During the past 40 years, the nature of funding for university research has changed significantly. At the same time, the nature of university research and attitudes within the research community have evolved. The coupled changes have occurred gradually, and for this reason, many of them have gone unnoticed. Those changes that are noticed are rarely subjected to critical assessment because of the symbiotic relationship between funding agencies and universities. The result of the gradual, unchecked changes in funding agency practices and in the research community is that the funding system for academic science and engineering is significantly flawed.
Research in the US has remained robust, at least as judged by the number of papers submitted to Physical Review and Physical Review Letters 1 and the number of foreign students coming to the US for advanced study. There is, however, no guarantee that the strength of the US contribution to science will continue. Over the past 200 years, scientific leadership has passed from France, to England, to Germany, and finally to the US. If the US is to maintain this leadership, its scientists must recognize and make the most of the factors that contribute to the strength of US science. Scientific quality depends, in part, on the manner in which science and engineering are supported; hence the issue of scientific support deserves close scrutiny.
The following comments are based on observations made during a 50-year research career in the physical sciences and engineering that includes a nearly 13-year period in which I was director of the Frederick Seitz Materials Research Laboratory. They are one person’s views, intended to stimulate discussion of issues that affect university researchers, an extremely important part of the university community. While many of the detailed comments are drawn from observations of the condensed matter physics, materials science, and engineering communities, I believe they have broader relevance.
University research
As shown in figure 1, academic research support in the US increased from $993 million in 1962 to $26.3 billion in 1998. The same figure shows the growth in academic research support in constant 1992 dollars. Over the period 1962–98, the percentage of academic research funds devoted to basic research fell somewhat, from 74.2% to 68.7%. (Statistical data are derived from reference 2.) In the same period, university research changed from an endeavor carried out in a small number of universities to a major component of education at almost all universities and at many colleges. The sophistication of research has increased, particularly if it is measured by the increased complexity and corresponding increased cost of research instruments. Large instrumentation facilities—for example synchrotron, neutron-scattering, and high-field magnet facilities—have generally been established in the past 40 years. Often these facilities are operated by Department of Energy (DOE) labs rather than by universities. The cost of providing instrumentation and the operating costs involved in using the facilities, though, are a part of the academic research budget.
Figure 1. Support for US academic research over the period 1960–98 in dollars unadjusted for inflation (red) and in constant 1992 dollars (blue). The curves reflect contributions from federal, state, and local governments; academic institutions; industry; and other sources. Data are from ref. 2, chap. 6.
Figure 1. Support for US academic research over the period 1960–98 in dollars unadjusted for inflation (red) and in constant 1992 dollars (blue). The curves reflect contributions from federal, state, and local governments; academic institutions; industry; and other sources. Data are from ref. 2, chap. 6.
The cost of university research programs has increased faster than the cost of living or the (usually) more rapidly rising gross domestic product price deflators used by NSF to adjust for inflation. The increases have resulted from many factors. Chief among these are the higher cost of sophisticated instrumentation as exemplified by the transmission electron microscope shown in figure 2(a); increased operating expenses, such as those associated with the use of large facilities; and increased personnel costs. The first two factors are difficult to quantify, though figure 2(b) does indicate the dramatic rise in the cost of a TEM over the past 30 years. Data on personnel costs are more readily available. Information contained in research proposals to the federal government shows that during the period 1972–99, the salaries of postdocs and graduate students at the University of Illinois increased by 5.4% and 81%, respectively, in constant dollars. Nationwide, postdoc and graduate student salaries have increased at comparable rates.
Figure 2. The transmission electron microscope (TEM) is significantly more sophisticated and expensive than it was 30 years ago. (a) This modern TEM incorporates many imaging, diffraction, and analytic techniques, and requires two scientists to operate it. (b) The costs of a TEM over the period 1970–2000 in constant 1992 dollars. The increasing costs reflect the increasing sophistication of the instrument.
Figure 2. The transmission electron microscope (TEM) is significantly more sophisticated and expensive than it was 30 years ago. (a) This modern TEM incorporates many imaging, diffraction, and analytic techniques, and requires two scientists to operate it. (b) The costs of a TEM over the period 1970–2000 in constant 1992 dollars. The increasing costs reflect the increasing sophistication of the instrument.
Academic research awards have not kept up with the inflated costs of carrying out research over the past 30–40 years. As an example, NSF division of materials research single-investigator grants have actually declined in constant 1972 dollars from an average of $38 010 in 1972 to $30 090 in 1999. 2,3 Overall, academic research awards have declined in constant dollars since 1962. NSF individual grants now typically support only one or two graduate students. Even so, the salary increases for postdocs and graduate students have forced universities to pay for operating costs that had previously been covered by research grants.
