The Economics of Science and Technology Charles Wessner9

Home , Economics of science

David B. Audretsch1 Barry Bozeman2 Kathryn L. Combs3 Maryann Feldman4 Albert N. Link5 Donald S. Siegel6 Paula Stephan7 Gregory Tassey8 The Economics of Science and Technology Charles Wessner9

ABSTRACT. This paper provides a non-technical, accessible Science and technology policy are even more introduction to various topics in the burgeoning literature on important in the “new” economy, with its greater the economics of science and technology. This is an interdisci- emphasis on the role of intellectual property and plinary literature, drawing on the work of scholars in the fields knowledge transfer. Thus, it is unfortunate that of economics, public policy, sociology and management. The aim of this paper is to foster a deeper appreciation of the eco- most individuals rarely have the opportunity to nomic importance of science and technology issues. We also study this subject. As a result, the general pub- hope to stimulate additional research on these topics. lic poorly understands the antecedents and conse- JEL Classification: O3 quences of technological change. It is clear in the report that most Americans are not well-informed about public policy issues relating to science and technology. As shown in Table I, individuals rank science and technology policy issues relatively high in terms of interest, 1. Introduction yet noticeably lower in terms of their self-assessed Science and technology have long been regarded knowledge about the issues. However, it is pre- as important determinants of economic growth. cisely these issues that may be most critical in Edwin Mansfield (1971, pp. 1–2), a pioneer in the determining long-run economic growth. economics of technological change, noted: The purpose of this paper is to provide an overview of salient topics in the economics of Technological change is an important, if not the most important, factor responsible for economic growth science and technology. We devote considerable without question, [it] is one of the most important determi- attention to historical and institutional informa- nants of the shape and evolution of the American economy. tion concerning these issues, because we believe that an understanding of the current situation depends to a large extent on an understanding of how this literature and the institutions that sup- 1Indiana University port science and technology evolved.1 2Georgia Institute of Technology 3University of St. Thomas The remainder of this paper is organized as 4Johns Hopkins University follows. Section 2 provides some initial defini- 5University of North Carolina at Greensboro tions. In Section 3, we summarize major pub- Greensboro, NC 27402 lic policy initiatives toward science and technol- E-mail: [email protected] ogy in the United States from the colonial period 6University of Nottingham and Rensselaer Polytechnic Institute 7Georgia State University to the present. Our up-front emphasis on policy 8National Institute of Standards and Technology underscores the subtlety of partnerships involving 9National Academy of Sciences the public and private sectors that have emerged

Table I Section 7, tax incentives in Section 8, research col- Indices of public interest and self-assessed knowledge in laborations in Section 9, and public/private part- selected policy issues, 1999 nerships that subsidize research in Sections 10 and Public 11. These mechanisms, and their relationshipto interest Policy issues Informedness firm behavior, are summarized in Section 12. All of the policy mechanisms discussed in Sec- 82 New medical discoveries 53 tions 7 through 11 are designed to stimulate the 71 Environmental pollution 48 71 Local school issues 58 private sector’s demand for R&D resources. For 67 Issues about new scientific demand mechanisms to be effective there must discoveries 44 also be a supply response in terms of the edu- 65 Use of new inventions and cation and increased availability of scientists and technologies 43 engineers. This is the topic of Section 13. As with 65 Economic issues and business all public-sector initiatives, there is also an issue conditions 50 64 Military and defense policy 44 of public accountability. How effectively are public 55 Use of nuclear energy to research funds being spent? The history of public generate electricity 29 accountability and related evaluation methods are 53 International and foreign policy 40 described in Section 14. Section 15 constitutes our 51 Space exploration 37 summary statement regarding the current state of 47 Agricultural and farm issues 33 the field of economics of science and technology Source: Science and Engineering Indicators—2000, Appendix and future directions. Tables 8-2 and 8-5. over the last three centuries. These “public/private 2. Fundamental concepts partnerships” have evolved from government’s As in any new field—and we view the economics desire to steer private investment towards certain of science and technology as an emerging field that types of scientific activity and the development draws on concepts from numerous disciplines— and use of new technologies. Thus, the federal there are several fundamental concepts. Thus, we government has attempted to establish an environ- begin with several definitions. ment that is conducive for private sector invest- In everyday conversation, terms such as science ment in research and development (R&D), as well and technology, as well as