In J. W. Sommer, ed. (The Independent Institute, San Francisco):
The Academy in Crisis: The Political Economy of Higher Education.
New Brunswick & London: Transaction Publishers, 1995, for Independent Institute.
Recipient of the Sir Antony Fisher International Memorial Award (1996).
SOCIAL WARRANTS FOR FUNDAMENTAL RESEARCH: IDEATIONAL VS. INSTRUMENTAL
The Contemporary View: "Instrumental" Science
Resource Allocation: Ideational vs. Instrumental Science
Instrumental Science vs. Physical Capital
Acknowledgements This paper was initially undertaken in 1976 at the request of Bruce L. R. Smith as part of a larger study of the state of academic science, supported by the National Science Foundation. At the request of John W. Sommer, it was resuscitated for presentation at the meetings of the Association for Public Policy Analysis and Management (Austin, Texas, October 1986) and inclusion in the present volume. I would note the contributions, direct and indirect, of the late Derek J. de Solla Price and of W. Lewis Hyde, Adair L. Waldenberg, Merton J. Peck, Kenneth R. Janson and (especially) Karol I. Pelc, none of whom, of course, can be deemed responsible for the conclusions reached.
The premise of this essay is that the fundamental inadequacy of the prevailing economics of fundamental research is the failure to identify what it is that research contributes to society. When this issue is clarified, a more operational, decentralized, although perhaps no less difficult, approach to the issue of the allocation of resources to fundamental research becomes possible.
As discussed by Habermas,(Note 1) central to this classical view of science was the perception of a virtually unbridgeable chasm between "theory" (including science) and "practice" (including technology): The concern of theory was with the "immutable essence of things," while the conduct of practical affairs was "pragmatically practiced according to traditional patterns of skill." In fact, this chasm was bridged, but only indirectly: Theory "obtain[ed] practical validity only by molding the manner of life of men engaged in theory." Thus, the capacity to comprehend (and engage in) theory, to perceive the "immutable essence of things," provided an ethical orientation to action and thus represented a sociocultural qualification required of those whose practical actions would have consequences for others. Knowledge of theory provided the moral sanction for practical action, while practical action itself required only pragmatic, instrumental, technical qualifications which were quite unrelated to theory.
From this perspective fundamental research is viewed as an investment in a stock of useful, directly applicable knowledge. Thus, through research chemistry contributes knowledge directly employed in the chemical industry, physics contributes the knowledge necessary for nuclear weapons and high-temperature superconductors, etc. This view of science as a body of knowledge, augmented through research, the applications of which are of direct benefit to society, underlies the conventional differentiation between basic research, applied research and development. Research results in an augmentation or net addition to the stock of knowlwedge: Basic research provides the superstructure of these net additions, while applied research fills in the crevices of direct relevance to particular applications. Development contributes the last engineering details. The essential point is that there is perceived to be a unidirectional flow from basic research to applied research to development, and the value at each stage derives from the stream of benefits flowing from the practical actions made possible by the advance of knowledge. (Note 2)
In principle, either the intrinsic-utility or investment-in-knowledge conceptions can be utilized, at least conceptually, to derive a social demand for and optimal allocation of resources to fundamental research:
Thus, it is not surprising that appeals for support of science as an investment in prospectively useful knowledge commonly degenerate into appeals for support of science as an end in itself. Because it is impossible to anticipate what benefits will flow from fundamental research, it is inappropirate to ask the scientist to justify social subvention of his research; that research becomes its own justification.
Entirely apart from the issue of operational usefulness, a serious difficulty associated with both of the dominant conceptions of fundamental research (social-consumption and investment-in-knowledge) is that they call into serious question the rationality of the observed rate at which resources are allocated to research.
Whatever the level of utility derived by the "representative" member of society from the existence of scientists engaged in "science," that utility would not be expected to be greater than the utility associated with, e.g., the arts. Knowledge of and appreciation for the arts may not be universal, but popular awareness of the arts must be substantially greater than popular awareness and appreciation of science. People actually have first-hand exposure to symphony orchestras and operas, while virtually no nonscientist observes or is even aware of (much less obtains utility simply from the conduct of) scientific research. And, while institutions such as the Smithsonian, the Chicago Museum of Science and Industry and Munich's Deutsches Museum offer popular entre to the wonders of science to significant numbers of people (although their emphases, and the interests of their visitors, focus primarily on technology rather than science), the number visiting art museums must be substantially greater. Yet the social resources devoted to science exceed dramatically the resources devoted to the arts (on the order of 15:1).
It might be responded that the supply of inputs (especially of capable personnel) to social-utility-generating art is substantially less elastic than to social-utility-generating science. Specifically, it would be argued, social utility, whether of art or science, flows not simply from artists or scientists engaged in art or science; rather, social utility is a function of the quality of artistic and scientific endeavor, while quality, in either art or science, is the consequence of the artistic or scientific talents of the practitioners, talents which are not uniformly distributed over the population. Thus, as the allocation of resources, to either art or science, is increased, the talents of marginal personnel (the last added to the cadre of practitioners) must decline, implying diminishing marginal social utility.
To explain the disproportionate allocation of resources to science relative to art on grounds of inelasticity of supply of talent, however, requires that the distribution of artistic talent be substantially more highly skewed than the distribution of scientific talent. In light of substantial evidence of an extremely skewed distribution of scientific talent (as discussed further below), this seems highly unlikely. (Note 4) At the least, it is difficult to believe that differences in the distributions of artistic and scientific talent can explain the observed order-of-magnitude difference in social resource allocation.