Nowadays, significant equipment purchases are based on competitive grants that require universities to foot part of the bill for the equipment, provide high-quality space, and supply some operational costs. These obligations are not part of the negotiated indirect cost reimbursements for building and ground maintenance, libraries, administration, and so forth. The cost sharing required by grants “certainly favors the well-endowed over public universities and is divorced from scientific quality,” according to a senior research colleague who also holds a half-time administrative position.
University administrators often claim, and it may well be, that universities lose money on federally supported research, particularly with federal agency grants (for example, from the National Institutes of Health or the US Department of Agriculture) that cap their overhead at low values. Even so, universities support the competition of their faculty for research grants: They argue that such support is necessary to attract the highest quality faculty. Not all universities, though, are equally capable of maintaining a first-rate research program. While recognizing that high-quality research is of great value to society and improves the quality of education, the university research communities should realistically assess the role of research at their institutions. The results of such an assessment might lead to a research concentration in fields in which the university is especially strong, or to a rededication to teaching.
Large research groups
Over the past 40 years, faculty research groups in the major research universities have increased in size. As a result, faculty must compete for and obtain significantly larger amounts of funding to support their research. I’ve already mentioned that the size of individual research grants has not kept pace with the rise in research costs. Furthermore, in many cases, support is granted for a limited time that is not adequate for completion of the proposed research. The limited support may reflect the belief that other funds will be used to supplement the amount given by the agency. Such an assumption places great stress on the research group, since the supplementary funds are often not forthcoming.
Because the NSF mathematical and physical sciences directorate funds only a fraction (about 35% in fiscal year 2000) 3 of the proposals it receives, faculty must spend a good deal of time preparing proposals. Young faculty members are becoming research managers, working on proposals, reports, and so forth, rather than doing research and following their best ideas. Since the early years are often among the most productive, this is a very unfortunate development.
Reinforcing the researcher-turned-manager phenomenon is the escalation in the number of publications considered to be acceptable for promotion or tenure. In the 1960s, 10 high-quality publications in refereed journals were adequate for receiving tenure at the University of Illinois. At present, this number would be unacceptable. The recognized need for large numbers of publications is a driving force behind faculty desire for large research groups—larger groups mean more students writing papers on which faculty members are coauthors. While this trend is generally decried by senior faculty and administrators, woe be unto faculty members who submit a relatively small number of high-quality publications at the time of their tenure consideration. They’d better hope for a much more enlightened faculty committee than is usually the case.
Funding agencies
With the exception of NSF, funding agencies have particular missions and, to varying degrees, research funding has been tied to these missions. Over time, there has been a varying mix of support for basic and applied science and engineering. The balance between basic and applied funding has reflected the views of the agencies’ leadership and, to some extent, the personnel who award and monitor the contracts. The variety of sources of research funds and the corresponding diverse attitudes of those who control the funds is one of the strengths of US research funding. Research ideas may be considered from many perspectives, and the abundance of funding opportunities protects against the constraints on innovation that would occur if, as is proposed in Congress from time to time, a single department of science were to control funding for academic research.
The variety of funding sources also necessitates a range of interactions between agencies and researchers. Some agencies, such as the Defense Advanced Research Projects Agency (DARPA), emphasize support of large, multi-investigator programs. DOE commits much of its funds to its national laboratories and is also a strong supporter of the large facilities used by researchers at universities. It tends to emphasize large programs, but does have a program to support individual investigators. Research support by Department of Defense (DOD) agencies other than DARPA is focused on individuals. NSF supports both individuals and a range of larger integrated programs.
The large multi-investigator programs supported by DARPA are generally mission-directed applied research, while those of NSF are a mixture of basic and applied research efforts. The percentage of funds committed to basic research fell somewhat over the period 1962–98 and, although the data are not readily available, I strongly suspect that a significant shift of support away from individual investigator programs to multi-investigator efforts has occurred (see the article “Physics In a New Era” by Thomas Appelquist and Donald Shapero in PHYSICS TODAY, November 2001, page 34).
Almost all federal funding agencies claim to depend primarily on external reviews, but in practice, peer review varies, even within a single agency. DOD is certainly an agency with a mission, but it does have funds earmarked for basic research. I have the impression that, even when considering proposals for these funds, DOD personnel obtain external reviews but make their decisions primarily on their views of agency needs. There is thus a great deal of reliance on the perspective and impartiality of the contract monitors.
In nonmission agencies, the discretionary judgment of the agency personnel should, and generally does, play a much smaller role in the selection process. Nonetheless, agency personnel occasionally have a significant influence on the process through the reviewers they choose, the emphasis they place on reviews, and so on. In some cases known to me, judgments of agency personnel have been unwarranted and excessively intrusive.
