent of the federal funding—in the
amount of $433 million—with NSF a distant second at 19 percent. Despite a cut of 25 percent from 1993 to 1999 in overall research funding, DOD managed to provide $538 million for computer science in 1999, compared with $528 million in 1993. Meanwhile, other agencies substantially increased funding from 1993 to 1999—NSF by 110 percent (from $142 million to $297 million), DOE by 339 percent (from $115 million to $506 million), and NIH by 209 percent (from $20 million to $62 million). In 1999, DOD accounted for 35 percent of the federal funding, DOE for 33 percent, NSF for 20 percent, and NIH for 4 percent.
The ratio of basic to applied research in computer science was 1 to 2.5 in 1999 and this was down from 1 to 1.5 in 1990. Although federal funding of computer science research more than doubled during the 1990s, funding for basic research and university research increased much less, by 60 percent and 47 percent, respectively. The share of federal funding going to basic research fell from 40 percent in 1990 to 29 percent in 1999. In 1999, universities were receiving one-third of all federal funding for computer science, compared with half in 1990.
The trends toward more applied research and research done outside the universities are consistent with the federal cross-agency initiative on applied computing and network research that was in effect during most of the 1990s. It is also consistent with the fact that much of the DOE program is conducted in federal laboratories rather than universities. In addition to scientific opportunities and technological ripeness, the increase in federal funding of computer science research (and research in related fields) was driven by the high priority accorded it by successive administrations.
The effort began as a special interagency initiative in the FY 1992 budget. This initiative was called the High-Performance Computing and Communications (HPCC) Program. HPCC, which was initiated by the Federal Coordinating Council for Science, Engineering, and Technology under President George Bush, was later taken up by the National Science and Technology Council under President Clinton. In the FY 2002 budget, the effort is continued as the Networking and Information Technology R&D Program. Actual budget authority for the interagency initiative tracks the increase in spending on computer science research (Table 1).
TABLE 1 Funding of Interagency Networking and Computing Initiative, FY 1992–2002 (in millions of 1999 dollars)
Federal funding of research in electrical engineering increased from $780 million in 1990 to $984 million in 1993 (26 percent), then fell to $640 million in 1997 (35 percent), before increasing again to $699 million in 1999 (see Figure 7). The field had 10 percent less funding in 1999 than in 1990 (29 percent less than its high point in 1993).10
In 1990, most federal funding for electrical engineering (84 percent) came from DOD. Unlike in the case of computer science, DOD did not try to sustain or increase its level of support of electrical engineering. DOD sharply reduced
its support from 1993 to 1997. In 1999, the field had 31 percent less funding from DOD than in 1993 and 15 percent less than in 1990. Also unlike the computer science case, other agencies did not increase the amount of support enough to offset much of the DOD cut.11 The Department of Commerce increased its support by $26 million from 1990 to 1999. DOE and NASA each provided about $5 million more in funding in 1999 than in 1990, but the 1999 levels constituted cuts from 1993 levels. NSF cut funding by $17 million from 1990 to 1999 (22 percent).
Within a shrinking budget, federal agencies increased their emphasis on basic research and research conducted at universities on electrical engineering, but most of the funding still went to applied research and research conducted in non-university settings. The ratio of basic to applied research increased from 1 to 3.4 in 1990 to 1 to 2.7 in 1999. Although federal obligations for electrical engineering declined by 10 percent from 1990 to 1999, funding for basic and university research increased, by 4 percent and 14 percent, respectively, indicating that the cuts were made in applied research and performers other than academic institutions.12 Basic research rose from 23 percent of total federal funding
in 1990 to 27 percent in 1999. Universities accounted for 22 percent in 1990, compared with 28 percent in 1999.
NATIONAL TRENDS IN SUPPORT OF R&D RELATED TO LIFE SCIENCES AND IT
The impact of federal funding of research is affected by trends in funding by other institutions. These include industry, universities, and other organizations such as nonprofit research institutes, state and local governments, and foreign research institutions. The federal government was the largest source of R&D funds from World War II until 1980 and continued to be the largest source of research funds until 1995, in real terms (see Figure 8).
Industry provides 51 percent of total research funding and the federal government provides 34 percent, but the roles are reversed for basic research. The federal government provides the largest portion of basic research support, 49 percent, and industry provides 34 percent (see Figure 9).