Also relevant here is the issue of the relative elasticities of supply of physical capital goods and of scientific talent. While capital goods sectors may exhibit relative inelasticity of supply in the short run, there is no reason to believe that this elasticity is not quite high in the intermediate and long run, when industry capacities can be adapted. In contrast, as suggested in the discussion of the relative elasticities of supply of scientific versus artistic talents, the distribution of scientific ability or competence is highly skewed, while this ability or competence is the primary determinant of the contribution of research to new knowledge. Thus, available evidence suggests that the marginal researcher, the last one admitted to the ranks of the active scientific community, contributes little or nothing to the stock of scientific knowledge.
Specifically, the distribution of scientific productivities is such that, in virtually all fields at all times, the greatest share of scientific output is produced by a very small fraction of all active scientists. Consider, for example, the cohort of young scientists publishing in the scientific literature for the first time in a given year, examined by Price and Gürsey. (Note 6) Over the 30 to 40 year professional lifetime of members of the cohort, 25 percent of the cohort's members will contribute more than 75 percent of the scientific papers attributable to the cohort, and the most productive five percent will account for more than half of the cohort's published output. Two-thirds of the cohort will in fact vanish after their first appearance in the scientific literature. If any sort of quality adjustment were applied to these publications, the skewness of the distribution of scientific productivity would almost inevitably be found to be even more extreme.
The general conclusion must be that a very small proportion of ever-active scientists and scholars must be credited with the greatest bulk of contributions to scientific and scholarly knowledge, even allowing for a substantial margin of error in the computations and for possibly significant variations over fields. Thus, at the margin, resources devoted to research produce very little of ultimate scientific value. Assuming even a highly imperfect mechanism of selection (and, most importantly, of selfselection) of scientists, i.e., for predicting an individual's future scientific productivity (contingent on his pursuit of a scientific career), for all practical purposes it can be reasonably concluded that the scientific world (and the national economy) would lose little in terms of scientific knwoledge as a result of, e.g., ex ante abortion of fifty percent of a cohort which would otherwise reach the stage of active participation in fundamental scientific work at the research front; whatever the reduction in support for scientific work, the reduction in the level of scientific output (additions to knowledge) would certainly be much less than proportionate.
Thus, for example, in a stylized model of fundamental science Dresch and Janson find that a roughly two-fold (94 percent) increase in the size of the cadre of fundamental scientists, from 0.47 to 0.91 percent of the underlying population, would result in only a 50 percent increase in the quantity of scientific output (e.g., publications) and a 36 percent increase in the scientific value of these contributions, while a further 225 percent increase in the science cadre, to three percent of the population, would increase the quantity of output by less than 100 percent and its scientific value by only 40 percent. (Note 7)
In short, because scientific productivities are not uniformly distributed over the population, employment of a greater fraction of the labor force in the scientific enterprise would be expected to result in rapidly diminishing returns as measured by contributions to knowledge. Even if the long-run returns to new knowledge were predictable, substantial and nondiminishing, the stream of returns to the marginal, i.e., small, scientific contributions of the marginal, i.e., last hired, scientist, engaged in either basic or applied research or in development, would eventually be less than the benefit which would derive from alternative employment. At that point, funding of research (as an investment in knowledge or its applications) should cease. Thus, the skewed distribution of scientific ability results in a highly inelastic supply of scientific knowledge, placing a severe brake on the flow of resources to investments in knowledge, while no similar brake is imposed on investments in physical capital.
If there was any relationship historically betweeen theory/science and practice/technology, it was the reverse of that conventionally posited currently: Advances in science were driven by advances in technology. Modern astronomy developed from the practice of astrology and from improvements in lenses, chemistry from the evolution of the dye and related chemical industries, classical thermodynamics from the development of the steam engine. In essence, practical action, experience, provided numerous uncontrolled experiments through which instrumental knowledge (technology) was augmented, and these augmentations of instrumental knowledge made possible advances in theoretical knowledge as well. (Note 8)
The transition from the classical, intrinsic-utility, conception of the role of science to the current instrumental, investment, conception can be argued to have been driven by a fundamental change in this historically descriptive relationship between science and technology. However, the characterization of the newly emerging relationship between science and technology as involving a unidirectional flow from the former to the latter is, at best, a seriously misleading caricature.
Perhaps the most misleading aspect of the conventional, unidirectional view of the science-technology relationship is the implicit attribution of purposiveness to scientific development. Scientific advances, supposedly, are motivated by their prospective practical uses or applications. In fact, when the consequences of "scientific" work can be predicted with any degree of certainty, the effort is inevitably primarily of an engineering nature. Fundamental knowledge may be augmented in the process, but this is essentially a byproduct, even an unintended byproduct, of technological work. Stated somewhat differently, science may have become susceptible to direct application, but specific prospective applications do not motivate or induce specific augmentations of scientific knowledge.
A more descriptive characterization of the evolving relationship between science and technology would emphasize reciprocity, multidirectional influences and flows. Augmentations of scientific knowledge may not be motivated by specific practical consequences, but effective practical action necessitates, is dependent upon, scientific knowledge: Scientific developments do have practical, technological implications. The technologist or engineer of several centuries ago would have benefitted little from a theoretical knowledge of Newtonian physics; in contrast, the contemporary technologist is doomed to failure or, at the least, inferiority or inefficiency in the absence of fundamental scientific knowledge. (Note 9) Although the eventual feasibility of nuclear weapons cannot be argued to have provided the motivation for the development of nuclear physics, the effective realization of the objectives of the Manhattan Project would have been impossible in the absence of research-front knowledge of nuclear physics. Historically, effective contributions to science required the scientist's grasp of engineering, but not vice versa. Today, the engineer is comparably dependent on a research-front knowledge of science. For example, the key personnel on the Manhattan Project were scientists functioning as engineers, because only scientists possessed the fundamental knowledge necessary for the engineering effort. The relationship between science and engineering, between theory and practice, truly has become reciprocal.