invention and innova- as one in which the public and private sectors can tion, are often used interchangeably. However, for be partners in undertaking innovative activity. academics and policymakers there are important Section 4 emphasizes the role of technology in distinctions that give each of these terms a unique economic growth and sets the stage for under- meaning. standing the scope of science and technology pol- Science, in a broad sense, is the search for icy mechanisms used to maintain national growth. knowledge, and that search is based on observed Fundamental to all such policy instruments is facts and truths. Thus, science begins with known the relationshipamong investment in R&D, tech- starting conditions and searches for unknown nological advancement, and economic growth. end results (Nightingale, 1998). Technology is the Dimensions of R&D are described in Section 5. application of new knowledge learned through The following section introduces the second part science to some practical problem. Technological of the primer by emphasizing the entrepreneurial change is the rate at which new knowledge is dif- nature of firms, both to innovate and to respond to fused and put into use in the economy. policy initiatives. Section 6 advances an economic Closely related to science and technology are rationale for government’s role in the innovation the concepts of invention and innovation. Follow- process. An articulation of such a role has long ing Bozeman and Link (1983, p. 4): been absent from science and technology policy The concepts commonly used in connection with innovation debates. Having set forth this rationale, four pol- are deceptively simple. Invention is the creation of some- icy mechanisms are discussed from a historical as thing new. An invention becomes an innovation when it is well as an economic perspective: patent laws in put in use. The Economics of Science and Technology 157

Innovations may be new products, new processes, perception of the opportunity is the fundamental or new organizational methods that are novel and first step. The consequent step is the ability to act add value to economic activity. Thus, invention on that perception. What defines the entrepreneur parallels the concept of science and innovation is the ability to move technology forward into parallels the concept of technology. innovation. The technology may be discovered or It is useful to think of an innovation as some- developed by others. The entrepreneur is able to thing new that has been brought into use. Thus, recognize the commercial potential of the inven- this innovation represents, in a sense, a new tion and organize the capital, talent, and other underlying technology.2 Embedded in this distinc- resources that turn an invention into a commer- tion between invention and innovation is a process cially viable innovation. whereby inventions become applied. This pro- What are the requisite resources needed for cess is central to what we call entrepreneurship. action, which takes the perception of an opportu- Entrepreneurship is a process involving the organi- nity forward to result in an innovation? One obvi- zation of resources, and the output of that process ous answer is research and development (R&D), is an innovation.3 Of course, for entrepreneurship that is, the commitment of resources to inven- to have economic value the resultant output or tion and innovation. R&D not only provides a innovation must have economic value. stock of knowledge to encourage perception but From an economic perspective, the concept also the ability for the firm to foster action. How- of entrepreneurial innovation can be traced back ever, firms that do not conduct R&D can still be to the Physiocrats in France in the mid-1700s. entrepreneurial, as discussed above. In such firms, Baudeau (1910, p. 46) referred to a process innovations are likely to be introduced rather than guided by an active agent, which he called an produced. Such firms act in an entrepreneurial entrepreneur, within a capitalistic system4: manner by hiring creative individuals and providing them with an environment conducive for the Such is the goal of the grand productive enterprises: first blossoming of their talents. to increase the harvest by two, three, four, ten times if pos- Consider R&D-active firms. The R&D they sible; secondly to reduce the amount of labor employed and so reduce costs by a half, a third, a fourth, or a tenth, conduct serves two general purposes. First, it pro- whatever possible. vides the resource base from which the firm can respond to an opportunity with perceived strate- Embedded in this conceptualization of entrepre- gic merit or technical opportunity that allows the neurshipis the notion of an innovative process, firm to developa commercial market. Second, one perhaps as simple as the perception of new those scientists involved in R&D are the inter- technology adopted from others so as to increase nal resource that facilitates the firm’s being able agricultural yield, or one as refined as the actual to make decisions regarding the technical merits development of a new technology to do the same. of others’ innovations and how effectively those When the process is completed, and when the innovations will interface with the existing tech- innovation is put into use, there will be an increase nological environment of the firm. The firm may in productivity, and possibly, substitution of capi- choose to purchase or