The federal government may not need to continue funding research in areas where industry is investing its own funds. However, there may be areas that should receive more emphasis in federal research priorities despite increased investment by industry. These areas require attention because they are still too risky for industry to fund or are at a stage of research where companies cannot be
sure of recouping their investment. Industry typically spends most of its research funding on development and short-term, applied research rather than on the long-term, fundamental research needed to generate new technologies in the future. This section looks at the relative roles of the federal, industry, academic, and nonprofit sectors in funding health and IT research.
We will approach this issue somewhat indirectly. There is no consistent database of information across sectors on the field of research that is being funded, such as biotechnology or information technology, or on the stage of research (basic or applied). The analysis in Part II used the annual NSF survey of federal funds for R&D. This survey collects information from federal agencies on the fields of research they fund and on whether the research is basic or applied. Another NSF survey—the annual survey of academic R&D expenditures—collects information from colleges and universities on how much federally and non-federally funded R&D they perform. The academic survey uses a set of fields very similar to the classification of fields used in the federal funds survey, but the inclusion of development with research makes the results non-comparable with the data collected in the federal funds survey. A third NSF survey—the annual survey of industrial R&D expenditures—classifies research of each company by a single industrial classification (SIC) code, such as “drugs and medicines” (SIC-283) and “office, computing, and accounting machines” (SIC-357), rather than by field of research, such as chemistry, computer science, and mathematics. The survey greatly expanded coverage of the service sector a few years ago, so data on new categories of R&D such as software R&D only go back a few years.
In the most recent edition of Science and Engineering Indicators, the National Science Board (NSB) and NSF explored the possibility of conducting “cross-sector field-of-science classification analysis”.13 Here, we will look at an updated and slightly modified version of the NSB analysis of R&D in life sciences and information technology, areas chosen by the NSB because they can be “associated with academic fields of study and with industrial end-projects that tend to be associated with those fields.” The NSB graphs and text covered the period from 1985 through 1997.14 Here, the graphs have been updated to 1998 with survey data released since NSB issued the 2000 Indicators, and they have been extended back to 1981, to give additional historical context.15
Although company-funded R&D in the life sciences, which includes pharmaceuticals and biotechnology, has increased tremendously, federal funding for R&D in the life sciences is still larger (see Figure 10). Federal funding of life sciences R&D was $13.8 billion in 1998, compared with industry funding of $12.8 billion. Other sources of funding for university research in life sciences and bioengineering also increased, although at a much lower rate.16 Industry and U.S. Department of Agriculture (USDA) funding related to food and agricultural biotech R&D was fairly flat.
In aggregate, these sources of R&D increased from $15.0 billion in 1981 to $36.7 billion in 1998, a real increase of 145 percent.17 Federal funding has constituted a large but declining share of the total funding—55 percent in 1981 and 43 percent in 1998. Federal investment increased steadily, so this shift in shares resulted from the steep increase in R&D spending by pharmaceutical and biotechnology companies throughout the period.
According to NSF’s survey of federal funds, half the federal funding was for basic research in 1998 ($8.0 billion of $15.8 billion). Assuming that the biopharmaceutical industry devotes the same percent of its R&D funding to basic research that drugs and medicine companies reported to NSF in 1998–17.4
percent—and that one-third of nonfederal funding of university R&D is for basic research, approximately $12.5 billion, or 34 percent of the total funding of life sciences R&D, went to basic research in 1998.18
Information Technology R&D
Two sources of funding account for most R&D in information technology. Company-funded R&D in electrical equipment varied between $10 billion and $14 billion in the 1980s, then grew steeply in the 1990s to nearly $25 billion in 1998 (in constant 1999 dollars; see Figure 11). Company-funded R&D in office, computing, and accounting machines increased from $6 billion to $28 billion in the 1980s, then declined sharply to less than $5 billion in the mid-1990s. It jumped back up to $13 billion in 1997 before falling to $9 billion in 1998.19
Federal funds for research in mathematics and computer science increased steadily over the period, from $470 million in 1981 to $1.9 billion in 1998, but federal funding of electrical engineering research, which was between $900 million and $1 billion a year in the 1980s, declined after 1993 to $650 million in 1998. Nonfederal academic R&D funding increased steadily from $80 million in 1981 to $350 million in 1998.
According to the NSF industry survey, which expanded its coverage of the service sector in 1995, R&D on software and data processing services is also large and fast-growing. It rose from $9.1 billion in 1995 to $14.5 billion in 1998.