At this point a third perception of the social role of fundamental research begins to emerge: In fact, knowledge does "advance," albeit in a manner which is unpredictable and with uncertain practical consequences. And, in a variety of pursuits a firm command over the margin of knowledge is a prerequisite of effective performance. Finally, research scientists do perform social functions other than research, and their effectiveness in these other functions is affected by, indeed is dependent upon, their participation in research.
From this third conception of the social functions of and warrants for fundamental research, very different from the intrinsic-utility and investment-in-knowledge conceptions, the purely scientific output of the research performer is irrelevant, no more than a fortunately accidental (and generally inframarginal) byproduct. The key issue concerns the effectiveness of the performer in some other dimension. Thus, for example, the value of the research of the academic scientist resides not in the articles he publishes, which are, in any event, read only by other (primarily academic) scientists (the most intensely interested of whom will have seen the article in draft or preprint form long before its formal publication (Note 10)), but in the quality of his students, the value of his advice as a consultant to government or industry, etc. More generally, the value of research is reflected not in the "new knowledge" (if any) which is produced and not in the (possibly nonexistent) "applications" of these (also possibly nonexistent) augmentations of knowledge, but in the capability of the researcher to produce on contract something which it is possible for a customer to order on the basis of knowledge at hand, whether a neutron bomb, an explanation for aquired immune deficiency syndrome, an adequately educated student, ... Thus, as characterized by Price,
.... society, somehow manifested, decides that some new understanding or new data is needed for a job (for example to locate a new airport, inoculate a population against some disease, advise on the building of various alternative energy plants, etc.)... [T]he research function [is] clear. It is a question of knowledge that society needs and must provide ... either through contract or through an agency charged with such a mission. In either case the customers know what they want and can later decide whether their wants have been satisfactorily provided in relation to the costs and competences. The service provided is then in the nature of an investment good and can be evaluated as technological research in terms of whether the outcome measures up to the service specified or contracted for. (Note 11)Most generally stated, these directly-valued functions involve (a) technological applications of expertise, not for the purpose of augmenting knowledge but rather for the purpose of achieving what is perceived (by the customer as well as by the supplier) to be a practical, attainable objective, and/or (b) the development of students possessing research front knowledge and capable of providing these technological applications of expertise.
The Manhattan Project, the space program, the Strategic Defense Initiative and the searches for causes and cures of specific diseases constitute particularly visible technological applications of research front expertise. In practice, it may be difficult to distinguish these directly-valued "engineering" services of research-front scientists from their fundamental research, and the distinction is certainly not well captured by the concepts of either development or applied research in the usual R&D trilogy. For present purposes it is not necessary to develop a fully operational definition of this activity. The essential points are that it is possible (a) for the customer (perhaps with expert assistance) to identify the objective or product, (b) for a suitably trained individual or group to produce and deliver something approximating the desired objective or product, and (c) for the effectiveness of the producing individual or group (or, for the quality of the product) to be enhanced through knowledge of research-front developments. Knowledge in a more general sense may be augmented in the process, or the technological product so produced may, in addition to its primary objective, create the opportunity to acquire more fundamental knowledge hitherto unavailable, but these byproducts will not have provided the underlying rationale for the endeavor.
Viewed in these terms, the basic valuational question relevant to the support of research has the form: How much more effective as a producer of students, vaccines for heretofore unknown diseases, sophisticated military hardware, ... is a person who has been engaged in research than is one who has not been engaged in fundamental research? Increments in effectiveness can then be contrasted with increments in research activity, providing the basis for identifying the socially optimal allocation of resources to research.
In this context, research has value precisely because knowledge does, indeed, advance. A casual nonparticipant observer can claim a command over any new knowledge only after an active participant in its development takes time to "popularize" the achievement. If we assume, arbitrarily but not unreasonably, that it takes ten years to fully establish and work out any significant change in the state of science, to the point that it can be put forth, even to professionally interested nonparticipants in its development, as the new orthodoxy, and if over the period of a decade knowledge in a given field roughly doubles, as indicated by the rate of growth of scientific literature, then a "mere observer," even one professionally trained, will be operating, on average, on the basis of knowledge which is fifty percent obsolete.
A parallel argument is developed by Price, (Note 12) who notes that the investment required to bring an individual to the research front is roughly four years of postbaccalaureate training. With a growth rate of knowledge (scientific literature) of seven percent per year (equivalent to the ten-year doubling period indicated above), in the absence of further investment in the maintenance of research front capabilities, the Ph.D. scientist would obsolesce at an annual rate of seven percent. Considering that, as a means of obtaining access to research front knowledge, the alternative to further investment in an established scientist is the training of a new Ph.D., an annual rate of investment in the maintenance of research front capabilities equal to 28 percent of the scientist's time (i.e., seven percent of four years of graduate education) is justified. Because this quantification assumes, most importantly, that an established scientist is, on average, no more efficient than the average graduate student in the development and maintenance of research front capabilities, when selection and selfselection of those scientists whose research front capabilities are sustained would be expected to leave only the more efficient in this group (with others diverted into occupations not requiring research front knowledge, e.g., management and administration), it would appear to offer an upper bound on the necessary rate of investment in the maintenance of research front knowledge.
In light of the uncertainties associated with new knowledge, e.g., the unanticipated side-effects which may accompany a new drug, the as-yet-unrecognized environmental consequences of a new source of energy, or the adverse strategic responses of a potential adversary to installation of a newly feasible weapons system, in many applications a substantial obsolescence may actually be desirable. But, for most purposes this degree of obsolescence would entail serious costs, as for example, in failing to consider the implications of current research front developments in decisions regarding substantial investments in a particular energy or weapons system which might be doomed to obsolescence before it would even reach the stage of operation.