license this technology or tal for labor. undertake a new R&D endeavor. In this latter We have defined entrepreneurship as a process: sense, one important role of R&D is to enhance an output is the promotion of one’s own inno- the absorptive capacity of the firm.5 vation or the adoption of another’s innovation. Thus, the role of R&D in enhancing the The term entrepreneurship is commonly used to absorptive capacity of the firm goes beyond simply refer to a businessman or even to a risk taker. assessing the technical merits of potentially pur- We use the term entrepreneur in a much broader chasable technology. It allows the firm to interpret sense; an entrepreneur is one who perceives an the extant technical literature, to interface when opportunity and has the ability to act upon it. necessary with the research laboratories of oth- Hence, entrepreneurship is a process that involves ers, in a research partnership relationship, or to both perception and action. The perception of acquire technical explanations from, say, a federal the opportunity may be influenced by changes in laboratory or university laboratory; or simply to strategic directions or competitive markets, but solve internal technical problems. 158 Audretsch et al.

An understanding of these historical events Strategic direction of the firm and players is important because it allows one to Entrepreneurial response Innovation Value Added Competitive market understand the environment in which innovation conditions takes place and the genesis of the assumptions Figure 1. The entrepreneurial process: an initial look. that underlie the public policies that influence this environment. It also illustrates the evolving role Figure 1 provides an initial view of what we that the government has taken in promoting sci- term the entrepreneurial process. This initial anal- ence and technology. ysis will be expanded upon, but here it introduces the following concepts: The colonial period • The organization, typically a private firm, has a focus that results from its agreed upon strate- The first member of the Royal Society of London gic direction. This strategic direction, coupled to immigrate to the Massachusetts Bay Colony with competitive market conditions, generates was John Winthrop, Jr. in 1631, just a few years an entrepreneurial response. after the founding of the Colony. As a scientist, he is credited with establishing druggist shops and • The purposive activity associated with the entrepreneurial response leads to an innova- chemistry laboratories in the surrounding villages tion. to meet the demand for medicine. According to UNESCO (1968, p. 9), these ventures were “per- • There are market forces at work that are, in part, beyond the influence of the firm and haps the first science based commercial enterprise these forces determine the economic value of of the New World.” the innovation and hence the value added to Before the turn of the eighteenth century, the project as well as to the user of the inno- colonists made noticeable advances toward what vation. may be called a scientific society, organizing scientists who came from England and other Euro- There is a subtle distinction between entrepre- pean countries into communities that promoted neurship or the innovation process and the process scientific inquiry. In 1683, the Boston Philosoph- of science and discovery. As noted above, science ical Society was formed to advance knowledge in moves from starting conditions toward unknown philosophy and natural history. Benjamin Franklin results whereas the innovation process starts with formed the American Philosophical Society of an anticipated intended result and moves toward Philadelphia in 1742 to encourage correspon- the unknown starting conditions that will pro- dence with colonists in all areas of science. It duce it. later merged with the Franklin-created American Society to promote what Franklin called “useful knowledge,” and it still exists today. This 6 3. Historical background combined Society focused on making available The history of science, technology, and economic advancements in agriculture and medicine to all growth in the United States was greatly influenced individuals by sponsoring the first medical school by the scientific discoveries and university infras- in America (also supported by the Pennsylva- tructure within Europe at the time of colonization. nia House of Representatives). Thus, Franklin’s While difficult to pinpoint how or which specific Society was a hallmark of how public and private elements of scientific and technical knowledge dif- sector interests could work together for the com- fused across the Atlantic, certain milestone events mon weal.7 can be dated and pivotal individuals can be sin- Influenced by the actions of Pennsylvania and gled out. This background gives us an appreciation later Massachusetts with regard to sponsorship of for the role that science and technology resources scientific institutions, the establishment of national played in the developing American nation and universities for the promotion of science was contributed to shaping the preeminent role that first discussed at the Constitutional Convention the former colony achieved. in 1787. However, at that time the founders of The Economics of Science and Technology 159 the Constitution believed educational and scien- were directed toward manufacturing and trans- tific activities should be independent of direct portation. In fact, the