Federal R&D funding is much smaller for IT than for life sciences (although DOD-funded development of military equipment is not included here). As approximated here, IT-related R&D totals about $40 billion, not including software R&D. Federal funding accounts for just 6 percent of the total, compared with 43 percent of life sciences R&D funding.
The matter of funding long-term, fundamental research is quite different in IT than in the life sciences. The $2.5 billion in federal funding of IT R&D includes a total of $933 million for basic research. If we assume that IT firms
invest 5 percent of their R&D funding in basic research, and that half of academic R&D funding is for basic research, then national funding of basic research was $3.0 billion in 1998. That was 8 percent of the total spent nationally on R&D in IT (7 percent if software R&D is included).20
RECENT TRENDS IN FEDERAL FUNDING OF THE BASIC RESEARCH BASE
Merrill, McGeary, and Henderson recently updated a 1999 study by McGeary and Merrill on trends in federal funding of fields of science and engineering.21 The first study analyzed data on actual federal research expenditures from FY 1990 through FY 1997, especially trends after 1993, when pressures from the federal budget deficit and reductions in the defense budget had stopped real growth in federal research budgets.22 The update extends the analysis to FY 1999, the latest year for which there are data on actual expenditures.
The 1999 study tested the hypothesis that certain fields of science and engineering would do less well in budgetary terms because the agencies that provide most of their federal research funding were reducing their level of investment in research. Indeed, in 1997, although the level of federal research spending was nearly the same as it had been in 1993, a number of agencies were spending less on research than they had in 1993, including DOD (–27.5 percent), Department of the Interior (–13.3 percent), Department of Agriculture (–6.2 percent), and DOE (–5.2 percent).23 Meanwhile, NIH’s research budget was up by 11 percent. In fact, federal research spending, not counting the NIH, was down by 6 percent rather than remaining flat.
The study found, however, that the fate of individual fields did not necessarily mirror that of their chief benefactor. For example, computer science, which received the majority of its support from DOD before 1993, continued to grow strongly despite the downturn in overall DOD research funding. Conversely, bio-
logical sciences research, which receives most of its funding from NIH, had barely increased its funding despite the rapid increase of the overall NIH budget. Several factors might account for that discrepancy: (1) agencies with reduced research budgets did not necessarily cut all fields, or did not cut them the same; (2) other agencies stepped up their support; or (3) some combination of these factors.
Nevertheless, most fields in the physical sciences and engineering were disproportionately negatively affected because they received most of their support from DOD and DOE, and only a few have been able to increase support from other agencies. The fields with less federal funding included chemical, civil, electrical, and mechanical engineering; mathematics; physics and chemistry; and geology. Four—electrical and mechanical engineering, physics, and geology—had at least 20 percent less funding than 4 years before.
The update found that the funding situation had eased somewhat by 1999. Federal obligations for research were 12-percent greater in 1999 than in 1993. Moreover, all agencies except DOD and the Department of the Interior now had larger research budgets, although NIH accounted for most of the net growth in research funding since 1993.24 Fewer fields had less funding than in 1993–7 of 21 compared with 11 of 21 in 1997—but 5 of them (including electrical engineering) had at least 20 percent less funding compared with 4 in 1997. Meanwhile, the number of fields with at least 20 percent more funding than in 1993 went from 1 in 1997 to 5 in 1999, including biological sciences, medical sciences, and computer science (see Figure 12).
Despite increased funding, the new pattern of allocation developed during the budget cuts of the 1993–1996 period remained. In some respects, the gap that had opened between certain fields at the top (especially biomedical and computer science) and those at the bottom (electrical and mechanical engineering, physics and chemistry, and geology) was widening.
The “Balance” Issue
The divergent trends in federal funding of fields of research in the 1990s, reinforced by the effort to double the NIH budget in 5 years beginning in FY 1999, have raised concerns among scientists and engineers, industrial leaders, and science and technology policy analysts about a possible imbalance in the federal research portfolio. Such concerns extend even to the long-term health of biomedical research itself.
Harold Varmus, formerly director of the NIH, stated his conviction that “NIH can only wage an effective war on disease only if we—as a nation and a scientific community, not just as a single agency—harness the energies of many
disciplines, not just biology and medicine. These allied disciplines range from mathematics, engineering, and computer sciences to sociology, anthropology, and behavioral sciences.”25
The Ad Hoc Group for Medical Research Funding, which strongly supports doubling the NIH budget, has also supported more balanced funding of the range of research fields. They have noted that
“…a strong federal research enterprise requires a balanced portfolio across the scientific and engineering disciplines. Many breakthroughs in medical research and treatment, such as magnetic resonance imaging (MRI), have come from advances in the physical sciences that were developed from basic research in physics, chemistry, and mathematics. Continued progress in medical research depends on continued advances in other areas of science and engineering.”26
The House Committee on Science, in its annual report of views and estimates, has also expressed concern about the growth of NIH relative to other disciplines also important to scientific progress, including in biomedical fields.