Interdisciplinary interactions and outreach
NSF has developed a number of multi-investigator programs, including the Materials Research Science and Engineering Centers, the Engineering Research Centers (ERC), and the Science and Technology Centers (STC), designed to achieve a number of goals in addition to those focused on science and engineering. Grant applications to these programs are judged on the quality of proposed research but also on outreach efforts and on the added value provided by interdisciplinary interactions within the university or between the university and industry. Outreach efforts, which are greatly emphasized, include, for example, programs for K–12 students and educators. While I believe that NSF multi-investigator programs have enhanced interdisciplinary interactions and that the outreach goals are commendable, I have concerns about the emphasis on outreach as proposals are evaluated. I am sympathetic to an active, well-respected colleague of mine who has commented, “Social engineering has distorted the NSF to a ludicrous degree. You almost can’t get a research grant anymore without making up some story about how it will impact K–12. The NSF needs a reformation.” More-over, I note that there is little assessment of either the success of the programs’ outreach efforts or their effect on the research funded.
Both the amounts of funds granted and the amount of time over which proposals are supported are often inadequate for the research programs NSF desires to support. Grants are distributed among a large number of investigators who must look for additional support elsewhere. For some programs (ERC and STC are examples), NSF assumes that industry will pick up the slack when NSF funding runs out. Unfortunately, the assumed industrial aid has not been forthcoming. Industry appears willing to supplement NSF support, but is generally unwilling to pay the entire cost of research. Hence, even research programs that meet all of the goals set forth in their funding proposal are generally forced to disband once NSF support terminates.
NSF grants often contain sunset clauses that define the maximum term of the grant. Grants should not become entitlements, but it is clear to me that terminating a project after a fixed period of time regardless of the performance of the research group isn’t an effective way to spend research funds. Whether project funding continues should be based on peer review of the quality of a renewal proposal.
A nice feature of the multi-investigator programs is that they allow for the acquisition and maintenance of needed experimental facilities. These facilities, though, need to be maintained even after the grant that enabled their purchase runs out.
Multi-investigator program grants are so desired by universities that maintaining the quality of the review process has been difficult. Often so many proposals are submitted that qualified reviewers are difficult to obtain. Reviewers often submit grant proposals as well, and are aware that they and the peers they are reviewing are chasing limited funds. I have the impression that, nowadays, peer reviewers are less willing to report favorably on proposals that disagree with their particular point of view. A colleague noted with evident frustration, “I have more control over my competitors than I do over my own agenda. I can block their research in 10 minutes with a bad review of a proposal. It takes me one to two months to write a proposal of my own, but it can be shot down trivially, no matter what it says!”
In 1979, NSF established the Experimental Program to Stimulate Competitive Research (EPSCOR), which sets aside funds for use in states lesser able to compete for research grants. The goal of improving research in disadvantaged states is laudable, but it has not been shown that the goal is being met. Moreover, EPSCOR lends credence to the sentiment expressed by an active researcher that “NSF is scandalously terrified of recognizing excellence at the expense of egalitarianism.” It may be time to determine the effectiveness of EPSCOR. If effectiveness cannot be established, it would be difficult to justify program expenditures at a time when research funding is insufficient to support excellent research proposals.
Industrial interactions
Industrial research support at universities has increased from $79 million in 1972 to $1896 million in 1998 (in constant 1972 dollars), at which time it corresponded to 7.2% of academic funding. 2 Industry supports academic research motivated by the possibility of obtaining intellectual property that could be commercially viable. Most universities seek and welcome this new source of research funding, but industrial funding comes with its own set of problems.
Determining intellectual property rights often is the most contentious point between universities and industrial firms. Other difficulties include the incompatibility of the time scales between university research and industrial needs, questions about who decides the direction of the research, limitations on open publication of results, and issues concerning the continuity of funding.
Universities, which want to protect their patent rights, deserve some of the blame when disputes over intellectual property rights strain their relations with industry. But industries deserve a share of the blame as well. In some cases, industrial sponsors want to own the intellectual property of an entire laboratory although they are sponsoring only a small part of the research. What is generally overlooked in this tug-of-war is that, in most fields, there is a 10- to 30-year period between the fundamental research and the development of a viable product based on that research.
The problems attendant with industrial funding of university research can and should be settled among the faculty, universities, and the industry sponsors before the onset of research. Almost all university research involves educating graduate students, and it is critical that students have the opportunity to choose a suitable research topic and to have the funds needed to carry out their thesis projects. In many cases, agreements between industry and universities have worked satisfactorily. In others, unneeded rancor resulted from misunderstandings and different expectations.