Secretary of the Treasury— national governmental control. But, they felt that the Department of the Treasury being the most the national government should remain an influ- structured executive department at that time— ential force exerting its influence through indirect directly funded the Franklin Institute in Philadel- rather than direct means. For example, Article I, phia to investigate the causes of these problems. Section 8, of the Constitution permits the enact- This action, driven by public concern as well as ment of patent law: the need to developnew technical knowledge, was the first instance of the government sponsoring The Congress shall have the power To promote the research in a private-sector organization. progress of science and useful arts, by securing for limited In 1838, the federal government again took a times to authors and inventors the exclusive right to their respective writings and discoveries.8 lead in the sponsorship of a technological innovation that had public benefits. After Samuel Morse However, Thomas Jefferson championed a demonstrated the feasibility of the electric tele- more direct role for the government in the area graph, Congress provided him with $30,000 to of science. While president, Jefferson sponsored build an experimental line between Baltimore, the Lewis and Clark expedition in 1803 to advance Maryland, and Washington, DC. This venture was the geographic knowledge of the nation, thus mak- the first instance of governmental support to a pri- ing clear that “the promotion of the general wel- vate researcher.9 fare depended heavily upon advances in scientific Public/private research relationships continued knowledge” (UNESCO, 1968, p. 11). In fact, this to evolve in frequency and in scope. In 1829, action by Jefferson set several important prece- James Smithson, gifted $500,000 to the United dents including the provision of federal funds to States to found an institution in Washington, individuals for scientific endeavors. DC for the purpose of “increasing and diffusing Although the Constitution did not set forth knowledge among men” (UNESCO, 1968, p. 12). mechanisms for establishing national academic Using the Smithson gift as seed funding, Congress institutions based on the founders’ belief that the chartered the Smithsonian Institution in 1846, and government should have only an indirect influence Joseph Henry became its first Executive Officer. on science and technical advancement, the need Henry, a renowned experimental physicist, contin- for a national institution related to science and ued the precedence of a federal agency directly technology was recognized soon after the Revolu- supporting research through grants to individ- tionary War. For example, West Point was founded ual investigators to pursue fundamental research. in 1802 as the first national institution of a sci- Also, the Institution represented a base for exter- entific and technical nature, although Connecticut nal support of scientific and engineering research established the first State Academy of Arts and since, during the 1850s, about 100 academic institutions were established with science and engi- Sciences in 1799. neering emphases. In the early 1800s, universities began to empha- Thus, the pendulum had made one complete size science and technical studies, and in 1824 swing in the hundred years since the signing of the Rensselaer Polytechnic Institute was founded in Constitution. In the early years, the government New York State to emphasize the application of viewed itself as having no more than an indirect science and technology. The American Journal of influence on the development of science and tech- Science was the first American scientific publica- nology, but over time its role changed from indi- tion, followed in 1826 by the American Mechanics rect to direct. This change was justified in large Magazine. part because advances in science and technology The social importance of the government hav- are viewed as promoting the public interest. ing a direct role in the creation and application of technical knowledge was emphatically demon- Toward a national infrastructure strated in the 1820s and 1830s through its support of efforts to control the cholera epidemic of 1822. Scientists had long looked toward the European Also during that time period federal initiatives universities for training in the sciences, but now an 160 Audretsch et al. academic infrastructure was beginning to develop in particular. What they experienced firsthand in the United States. Harvard University awarded were the strong ties between European industries its first bachelor of science degree in 1850. The and graduate institutions. European companies development of an academic science base and the invested in professors and in their graduate stu- birth of technology-based industries (e.g., the elec- dents by providing them with funds and access trical industry) in the late 1850s established what to expensive materials and instruments. In return, would become the foundation for America’s tech- the firms gained lead-time toward new discover- nological preeminence. ies, as well as early access to the brightest grad- In 1863, during the Civil War, Congress estab- uate students as soon as they completed their lished the National Academy of Sciences. The fed- studies.10 This form of symbiotic arrangement eral government funded the Academy but not the became the norm for the European-trained scien- members