“The Committee looks forward to working with the administration and our Congressional colleagues to try to develop ways to determine whether the current portfolio is too heavily weighted toward NIH, and, if it is, to figure out what a balanced portfolio would be.”27
Finally, in a recent editorial, Donald Kennedy, editor of Science, asked, “Does it really make sense for some pieces of the enterprise to be treated very well indeed and others to be held back or cut? There are good reasons for thinking it doesn’t.”
“In the first place, an increasing proportion of the important problems in science are interdisciplinary in character. At Science, we have published contributions to nanotechnology that come from disciplines as diverse as chemistry, materials science, and electrical engineering. The climate sciences, on which we will depend in formulating international policies, draw from paleontology, oceanography, and atmospheric chemistry. The dramatic scientific gains that will flow from the sequencing of the human genome will be harvested not only by molecular biologists but also by specialists in bioinformatics, trained in such disciplines as mathematics and computer science. Nurturing fields such as these requires a balanced portfolio.”28
The current administration is less concerned about the balance issue, arguing that the research enterprise is expanding and healthy because of the increase in industrial R&D funding. At his confirmation hearing in January 2001, Mitch Daniels, Director of the Office of Management and Budget, acknowledged that the federal government has an important role to play in funding research that the private sector finds too risky or too long-term to invest in and said his office would “work to develop an appropriately balanced program of federal research.” At the same hearing, Daniels’ deputy director, Sean O’Keefe, spoke of the need for a balanced portfolio of research funded by both the federal government and the private sector and said that, from that perspective, the current portfolio is balanced.29
Trends in Federal Funding of Basic Research
Trends in basic—that is, long-term, fundamental research funding are important because industry tends to focus its funding on applied research and development. In the IT area, for example, Irving Wladawsky-Berger, General Manager of IBM’s Internet Division, told the President’s Information Technology Advisory Committee (PITAC) that government must take the lead in funding long-term research because industry must focus on being successful in the marketplace by developing competitive products and services and is facing increasingly short development cycles. “While the IT industry invests significantly in R&D, the bulk of the investment is product development (90 percent), and the bulk of the remaining is short-term, applied research, with only a few larger companies doing any long-term, basic research.” Meanwhile, “R&D has declined as a percentage of revenues due to competitive pressures on prices and profit margins, putting further pressure on long-term research.”30
The trends in basic research are similar to those in total research described above (see Figure 12), although federal funding of basic research was not cut as much and recovered faster (see Figure 13). In 1997, when federal obligations for basic research were nearly 3 percent more than in 1993, 6 of the 12 fields in engineering, physical sciences, and math/computer science had less funding for basic research than in 1993. And 2 of the 12 (electrical and mechanical engineering) had more than 20 percent less. In 1999, when federal basic research funding was 17 percent more than in 1993, 6 of the 12 fields still had less than in 1993. And 2 of the 12 (chemical and mechanical engineering) had more than 20 percent more. In the life sciences, 1 of the 4 fields (medical sciences) had more funding for basic research in 1997 than in 1993; in 1999, all 4 had more funding than in 1999.
Federal support of the life sciences fields increased substantially during the 1990s. Annual funding for basic research in the biological sciences went from $3.3 billion in 1990 to $4.2 billion in 1999, a real increase of 27 percent (see Figure 14). Funding for medical sciences increased even more in the 1990s (76 percent) and was close in amount to that for the biological sciences by 1999 ($4.0 billion). The other two life sciences fields—environmental biology and agricultural science—were cut for several years in the mid-1990s and were about even with their 1990 funding levels in 1998. Each received a substantial increase in 1999, which put them ahead of 1990 by 37 percent and 13 percent, respectively.
Because 1999 was the first year of the 5-year campaign to double the NIH budget, the rate of growth for basic research in the biological and medical sciences can be expected to increase substantially from the 2.7 percent and 6.5 percent annual compound rates of growth they experienced from 1990 to 1999, respectively. From 1998 to 1999, for example, funding increased by 14.4 percent for biological sciences, and by 15.0 percent for medical sciences.