Especially in the domain of advanced education and training, obsolescence of the indicated magnitude would imply probably fatal inefficiency. Imagine a "graduate faculty whose active participation at the research front had ended ten or more years earlier and who would, in consequence, be operating with 50 percent obsolete knowledge, capable of bringing graduate students only to the research front as it had existed ten years earlier. If it is assumed, reasonably, that the cost of bringing a graduate student to any research front is independent of whether that front is current or obsolete, then the only cost of bring the student to the current front rather than to the decade earlier front is the faculty research necessary to stave off obsolescence.
The issue with respect to that faculty research, then, is this: Is the value of a nonobsolete new Ph.D. greater than that of an obsolete new Ph.D. by an amount sufficient to justify support of that faculty research necessary to produce the former rather than the latter? That is, we must compare the current value of the "new" Ph.D. who has been kept on ice (placed in suspended animation) for a decade with that of a new Ph.D. whose training has brought him to the current research front. Such a comparison, obviously, is difficult to make, since the world provides no obvious natural experiments on which to base the calculations.
One possible index of the relative value of a nonobsolete to an obsolete new Ph.D. might be provided by lifetime earnings differentials of Ph.D.s related to the quality of graduate program, on the assumption that the "quality" index reflects differences in obsolescence of faculty. However, even if that assumption is justified, as it probably is, the better schools are more selective, suggesting that a significant fraction of the differential may be related to differences in innate talent, ability and prior achievement. On the other hand, differences in the post-schooling employments of Ph.D.s from different quality classes of graduate schools suggests that graduates of the better schools tend to select employment with relatively lesser pecuniary income and greater nonpecuniary perquisites (interesting and challenging colleagues, prestige, etc.). Thus, even the direction of bias would be uncertain.
In any event, if the difference in value could be determined, this would indicate the level of faculty research which would be justified solely on grounds of the implied benefit of less obsolete graduate students. If the (discounted) lifetime value of a 50 percent obsolete "new" Ph.D., Vmin, would be enhanced by v as a result of an additional decade of knowledge, i.e., if he were rendered nonobsolete, and if the annual ratio of Ph.D. production to graduate faculty were y, then the level of research support per faculty member which would be fully compensated by the augmented value of new Ph.D.s would be vy. More generally, the value of students (V) probably increases at a decreasing rate with increases in faculty research (R), reflecting a corresponding change in faculty obsoelescence (Z), decreasing at a decreasing rate with increases in the level of research effort. The "optimal" degree of faculty obsolescence, then, would be that level of Z for which (dV/dZ)(dZ/dR) = 1/y, measuring R in monetary terms. (Note 13)
The general point of the foregoing, to reiterate, is that, for purposes of decisions concerning the allocation of resources to fundamental research, research itself, or the products of that research, cannot be viewed as the essential output. Rather, research must be viewed as an activity essential for performance of other, directly-valued, functions. Specifically, research represents the process by which the capabilities of researchers, and hence their effectiveness in "education" and "production," are maintained over time, in the face of what would otherwise constitute an inexorable process of obsolescence.
Second, the intrinsic utility may derive in part from the recognition by others of the research accomplished or supported by an individual or group. In this case, the support of research is deemed to confer prestige on the supporter, and research is undertaken as a form of conspicuous individual or social consumption. Historically, such considerations (as well as intrinsic private utility) can be argued to have motivated the support of science by the nobility, (Note 15) by those of great mercantile or industrial wealth, (Note 16) and by organizations such as the church. In the twentieth century, one might argue that national prestige has motivated governmental support of fundamental science, (Note 17) with the status of a country determined in part by, e.g., its share of Nobel laureates. The resultant prestige may be of value not only in its own right but also because of the influence it confers in other domains (e.g., diplomatic) or because of social as well as scientific emulation (e.g., of the political or economic system capable of achieving such prestige in science). This type of motivation is one of several possible sources of what might be characterized as "international competition" in science, a phenomenon which might also be observed on the part of other (sub- or supranational) entities (e.g., corporations) as well.(Note 18) The critical point, however, is that science is of value not in its own right but because of its influence on the perceptions others have of the research-sponsoring entity, which will determine its level of support for science accordingly.
Finally, third, engagement in and/or support of research may be deemed necessary for the evidence it provides of the noninstrumental (e.g., social or cultural) qualifications for performance in other domains. Thus, as discussed above, the distinguishing character of the professions, historically, has been the possession by their members of appropriate (socially sanctioned) moral qualifications for the performance of functions having consequences for the welfare of others, with the capacity to comprehend and engage in science deemed to provide an ethical orientation to practical action. In this case, the level of scientific activity will reflect the level of demand for those personnel for whom science is deemed to provide necessary noninstrumental qualifications for performance and on the effectiveness with which the noninstrumental qualifications of these personnel can be regulated, either by the market or by governmental fiat.
To the degree to which access by others can be precluded, an entity may have some incentive to engage in research. However, the level of research which will be supported will be less than would be the case were the benefits of maximal exploitation to be recognized. The resultant underinvestment by private (socially subordinate) units and their motivations for precluding access by others are argued to provide the justification for governmental subvention of research, with provision of free access to the knowledge thus produced.
However, short of a global governmental subvention, governmental underwriting does not resolve the problem. Even if access by other countries to the results of any country's research can be prevented, global investment in research will remain nonoptimal. Although there may be "international competition" in science, much of the work supported in any country will be duplicative of that supported in others, and the progress of science in all countries will be retarded by comparison to a global regime under which access to each country's scientific accomplishments is freely available to all other countries.