affiliated with it who had “an obligation tists who were working in U.S. industries and U.S. to investigate, examine, experiment, and report universities. upon any subject of science or art in response to a By the turn of the century, it was widely request from any department of the Government” accepted among industrial leaders that scientific (UNESCO, 1968, p. 14). Then, as today, the knowledge was the basis for engineering develop- Academy is independent of governmental control. ment and was the key to remaining competitive. The Morrill Act of 1862 established the land Accordingly, industrial research laboratories soon grant college system thereby formally recognizing began to blossom as companies realized their need the importance of trained individuals in the agri- to foster scientific knowledge outside of the uni- cultural sciences. The Act charged each state to versity setting.11 There are a number of examples establish at least one college in the agricultural of this strategy. and mechanical sciences. Each state was given General Electric (GE) established the General 30,000 acres of federal land per each elected U.S. Electric Research Laboratory in 1900 in response Senator and Representative. An important out- to competitive fears that improved gas lighting growth of this land grant system was a mechanism would adversely affect the electric light business, or infrastructure through which state and federal and that other electric companies would threaten governments could financially support academic GE’s market share as soon as the Edison patents research interests. expired. Similarly, AT&T was facing increasing Although the federal government was encour- competition from radio technology at the same aging an infrastructure to support science and time. In response, AT&T established Bell Labo- technical research, it did not have a so-called in- ratories to research new technology in the event house staff of permanent professionals who were that wire communications were ever challenged. competent to identify either areas of national And as a final example, Kodak realized at the importance or areas of importance to specific agencies. In 1884, Congress established the Alli- turn of the century that it must diversify from syn- son Commission to consider this specific issue. thetic dyes. For a number of years Kodak relied on While many solutions were debated, including the German chemical technology, but when that tech- establishment of a Department of Science—an nology began to spill over into other areas such idea that resurfaces every few decades—the Com- as photographic chemicals and film, Kodak real- mission soon disbanded without making any rec- ized that their competitive long-term health rested ommendations much less reaching closure on the on their staying ahead of their rivals. Kodak too matter. One could conclude from the inaction of formed an in-house research laboratory. the Commission that it favored the decentralized Many smaller firms also realized the compet- administrative architecture that had evolved over itive threats that they could potentially face as a time as opposed to a centralized one. result of technological competition, but because of their size they could not afford an in-house facility. So as a market response, contract research labo- Toward an industrial infrastructure ratories began to form. Arthur D. Little was one Most scientists in the United States in the 1870s such contract research laboratory that specialized and 1880s had been trained in Europe, Germany in the area of chemicals. The Economics of Science and Technology 161

Just as industrial laboratories were growing address governmental needs, the Science Advi- and being perceived by those in both the pub- sory Board was multi-field and organized around lic and private sectors as vitally important to the impending national problems. The National Plan- economic health of the nation, private founda- ning Board was formed on the presumption tions also began to grow and to support university that there were areas of economic concern that researchers. For example, the Carnegie Institution required a national perspective rather than a of Washington was established in 1902, the Rus- field-of-science perspective. In 1934, the National sell Sage Foundation in 1907, and the Rockefeller Resources Committee replaced the National Plan- Foundation in 1913. ning Board, and it then subsumed the Science In the early-1900s science and technology Advisory Board. The bottom line was, after all of began to be embraced—both in concept and in the organizational issues were settled, that the fed- practice—by the private sector as the founda- eral government recognized through the formation for long-term competitive survival and gen- tion of these committees and boards that it had eral economic growth. and would continue to have an important coordinating role to play in the science and technology planning toward a national goal of economic well World War I and the years that followed being. Hence, the pendulum began to swing away Increased pressure on the pace of scientific from government having a hands-on role toward it and technical advancements came at the begin- having an indirect influence on planning the envi- ning of World War I. The United States had ronment for science and technology. been cut off from its European research base. In 1938, the Science Committee of the National Congress, in response, established the Council