Trends in federal obligations for basic research in engineering varied by field (see Figure 15). Metallurgy/materials engineering experienced substantial growth, especially from 1993 to 1996. From 1990 to 1999, funding for the field increased from $317 million to $481 million, or 52 percent.
Several other fields increased their funding for basic research for the first few years of the 1990s, then received less funding through 1999. Electrical engineering increased from $179 million in 1990 to $227 million in 1993, then declined to $186 million in 1999, 18 percent less than its high point in 1993. Chemical engineering increased from $92 million in 1990 to $120 million in 1992, dropped to $81 million in 1993, and steadily shrank to $55 million in 1999; it was down 40 percent for the decade (54 percent from its high point in 1992). Astronautical engineering rose from $76 million in 1990 to $108 million in 1992, then fell to between $60 million and $70 million from 1993 to 1999.
Funding levels for the other two fields were down for most of the decade but recovered in 1999. Aeronautical engineering had its two best years at the beginning and end. Funding for basic research in that field was $329 million in 1990 and $332 million in 1999, but it fluctuated between $275 million and $300 million during that period. Civil engineering was down from its high of $70 million in 1991 for most of the decade, jumping from $36 million in 1998 to $69 million in 1999.
The general trend for funding of basic research in the physical sciences was downward in the 1990s, at least after the first year or two (see Figure 16). Funding for physics peaked in 1991 at $1.9 billion, dropped to about $1.6 billion from 1994 to 1998, and was at $1.7 billion in 1999. Physics was down by 6 percent for the decade (by 12 percent from the 1991 high point). Similarly, chemistry was funded at $612 million in 1990, $636 million in 1992, and $555 million in 1999.
Funding for astronomy also peaked early (at $834 million in 1992) but ended the decade about where it started (at $717 million compared with $707 million in 1990).
The general trend for computer science was up, except in 1994 (see Figure 17). Basic research funding was $274 million in 1990 and $483 million in 1999, an increase of 60 percent. Applied research funding was up even more—by 163 percent.
The picture for mathematics was more mixed. Funding for basic research generally increased from 1990 to 1994, then dropped for several years before recovering in 1997. The funding level was $215 million in 1990 and $271 million in 1994. In 1995 and 1996, it dropped about $100 million, then jumped to about $250 million in 1997, 1998, and 1999. Thus, funding for basic research
in mathematics increased by 14 percent from 1990 to 1999, but the 1999 level was 6 percent lower than it was in 1994.
Federal and nonfederal funding of research related to the core fields of health (biological sciences and medical sciences) and IT (computer science and electrical engineering) increased substantially in the 1990s. It must be kept in mind, though, that federal funding plays a much larger role in the health sector. This contributes to a much greater emphasis on basic research in the health than in the IT sector.
The growth in federal funding stems from strong congressional interest in biomedical research and the high priority that successive administrations have given to a cross-agency initiative in high-performance computing and networking dating from 1992. This growth will continue and even accelerate because of the campaign, begun in 1999, to double the NIH budget in 5 years. A similar acceleration of growth may occur in IT in response to 1998 and 1999 reports of the PITAC, which recommended a near tripling of federal investment in IT R&D by 2004.31
See, for example, PITAC (1999). President Bush just extended PITAC until June 2003. See Executive Order of May 31, 2001, at http://www.whitehouse.gov/news/releases/2001/06/200106016.html.
Federal funding of research in a number of other fields—important to long-term progress in the health and IT sectors—has either not increased as much or, in the case of most fields in the physical sciences and engineering, has suffered cuts during the 1990s. This situation has given rise to a growing chorus of concerns about a possible “imbalance” in the federal portfolio of research.
Funding Trends in the Core Fields of Health and IT R&D
Federal funding has been moving into the hot fields central to health and IT R&D. For example, funding of biological sciences and medical sciences increased 28 percent and 65 percent, respectively, in real terms, from 1990 to 1999. Federal funding of computer science research increased 121 percent during the same period. Electrical engineering, however, experienced 10 percent less federal support.
Nonfederal funding patterns differ in health and IT R&D. It is impossible to know how nonfederal funding is distributed by field of research (e.g., computer science or biology), because industry R&D data are not collected by field. In this paper, we have attempted rough estimates of national R&D by including R&D funded by health and IT industries and nonfederal funds spent by academic institutions on R&D in the relevant fields, along with federal R&D funding. In health, for example, R&D funding in two industries—drugs and medicines and food—is counted, along with nonfederal expenditures on life sciences and bioengineering R&D by academic institutions. The industries representing IT are electrical equipment; office, computing, and accounting machines; and, beginning in 1995, computer and data processing services, which are included along with academic nonfederal spending on mathematics, computer science, and electrical engineering.