It seems to me that universities are convinced that the economic value of faculty research could benefit them financially. Considering the lack of success of almost all university intellectual property ventures, I can’t see what this conviction is based on. Only a few universities make a significant net profit on their intellectual property development efforts and it appears most of that income derives from medical and biological research. Success stories in those fields and in software development fuel the financial ambitions of both universities and faculty.
Contrary to reality, many faculty members think that their intellectual products are extremely valuable. In general, though, they are unwilling to risk developing their intellectual property themselves and prefer to have it patented and licensed to industry. In exceptional cases, faculty have formed companies to exploit the results of their university research and, in some of these cases, the universities have taken equity positions. A small number of faculty start-up companies have been successful.
Faculty ventures into commercial enterprises create a number of potential problems for both faculty and the university. Developing and operating start-up companies is a full-time endeavor: Such significant time expenditures by faculty clearly compromise their commitments to university teaching and research. Graduate students may be recruited to work in the commercial ventures without suitable compensation and to the detriment of their studies. University research projects may be selected to meet the needs of a small beginning enterprise. How university facilities are used in the commercial ventures is also a thorny issue.
Unless universities come to grips with the various difficulties I have discussed, they will increasingly have to deal with disputes between students and faculty advisers about ownership of intellectual property, questions about inappropriate use of university facilities, and commercial firms litigating against university researchers to prevent their results from affecting business ventures—the list goes on.
Recommendations to universities
In a number of ways, universities can improve the manner in which research is conducted and evaluated so as to mitigate many of the problems I have highlighted.
Universities should develop shared central facilities that are the common responsibility of the faculty and the university. The nature of these facilities and the services they provide should reflect the faculty’s research needs. Shared facilities would give all faculty access to needed instrumentation, and would relieve them of burdensome efforts directed at obtaining and maintaining complex instrumentation. Young faculty members would have access to facilities not customarily available to them in the early stages of their careers. Note that universities already maintain shared facilities—for example, animal care facilities—to support some areas of research.
Obtaining significant research funding increasingly requires a degree of cost sharing. Universities that are serious about research must recognize this development and have in place criteria for selecting from among requests for cost-sharing contributions. It no longer makes sense to deal with each request on an impromptu basis.
Universities need to make a serious effort to substitute quality for quantity in evaluating publications for promotion and tenure. One step in this direction could be to allow faculty to include only a limited number of publications in their promotion or tenure files.
The most important goals of the university are education, the creation of knowledge through research, and public service. As universities formulate policies governing their interactions with industry and, in particular, the commercialization of intellectual property arising from research, they must establish guidelines that least compromise these goals. Of particular concern should be the involvement of faculty in the operational activities of startup companies.
Recommendations to funding agencies
Federal funding agencies must choose their policies wisely to preserve the quality of university research. Many agencies hire university faculty as short-term employees. This practice should be increased, as it gives the university and funding-agency communities a great opportunity to understand each other.
Agencies should base their funding decisions primarily on the quality of the proposed research, the quality of the prior achievements of the proposing groups, and the mission needs of the agencies. They should provide a level of funding high enough and a term of funding long enough to support the research they enable. If, due to annual budgetary constraints, agencies are unable to commit to long-term funding, they could indicate their intent to continue funding if a project is going well. Termination of research support should be based on a critical assessment of the performance of the funded group, not on preconceived time limits.
The quality of proposed research, prior achievement, and continuing success should be judged by a peer review—the best means for assessing the quality of proposals, despite all its faults. To best ensure that proposals are assessed fairly, agencies must seek reviewers who are competent in the proposed areas of work and free from conflict of interest. No system is perfect, though, so there should be an appeals process that is independent of the agency components involved in making a contested evaluation.
Improving science education for all segments of society and extending research opportunities to groups that have previously not enjoyed such opportunities are commendable goals. But, unless there is clear evidence that coupling outreach to research is an effective way of achieving such desired social goals, research proposals should not be required to incorporate outreach.
Programs such as EPSCOR should be evaluated to determine whether they achieve their goals of improving research in disadvantaged states and such evaluations should determine their continuation. The “margarine method” of spreading research funds equally thin among all possible recipients is a waste of resources. Funding agencies should adopt the philosophy that, in general, “equality” in research funding means equality of proposal consideration, not the dispersal of funds equally among various groups of claimants.
REFERENCES
Howard K. Birnbaum ([email protected]) directed the Frederick Seitz Materials Research Laboratory of the University of Illinois at Urbana-Champaign from June 1987 to December 1999. He is currently an emeritus professor of materials science at Illinois.