In fact, in fundamental science it will be extremely difficult, if not impossible, to foreclose access, simply by the nature of the work involved. Within any system, e.g., country, science will have value only if it is relatively freely available to all. However, given the form in which fundamental science is disseminated, through publications and, even more importantly, through collaboration by scientists in "invisible colleges," free access within any domain will virtually foreclose prohibitions of access to external parties. Thus, it will be effectively impossible for any country both to insure free internal access and to foreclose external access.
Even in the case of the United States, a major source of fundamental research, these considerations have devasting consequences for the rationalization of the allocation of resources to research. If the U.S. contribution to total world investment in knowledge is measured, for purposes of argument, by the U.S. share of the world scientific literature, then it would appear that at least 60 percent of new knowledge generated by research would be available to the U.S. even if it invested nothing in research. Such a free-rider policy might be considered ethically wrong, and, more fundamentally, if it were pursued by all countries, a free-rider policy would obviously lead to zero global investment in knowledge. Nonetheless, at the margin it would appear that the U.S. would lose little in terms of the general stock of available knowledge as a result of even a significant reduction in the level of research support. Viewed differently, the access of the U.S. to global scientific "production" provides a further basis for anticipating an elasticity of supply of knowledge substantially less than that of physical capital. (Note 19)
The implication of this argument, however, is that there exists a necessary interdependence between the efficient level of scientific activity in any one country (or enterprise/firm) and the level of scientific activity in others, and vice versa. The greater the level of scientific activity elsewhere, the greater the incentive to engage in research in order to monitor and be in a position to exploit scientific achievements elsewhere. However, the effort required to monitor scientific developments elsewhere must be less than that required to achieve those developments, especially in light of the skewed distribution of scientific talent, which will imply rapidly diminishing augmentations of knowledge as the scale of research-front activity is increased; by implication, an increase in scientific activity elsewhere will generate a less than equiproportionate increase in the rate of augmentation of knowledge and hence will motivate a less than equiproportionate increase in the desirable level of "own" research effort. Thus, there presumably exists a stable, "equilibrium" relationship between research in any given country (enterprise/firm) and research in the rest of the world.
This relationship is represented stylistically in Figure 1, in which the desirable scale of fundamental research in country "A," denoted and measured either in (real) monetary terms or in full-time-equivalent employment of persons engaged in fundamental research, is represented as a function of fundamental research in country "B," and vice versa. The positive slopes of the curves indicate the increase in scale of scientific activity in each country motivated by additional research in the other country. However, because the slopes are increasing at, respectively, increasing/decreasing rates, there is a point of intersection beyond which neither party finds it desirable to devote additional resources to research. These slopes reflect both diminishing scientific returns to research in each country (attributable largely to the skewed distribution of scientific talent) and the fact that monitoring scientific activity elsewhere is less than a full-time occupation, i.e., that research front developments attributable to one full-time-equivalent scientist can be monitored by a scientist devoting less than full time to work at the front.
It should be emphasized that these positive reaction functions do not represent the operation of an "international (or interfirm) competition" in science. The objective is not to outproduce the other country (firm) but to be in a position to exploit the other country's (firm's) scientific achievements. Thus, ironically, the desire to appropriate the knowledge of others provides the motivation to contribute to the aggregate stock of knowledge, a rather interesting reversal of the free-rider position.
While the general characteristics of these research reaction functions are unambiguous, their precise positions (vertical/horizontal placements) are less determinate. In light of the preceeding discussion of the role of research, it can be anticipated that the level of research effort is ultimately determined by the domestic (or intraorganizational) extraresearch functions of scientists. The demand for technological applications of expertise determines, directly or indirectly, the efficient allocation of resources to research, where the issue of "the efficient allocation of resources to research" is essentially one of the number of persons for whom knowledge of research front developments is a key determinant of effectiveness in other pursuits. Thus, for example, a change in the composition of economic activity in country "A" toward "meditation" would shift its research reaction function to the left (assuming that "meditation" is not a technologically sophisticated activity), while a military confrontation would move the function to the right.
This analysis has profound implications for the institutionalization and financing of fundamental research. While it is not possible to fully explore these implications within the confines of this discussion, the following provides at least a suggestive sketch.
The principal conclusion of the foregoing discussion is that fundamental research serves primarily as the mechanism by which the capabilities of scientific personnel are maintained over time. These capabilities have value, however, only to the degree to which they contribute to the effectiveness of research personnel in other pursuits. If there were to be no future demand for scientific-technological capabilities obtained and sustained through participation in fundamental research, then there would be no justification for maintaining the capabilities of scientific personnel, and research support could fall to zero (or to that level which would be justified by the intrinsic social utility of science).
In this context, the party with the most direct and immediate incentive to bear the costs of research is the researcher himself, who can anticipate higher earnings (prices for his services) in the future if he is less scientifically obselete. Thus, in a pure market system the costs of research would be borne, in the first instance, by the scientist, who would recover these costs through subsequent earnings differentials attributable to research-maintained scientific knowledge and capabilities. Individual scientists would invest in their own capabilities (through research) up to the point that the return to that investment equaled the returns to other comparably risky investment alternatives. If there were an excess supply of nonobsolescent scientific capabilities at any point, the prices which could be commanded by research-front capabilities would be less than sufficient to compensate scientists for the costs of that research necessary to sustain these research front capabilities, with the result that fewer individuals would invest in the maintenance of their capabilities, i.e., some persons would make career changes for which obsolescent capabilities would be adequate, while others would reduce the rate of investment, permitting their capabilities to degrade slowly over time. Conversely, an excess demand for nonobsolescent scientific capabilities would imply service prices more than sufficient to compensate for the costs of research, motivating higher proportions of suitably-positioned individuals to acquire and maintain knowledge of research-front developments.