Nonfederal (primarily industry) funding of R&D related to health has apparently increased at a faster rate than federal funding. Federal health R&D, which is almost all research, accounted for 43 percent of the national investment in 1998 despite the strong buildup in industry funding. In IT, nonfederal funding of R&D appears to have increased at a rate slightly less than that for federal funding of computer science and electrical engineering, but federal funding still only constitutes a small part of IT R&D—less than 7 percent (5 percent if the software industry is included).32
Although R&D is up overall in both cases, the picture is more mixed in IT than in health R&D. The strong increase in federally funded research in computer science has been offset to some extent by the decline in funding for electrical engineering, down by 29 percent since 1993 (basic research by 18 percent).
Company funding of computer R&D is also down, although funding of computer peripherals, networking products, and software is probably up (these are not measured directly except, since 1995, software R&D).
The greater extent of federal support of R&D in health than in IT helps account for the larger share of basic research in health R&D. Drugs and medicines firms also devote a larger share of their R&D funding to basic research than electrical equipment and computer firms do—17 percent vs. less than 5 percent on average. As a result, basic research constitutes roughly one third of national R&D related to health, compared with less than one tenth of national R&D related to IT.
Trends in Federal Funding of Other Fields Contributing to Progress in Health and IT
Most other fields of research, including those that have been or promise to be important in health and IT in the long run, have received a shrinking share of federal investment during most of the 1990s. In some cases, they are receiving less in absolute terms than earlier in the decade. Fields increasing at a rate less than the biological, medical, and computer sciences since 1993 include environmental biology, agriculture, materials engineering, mathematics, atmospheric science, psychology, and the social sciences. Fields whose level of funding has fallen absolutely since 1993 include chemical engineering, electrical engineering, chemistry, physics, and geology (all but chemistry by 25 percent or more).
Most of the shift in the composition of federal funding of research occurred during the period from 1993, when federal funding stopped growing, and 1997, when growth resumed. The increases in biomedical and computer research were deliberate efforts on the part of Congress and the Administration, respectively. The cuts, however, were not the result of a high-level plan or policy. They were the byproduct of priority setting by the departments in light of their individual missions. Many of the fields with reduced funding (physics, chemical engineering, and electrical engineering) were primarily supported by DOD and DOE in the early 1990s, agencies that cut their research budgets substantially in the 1993–1996 period (by 22 and 8 percent, respectively).33
The shift in federal funding among fields was not a planned one in response to a formal assessment of national needs or research opportunities. It was a “resultant,” that is, it is what happened after each federal agency and its committees in Congress set priorities without regard to national-level considerations. But what levels are optimal for each agency, taken one by one, may not be
optimal overall, especially as the rationale for research in the physical sciences and engineering shifts from Cold War exigencies to economic growth and competitiveness. Industry’s need for a flow of research results in the physical sciences and engineering, and for physical scientists and engineers trained in academic institutions conducting the research, may exceed the future needs DOD has for research and talent in those areas, but there is no mechanism for making the adjustment.
Balance in the Federal Portfolio
The federal budget process should provide adequate and steady support for a diverse portfolio of research activities that reflects scientific opportunities and national needs and also provides for a healthy infrastructure of facilities and talented, well-trained personnel. At present, however, there is no well-accepted methodology for setting priorities across fields of science and engineering. It is difficult to achieve agreement within a field on the appropriate balance among funding current research projects, a state-of-the-art infrastructure of research facilities and equipment, and the training of new researchers for the future. The absence of clear guidance in the form of policy priorities, or of a process for analyzing the overall federal research budget, makes it difficult to balance funding among complementary fields of research. In such circumstances, one person’s idea of research portfolio imbalance can well be another’s notion of what constitutes a healthy investment in a “hot” area of research—though both may honestly seek to meet major national goals through their respective proposals.
In the case at hand, some core fields in health and IT research are expanding rapidly and are accounting for most of the increase in federal funding in recent years (three fields—the biological, medical, and computer sciences—received more than two-thirds of the net increase in federal funding of research from 1990 to 1999). And that trend has no doubt continued since 1999, the last year for which there are data on actual funding by field, and will continue at least through 2003, the final year of the 5-year effort to double the NIH budget. But is this disparity in rates of growth among fields’ funding leading to an imbalance among fields?