of Resources Committee issued a multi-volume National Defense in 1916 to identify domestic report entitled, Research—A National Resource. pockets of scientific and technical excellence. The Some important first principles were articulated National Academy of Sciences recommended to in that report. These principles have since then President Woodrow Wilson the formation of the formed a basis for economists and policy makers National Research Council to coordinate cooper- to rationalize/justify the role of government in sci- ation between the government, industry, and the ence and technology. The report is explicit that: academic communities toward common national • There are certain fields of science and tech- goals.12 The prosperity of the post-World War I nology, which the government has a Consti- decade also created an atmosphere supportive of tutional responsibility to support. These fields the continued support of science and technology. include defense, determination of standards, In 1920, there were about 300 industrial research and certain regulatory functions. laboratories, and by 1930 there were more than • The government is better equipped to carry 1,600.13 Of the estimated 46,000 practicing scien- on research in certain fields of science than tists in 1930, about half were at universities and the private sector. These are areas where over a third were in industry. Herbert Hoover was “research is unusually costly in proportion to Secretary of Commerce at this time. He adopted its monetary return but is of high practical the philosophy that (UNESCO, 1968, p. 18): or social value” (p. 25). Examples cited in pure and applied scientific research constitute a foun- the report include aeronautical and geological dation and instrument for the creation of growth and effi- research. ciency of the economy. • Research by the government “serves to stimu- In response to the Great Depression and the sub- late and to catalyze scientific activity by non- sequent national economic crisis, two important governmental agencies. In many fields, new events occurred in 1933. One was the appoint- lines of research are expensive and returns ment of a Science Advisory Board and the other may be small or long delayed. Industry can- was the establishment of a National Planning not afford to enter such fields unless there is Board. Whereas the National Research Council reasonable prospect of definite financial gain had been organized around fields of science to within a predictable future, and it is under such 162 Audretsch et al.

circumstances that the Government may lead The Office of Scientific Research and Development, of the way ” (p. 26). One example cited was which you are the Director, represents a unique experi- the Navy Department’s influence on the devel- ment of team-work and cooperation in coordinating sci- opment of the steel industry. entific research and in applying existing scientific knowledge to the solution of the technical problems paramount in war. There is no reason why the lessons to be found in this experiment cannot be profitably employed in times of peace. This information, the techniques, and the World War II and the years that followed research experience developed by the Office of Scientific Research and Development and by the thousands of scien- The involvement of the United States in World tists in the universities and in private industry, should be War II had a dramatic impact on the scope and used in the days of peace ahead for the improvement of direction of government’s support of science and the national health, the creation of new enterprises bring- technology. Prior to the war, there were about ing new jobs, and the betterment of the national standard of living. New frontiers of the mind are before us, and 92,000 scientists, with about 20 percent in govern- if they are pioneered with the same vision, boldness, and ment and the remaining 80 percent being almost drive with which we have waged this war we can create a equally divided between universities and the more fuller and more fruitful employment and a fuller and more than 2,200 industrial laboratories. Clearly, the fruitful life. United States had a significant scientific resource Shortly before asking Bush to prepare this base to draw upon for it war efforts. report, Senator Kilgore from West Virginia had In 1940, President Roosevelt established the introduced a bill to create a National Science National Defense Research Committee and asked Foundation. The Kilgore bill recommended giving Vannevar Bush, President of Carnegie Institution authority to federal laboratories to allocate public of Washington, to be its chairman. The purpose moneys in support of science to other government of this committee was to organize scientific and agencies and to universities. Clearly, this recom- technological resources toward enhancing national mendation gave a direct role to government in defense. It soon became apparent that this task shaping the technological course of the country required an alternative administrative structure. not only in terms of scientific direction but also In 1941, Roosevelt issued an Executive Order in terms of what groups would conduct the under- establishing the Office of Scientific Research and lying research. The bill was postponed until after Development (OSRD) with Bush as Director. The the war. OSRD did not conduct research; rather it real- Bush submitted his report, Science—The End- ized that there were