At a superficial level this model appears to be ludicrously at variance with reality. Clearly, researchers do not "finance" their research. Implicitly, however, this model may well be at least partially descriptive. If researchers did finance their research, their earnings, as consultants, contractors, faculty, ..., would necessarily be significantly higher than actually observed earnings. Thus, it is possible to interpret the costs of research as an implicit component of gross compensation, a component which the scientist elects to reinvest in the maintenance of his research-front knowledge and capabilities.
This implicit model of scientists' selffinancing of research may be highly descriptive in the case of those sciences, e.g., mathematics and theoretical physics, in which the dominant input into research is the scientist's own time. Consider a scientist for whom it is rational to devote 20 percent of his time to research (an investment which maximizes the present value of his lifetime earnings as a faculty member, consultant, ...). He is remunerated (in the form of salary or commissions) for the other 80 percent of his time which he dovotes to some other activity (teaching, consulting). Interpretively, had he desisted from research and devoted full time to the provision of income producing services, his earnings (in the short run) would have been 25 percent greater, a differential which he elected instead to invest in the maintenance of his capabilities.
Notwithstanding its stylized form, this model suggests interpretations of several observations drawn from reality. Notably, it explains the apparent decline with age in the proportion of time devoted to research. Ceteris paribus, any time devoted to research will generate higher returns for a younger scientist than for his older colleague, simply because of the longer period over which he can expect to provide services; thus, at least beyond some age it will be rational for the scientist to shift the allocation of his time from research to service provision. Under plausible circumstances, this reduction in time devoted to research with advancing age will also be reflected in increased observed earnings. On the one hand, the older scientist, having rationally reduced his investment in obsolescence-offsetting research, will experience a decline in earnings as a result of his progressively greater obsolescence; however, over some period (immediately following the reduction in his time devoted to research) this may be more than fully compensated by the higher proportion of his time devoted to service provision.
Consider, for example, Price's (upper-bound) estimate of the proportion of a scientist's time which should be devoted to maintaining his competence, 28 percent. Were he to reduce the proportion of his time devoted to research to zero, time devoted to service provision would increase by 38.9 percent Under Price's assumptions he would become obsolete at a rate of seven percent per year; plausibly, this would reduce his earnings capability by seven percent peryear. However, the effect of the increase in time devoted to service provision would outwiegh the effect of increasing obsolescence for the first 4.86 years after his cessation of research. Thus, maximization of lifetime earnings (net of the costs of research) would dictate cessation of research approximately five years before the expected termination of the working lifetime (and even earlier if the real discount rate is greater than zero).
A second phenomenon explained by this analysis concerns the budgets of solicited grant and contract research. Here the sponsor essentially is contracting for a specific (anticipatable) product, and the service provided by the contractor is not fundamental research. However, budgets for such research are commonly, indeed "notoriously," inflated; because these contracts typically provide only for cost reimbursement, this budgetary inflation is accomplished by exaggerating the time of the scientist required for to the contracted project. The contracting agency is prepared to pay the higher price because only thus can it obtain the services of a research front scientist. The scientist devotes the excess of budgeted over required time to the maintenance of his capabilities, i.e., to fundamental research. Thus, grant and contract support is frequently acknowledged in the fundamental scientific literature not because augmentations of fundamental scientific knowledge were the intentions of the sponsors but because fundamental research is a "pirate activity" carried out in association with the provision of other services, when effective service provision requires fundamental knowledge. Only in the event of an excess supply of particular research front capabilities will it be possible for the contracting agency to acquire such services at a price which does not provide for the maintenance of research front capabilities.
The important implication of the alternatives of explicit versus implicit inclusion of research support in the gross earnings of research-front personnel concerns the locus of risk bearing and, hence, the role of institutions. If support for research is explicitly incorporated in the earnings of the scientist, then the scientist is ultimately responsible for decisions concerning the appropriate rate of investment in the maintenance of his capabilities through research. Thus, his anticipations of the future earnings differentials associated with greater versus lesser allocations of resources to current research will determine the level of his investment in research. Obviously, however, this decision must be made in the face of significant uncertainty concerning future demands for knowledge and expertise. To the degree to which he is risk averse, his investment will be reduced by comparison to the socially optimal rate of investment in research-front capabilities, i.e., to the rate of investment which would be observed if ex post moral hazard and and ex ante adverse selection did not serve to constrain opportunities to insure against adverse market developments.
Institutionalization of research support, with the employing institution (a) compensating the scientist at a rate net of support for research, (b) underwriting the scientist's research and (c) recovering the costs of research through its charges for the scientist's services to clients (students, agencies contracting for specific services, etc.), provides a mechanism by which it may be possible to effectively pool the risks associated with investments in research-front capabilities. The employing institution, maintaining a balanced portfolio of research-front capabilities, thus can utilize the unanticipatedly high returns to certain types of capabilities to compensate for lower than anticipated returns to others.
A serious constraint on the capacity of the employing institution to pool the risks associated with individual decisions to engage in obsolescence-offsetting investments in scientific capabilities derives from the capacity of individuals whose capabilities are in unanticipatedly high demand to expropriate the excess of actual over expected payments for services. Risk pooling essentially means that future net earnings of the individual scientist will be dependent only on the previously anticipated (expected) state of demand, not on the state of demand actually confronted. Thus, the institution will incur losses on those whose expertise is in less than anticipated demand, recovering these through the higher-than-anticipated prices which can be commanded for the services of those whose capabilities are in unexpectedly strong demand. If, however, the scientist whose capabilities are in strong demand is free to exit the institution, either directly marketing his capabilities or associating himself with another institution, then the employing institution will not be able to retain title to these higher than anticipated earnings. Thus, provisions such as Constitutional prohibitions of involuntary servitude, which preclude even the voluntary waiver of the right to sever an employment contract with an institution, serve to severely constrain the risk-pooling functions of employing institutions.