There are several approaches to answering this question. One approach is to treat research funding as an investment portfolio in which uncertainty and risk are minimized through diversification.34 This approach is based on the uncertainty inherent in research.35 It is not possible to predict where and when break-
throughs will occur, and the payoffs may come years later, perhaps in another field. Who knew in 1945, when nuclear magnetic resonance was discovered in condensed matter, that that basic research finding would lead to the development of magnetic resonance imaging (MRI) technology for improved health care and research 30 years later?36 This approach would ask if all fields were being adequately funded, including those that are not necessarily identified with popular goals of research, because advances in any field may lead to a breakthrough in another. The increase in interdisciplinary research adds to the prudence of investing in a broad range of fields.
A second approach is to see where the United States stands in comparison with research in other countries in each field and make adjustments to ensure not only that the nation is preeminent in certain fields and among the world leaders in the other fields.37 International benchmarking avoids the problem of setting priorities across fields. This approach would ask if the reduced funding in fields such as physics, chemistry, geology, and some fields of engineering is putting the United States way behind the performance of other countries in those fields.
A third approach is to examine carefully the degree to which progress in high-priority fields might depend on advances in other fields. In this case, because of the sequencing of the human and other genomes, advances in computing are clearly going to contribute to health research and enable new applications; at the same time, computing advances may well be based on biological systems (see workshop proceedings in this volume). But advances in both areas—IT and health—have and almost certainly will continue to rely on contributions from other fields, including chemistry, physics, mathematics, various engineering fields, and the social and behavioral sciences. In this approach, for example, the federal research portfolio is out of balance to the extent that progress in health research depends on fields other than the biological and medical sciences that are not receiving funding commensurate with their role.
This situation came about because decision-making on health research historically has assumed that funding for the other important disciplines would be provided by other agencies (for example, chemistry, physics, and chemical engineering). To a large extent it was, until the end of the Cold War, reinforced by the federal budget deficit, caused DOD and DOE to reduce their research budgets.
The trend toward interdisciplinary work makes it more prudent to invest in a broad portfolio of research and not concentrate funding in a few fields. The editorial in Science by Donald Kennedy quoted earlier takes this approach in noting that progress in a number of hot fields—for example, nanotechnology,
global climate change research, genomics, and energy supply—depend on advances in a number of other fields. He concludes that capturing interdisciplinary complementarities requires a balanced portfolio.
A fourth and perhaps simplest way to approach the issue is not to ask if biomedical research or any other field is receiving too large a share of federal research funding. Instead, ask if there are fields that are underfunded with respect to scientific opportunities, including their contributions to other fields, especially fields central to the achievement of national goals such as better health, economic competitiveness, and environmental protection. International comparisons are useful here. This approach is also consistent with the reality that balance in the federal research portfolio cannot easily be achieved by reallocating funding from one agency to another. The R&D budget, including funding for research, is never considered as an integrated whole, especially in Congress. Each R&D program competes with other programs in its agency and the other agencies under the jurisdiction of its appropriations subcommittee, not with other R&D budgets in other appropriations subcommittees. Thus the argument shifts from saying, say, NIH gets too much to saying, say, physics does not get enough, in part because of its role in improving health among many other worthy applications such as computing and other information technologies.
Each of the approaches to assessing the federal research portfolio outlined above would examine the magnitude of the shift among fields that took place in the mid-1990s, focus on fields that stopped growing or began shrinking, examine their importance to the nation’s leadership in science and technology, and ask if the decentralized decision-making that led to no or negative funding growth makes sense from an overall point of view. The answer may well be yes, given changes in scientific opportunity or national goals or both, but currently there is no mechanism for addressing the question in the first place. In the early 1970s, when DOD, NASA, and the Atomic Energy Commission cut research funding and federal support of the physical sciences and engineering dropped sharply, NSF received a 20 percent increase, half designated for support of important research being dropped by the mission agencies.38 In the late 1970s, the Office of Management and Budget held funding in reserve for R&D program expansions and additions deemed advisable from the President’s point of view after review of the agency requests (Smith, 2001).39 There has been no similar plan to adjust the impact of shifts in agency priorities. Proponents of diminishing fields must make their best case to agencies and congressional committees with juris-
diction over their programs except perhaps to convince the House and Senate budget committees to give larger allocations to the relevant appropriations subcommittees.