Although an institution may be precluded from utilizing legally binding agreements to insure that those whose research is supported by the institution will not expropriate unexpectedly high returns to that investment, certain classes of institutions may be able to utilize moral suasion to achieve the same end. Canons of "professional ethics" and collective ideologies and values, e.g., those associated with the academic professions, can be viewed as providing, inter alia, for extralegal enforcement of voluntary agreements involving the waiver of the right of the individual scientist to expropriate the surpluses associated with unanticipatedly strong demands for his expertise. To the degree to which these are successful, the employing institution can serve a risk-pooling or insuring function. In actuality, however, success in restricting exit rights of employees is more likely to be attributable to collusion between collectively monopsonistic employing institutions than to the strength of the selfimposed extralegal constraints on individual mobility stemming from collective ideologies and values.
Even if employing institutions are incapable of performing a risk-pooling function, there may well be motivations for scientists to affiliate themselves with institutions. For example, the institution may perform an informational function for the market, providing a certification of the capabilities of the individual scientist, especially in the face of serious constraints on the capacities of those purchasing scientific services to effectively evaluate the capabilities of individual providers. Moreover, the conferring institution may be able to impose an effective change on the scientist for this informational imprimatur, providing a potential source of funds compensating for its selfinsurance losses.
This entire system breaks down, however, if the scientist and his employing institution are precluded from incorporating in the charge for his services the costs of that fundamental research necessary for the maintenance of the research-front capabilities of the scientist, even when the "market" would permit recovery of these costs (i.e., even if there were no excess supply of research-front capabilities and, thus, no lower bidder could be found). As suggested above, grant and contracts for specific services generally (i.e., in equilibrium or excess-demand situations) do include budgetary "slack" necessary to cover associated capability-sustaining fundamental research.
However, in the case of one of the largest direct and indirect purchasers of the services of research-front scientists, government, "cost-accounting standards" are intended to preclude this incorporation of an "implicit charge" for fundamental research in contracts for specific technological services. On the other hand, it has been impossible for the government not to recognize the necessity for research-front capabilities. Thus, to the degree to which prohibition of implicit funding of fundamental research has been successful, it has been necessary for government to introduce direct funding for this capability-sustaining research. Especially in light of the constraints faced by institutions in performing the risk pooling function for which they are uniquely qualified, they have been prepared to accept constraints on the prices which can be charged for scientific services in exchange for direct governmental funding of fundamental research, thus, explaining the emergence of fundamental-research-supporting agencies such as the National Science Foundation.
The adverse consequence of this creation of governmental agencies sponsoring fundamental research is that it breaks the connection between fundamental research and its justification, the maintenance of research front capabilities for which there is an extraresearch function or requirement. Thus, it becomes possible, e.g., for individuals performing no useful nonresearch functions to be supported only for purposes of their fundamental research. As a result, the supply of scientific talent to other activities (whether these do or do not require research front capabilties) may be seriously reduced.
Moreover, this allocation to government of the responsibility for financing fundamental research centralizes judgments concerning expected future requirements for research-front capabilities. While an individual may under- or overestimate the future demand for his particular research front capabilities, these individual errors are likely to be less than perfectly correlated across individuals, thus partially averaging out. In contrast, with a centralized determination, there will be no compensating under- and overestimations of future demands, since only one estimate (draw from the distribution) will be made. Also, given lags between the various phases of the process of determining levels of governmental support, e.g., between the assessment of "need," the authorization of funding and the actual awards of support, this governmentally managed system is likely to be seriously out of phase with actual market developments, even if these developments were accurately anticipated in the first instance. It is for this reason that a centralized system is likely to be characterized by successive boom and bust markets, recurrent expansions and contractions of supply of research front capabilities.
2. While not of direct relevance here, it is interesting to note that the transition from the classical, intrinsic utility perception to the contemporary instrumental-utility, investment-in-knowledge perception involves a fundamental change in the perceived nature of theory and science as well as in the relationship of theory/science to practice/technology. From insight, albeit partial, into the "immutable essence of things," theoretical knowledge itself becomes transitory. While individual observations may (but need not) be "immutable," the meaning or interpretation of these observations becomes provisional, tentative, subject to radical reformulation on the basis of future observations. While such radical reformulations occurred in the past, as in the revolutionary displacement of Ptolomeic by Coperincan astronomy, these reformulations were the exception rather than the rule; most importantly, they did not undermine the subjective perception that scientific knowledge was concerned with immutable essences. In contrast, among adherents of instrumentally utilitarian science, it would be difficult to find anyone who seriously believes in any immutable essence; contemporary theory is validated not by the insight it provides into the immutable but by its perceived usefulness, a usefulness which is at best temporary as new observations undermine existing theory.
3. Operationally, given any level of basic and applied research, the optimal level of development could be determined by the principle of marginal cost equal to (discounted) marginal benefit. Maintaining the optimal rate of development as a function of the rates of basic and applied research, the optimal rate of applied research, given the rate of basic research, could be determined by the same principle. The same procedure would then be employed to determine the optimal rate of basic researh.
4. On the relative elasticity of artistic talent, see J. Barzun, "A Surfeit of Art", Harper's (July, 1987).
5. This decomposition of economic growth into components attributable to specific sources of growth is taken from the "growth-accounting" analysis of E. F. Denison, Accounting for United States Economic Growth 1929-1969 (Washington, D.C.: Brookings Institution, 1974).
6. See D. de S. Price and S. Gürsey, "Studies in Scientometrics, Part 1: Transience and Continuance in Scientific Authorship," International Forum on Information and Documentation (Moscow, 1976) 1(2):17-24; reprinted in, D. de S. Price, Little Science, Big Science ... and Beyond (New York: Columbia University Press, 1986), in which also see Chapter 2 (Galton Revisited).
7. See S. P. Dresch and K. R. Janson, "Giants, Pygmies and the Social Costs of Fundamental Research, or Price Revisted." Technological Forecasting and Social Change, 32/4 (1987).
8. For a discussion of technological advance as the driving force of scientific development, see D. de S. Price, Science since Babylon (New Haven: Yale University Press, 1961).
9. Parenthetically, it might be noted that scientific developments may have one of two fundamentally different engineering effects. First, they may permit the scientifically knowledgeable engineer to accomplish something previously considered impossible. Second, they may permit something already possible to be accomplished more knowledgeably and efficiently.
Consider the second: In the absence of scientific knowledge, the engineer is forced to rely on experience. Thus, experience in a prior situation guides action in a current situation. However, different situations are rarely identical in all relevant respects, with the consequence that prior experience never provides a certain guide to current action. As a result, the engineer who must rely on experience is forced to take a fundamentally conservative stance. For example, in the absence of precise knowledge concerning the properties of materials to be used in the construction of a bridge, the cautious engineer "overbuilds," simply because he cannot anticipate the magnitude of the stresses which his materials will tolerate. In contrast, as noted by K. I. Pelc (seminar discussion), the scientifically knowledgeable engineer can "optimize," e.g., "precisely" tailor the construction of a bridge to the stresses to which is anticipated to be subjected. However, because the engineer's knowledge (of the situation, of future stresses, etc.) is never complete, to optimize rather than overbuild also implies an increase in the risk of failure. Thus, Roman aqueducts survive while contemporary bridges unexpectedly collapse.
With reference to the first effect of scientific knowledge, the capacity to accomplish something previously impossible, risks are inevitably encountered because of unanticipated concomitants of the intended outcome. On both counts, the scientization of engineering (the replacement of the conservative engineer by the scientific engineer) raises engineering risks. Of course, this increase in engineering risk is conjoined with, and can be considered part of the price of, the concomitant increase in engineering "power."
10. Thus, Derek Price, in Little Science, Big Science ..., argues that the primary objective (and function) of scientific publication is not to disseminate new knowledge but rather to claim credit for its discovery.
11. D. de S. Price, "An Extrinsic Value Theory for Basic and 'Applied' Research", in J. Haberer, ed., Science and Technology Policy (Lexigton, Mass.: Lexington Books, 1977).
12. Derek de Solla Price, "A Theoretical Basis for Input-Output Analysis of National R&D Policies," in Devendra Sahal, ed., Research, Development, and Technological Innovation (Lexington, Mass.: Lexington Books, 1980).
13. This more general formulation assumes that any given degree of faculty obsolescence can be sustained over time with a constant rate of research activity and that greater research activity is required to sustain a lesser degree of obsolescence, i.e., that it is easier (involves lesser research effort) to remain consistently further from the frontier.
14. If a corporation were to support fundamental research on purely ideational grounds, this would reflect the utility of its managers and would necessitate (a) the corporation's command of monopoly profits and/or (b) the insulation of the managers from the owners.
15. The Kaiser Wilhelm Gesellschaft, now renamed the Max Plack Gesellschaft (Society), provides an excellent example.
16. Thus, the Institute for Advanced Study was endowed by the mercantile wealth of Louis Bamberger and his sister, Mrs. Felix Fuld, while Rockefeller University was endowed by the industrial wealth of John D. Rockefeller. A more recent example is provided by Howard Hughes' endowment of the Howard Hughes Medical Institute.
17. Prestige-oriented national scientific institutions such as the Max Planck Gesellschaft, the Soviet Academy of Sciences and the National Science Foundation come readilly to mind.
18. Here a corporation might legitimately allocate resources to fundamental research on grounds of the "good will" generated thereby. Consider, for example, the good will derived by American Telephone and Telegraph Company from the prestige of Bell Labs.
19. For example, because the U.S. accounts for 40 percent of scientific "production," a doubling of U.S. output would increase the total scientific knowledge available in the U.S. by only 40 percent, implying an arc elasticity of at most 0.4. Further, this assumes that a doubling of inputs would result in a doubling of U.S. output, which, for reasons discussed, is probably unrealistic; thus, Dresch and Janson ("Giants, Pygmies ...") suggest an elasticity of scientific contributions (e.g., articles) with respect to resources (the size of the scientific cadre) of about 0.5, suggesting an elasticity of world output with respect to U.S. input of only about 0.2, i.e., a doubling of U.S. input would increase global output by only 20 percent. Focusing not on contributions but the scientific value of these contributions, Dresch and Janson indicate an elasticity with respect to resources of less than 0.4, suggesting at most a 10 percent increase in world scientific value as a result of a doubling of U.S. inputs.
Stephen P. Dresch, elected to the Michigan House of Representatives in 1990, was an unsuccessful 1992 Congressional primary candidate. A Ph.D. economist (Yale), he is former dean of Michigan Technological University's School of Business and Engineering Administration, research scholar at the International Institute for Applied Systems Analysis (Austria), chairman of the Institute for Demographic and Economic Studies, director of Yale University's program of Research in the Economics of Higher Education, and research associate of the National Bureau of Economic Research.