Ad Hoc Group for Medical Research Funding. 2001. “Ad Hoc Group for Medical Research Funding—FY 2002 Proposal,” January. At: http://www.aamc.org/research/adhocgp/fy2002.doc.
AIP (American Institute of Physics). 2001. “The Office of Management and Budget on Science and Technology.” FYI, The American Institute of Physics Bulletin of Science Policy News. No. 28(March 9).
Cockburn, I, R.Henderson, L.Orsenigo, and G.P.Pisano. 1999. “Pharmaceuticals and Biotechnology,” pp. 363–398 in National Research Council, U.S. Industry in 2000: Studies in Competitive Performance. Washington, D.C.: National Academy Press.
Gomery, R.E. 1995. “An Unpredictability Principle for Basic Research,” pp. 5–17 in A.H.Teich, S.D.Nelson, and C.McEnaney, eds., AAAS Science and Technology Policy Yearbook: 1995. Washington, D.C.: American Association for the Advancement of Science.
House Committee on Science. 2001. “Views and Estimates of the Committee on Science for Fiscal Year 2002.” March 16.
Kennedy, D. 2001. “A Budget Out of Balance.” Science. 291(March 23): 2337.
Lomask, M. 1976. A Minor Miracle: An Informal History of the National Science Foundation. NSF 76–18. Washington, D.C.: National Science Foundation.
Merrill, S.A., and M.McGeary. 1999. “Who’s Balancing the Federal Research Portfolio and How?” Science. 285(September 10): 1679–1680.
McGeary, M., and S.A.Merrill. 1999. “Recent Trends in Federal Spending on Scientific and Engineering Research: Impacts on Research Fields and Graduate Training,” Appendix A in National Research Council, Securing America’s Industrial Strength. Washington, D.C.: National Academy Press.
McGeary, M., and P.M.Smith. 1996. “The R&D Portfolio: A Concept for Allocating Science and Technology Funds.” Science. 274(29 November): 1484–1485.
National Academy of Sciences. 2001. A Life-Saving Window on the Mind and Body: The Development of Magnetic Resonance Imaging. Washington, D.C.: National Academy of Sciences, March 2001. At: www/beyonddiscovery.org/beyond/BeyondDiscovery.nsf/files/PDFMRI.pdf/$file/MRIPDF.pdf.
National Research Council. 1993. Science, Technology, and the Federal Government: National Goals for a New Era. Committee on Science, Engineering, and Public Policy. Washington, D.C.: National Academy Press
National Research Council. 1999. Funding a Revolution: Government Support for Computing Research. Computer Science and Telecommunications Board. Washington, D.C.: National Academy Press
National Research Council. 1999. Securing America’s Industrial Strength. Board on Science, Technology, and Economic Policy. Washington, D.C.: National Academy Press
National Research Council. 2001. Trends in Federal Support of Research and Graduate Education. Washington, D.C.: National Academy Press
National Science Board. 2000. Science and Engineering Indicators—2000. Arlington, VA: National Science Foundation (NSB 00–01).
National Science Foundation. 2000. Research and Development in Industry: 1998. Arlington, VA: National Science Foundation (NSF 01–305).
Office of Management and Budget. 2001. Historical Tables, Budget of the United States Government, Fiscal Year 2002. Washington, D.C.: U.S. Government Printing Office.
President’s Information Technology Advisory Committee (PITAC). 1999. Information Technology Research: Investing in Our Future. Report to the President. Arlington, VA: National Coordinating Office for Computing, Information, and Communications.
Smith, P.M., 2001. “Science and Technology in the Carter Presidency.” Paper presented at the Office of Science and Technology Policy 25th Anniversary Symposium, Massachusetts Institute of Technology, Cambridge MA, May 1. Forthcoming in symposium volume. Cambridge, MA: MIT Press.
Varmus, H. 1999. “The Impact of Physics on Biology and Medicine.” Plenary Talk, Centennial Meeting of the American Physical Society, Atlanta, March 22. At: www.mskcc.org/medicalprofessionals/presidentspages/speeches/theimpactofphysicsonbiologyandmedicine.html.
Wladawsky-Berger, I. 1999. “Information Technology: Transforming Our Society,” presentation at the seventh meeting of the President’s Information Technology Advisory Committee, February 17. See “Transformations” PDF at: www.ccic.gov/ac/agenda-17feb99.html.