pockets of scientific and less Frontier, to President Roosevelt on July 25, technological excellence throughout the country, 1945. In Bush’s transmittal letter to the president and through contractual relationships with uni- he stated: versities and industry and government agencies The pioneer spirit is still vigorous within this Nation. Sci- it could harness national strengths with a focus ence offers a largely unexplored hinterland for the pioneer on ending the war. One hallmark event from the who has the tools for his task. The reward of such explo- efforts of the OSRD was the establishment of ration both for the Nation and the individual are great. Scientific progress is one essential key to our security as the Los Alamos Laboratory in New Mexico under a nation, to our better health, to more jobs, to a higher the management of the University of California. standard of living, and to our cultural progress. What came about from the collective efforts of The foundations set forth in Science—The End- the resources acquired by the Office were not only less Frontier are: atomic weapons but also radar. By 1944, it was clear that World War II was • “Progress depends upon a flow of new sci- almost over. President Roosevelt then asked Bush entific knowledge” (p. 5). to developrecommendations as to how scien- • “Basic research leads to new knowledge.14 It tific advancements could contribute in the larger provides scientific capital. New products sense to the advancement of national welfare. In and new processes do not appear full-grown. his November 17, 1944 letter to Bush, President They are founded on new principles and new Roosevelt stated: conceptions, which in turn are painstakingly The Economics of Science and Technology 163

developed by research in the purest realms of academic researchers thought about the process science” (p. 11). of creating new technology. The so-called linear • “The responsibility for the creation of new sci- model set forth by Bush is often represented by: entific knowledge rests on that small body of men and women who understand the fun- Basic Research → Applied Research damental laws of nature and are skilled in the → Development techniques of scientific research” (p. 7). → Enhanced Production • “A nation which depends upon others for its → Economic Growth new basic scientific knowledge will be slow in its industrial progress and weak in its compet- Complementing Science—The Endless Frontier itive position in world trade, regardless of its was a second, and often overlooked, report pre- mechanical skill” (p. 15). pared in 1947 by John Steelman, then Chair- man of the President’s Scientific Research Board. • “The Government should accept new responsibilities for promoting the flow of new scientific As directed by an Executive Order from Pres- knowledge and the development of scientific ident Truman, Steelman, in Science and Public talent in our youth” (p. 7). Policy, made recommendations on what the fed- • “If the colleges, universities, and research eral government could do to meet the challenge institutes are to meet the rapidly increas- of science and assure the maximum benefits to ing demands of industry and Government for the Nation. Steelman recommended that national new scientific knowledge, their basic research R&D expenditures should increase as rapidly as should be strengthened by use of public funds” possible, citing (p. 13): (p. 16). • “Therefore I recommend that a new agency for 1. Need for Basic Research. Much of the world is these purposes be established” (p. 8). in chaos. We can no longer rely as we once did upon the basic discoveries of Europe. At the Bush recommended in his report the creation same time, our stockpile of unexploited funda- of a National Research Foundation. Its proposed mental knowledge is virtually exhausted in cru- purposes were to: cial areas. 2. Prosperity. This Nation is committed to a pol- develop and promote a national policy for scientific research and scientific education, support basic icy of maintaining full employment and full pro- research in nonprofit organizations, developscientific duction. Most of our frontiers have disappeared talent in American youth by means of scholarships and and our economy can expand only with more fellowships, and contract and otherwise support long- intensive development of our present resources. range research on military matters. Such expansion is unattainable without a stim- Bush envisioned a National Research Foun- ulated and growing research and development dation that would provide funds to institutions program. outside government for the conduct of research. 3. International Progress. The economic health Thus, this organization differed from Kilgore’s of the world—and the political health of the proposed National Science Foundation in that world—are both intimately associated with our Bush advocated an indirect role for government. own economic health. By strengthening our There was agreement throughout government that economy through research and development we an institutional framework for science was needed, increase the chances for international economic but the nature and emphases of that framework well-being. would be debated for yet another five years.15 4. Increasing Cost of Discovery. The frontiers of Science—The Endless Frontier affected the sci- scientific knowledge have been swept so far entific and technological enterprise of this nation back that the mere continuation of pre-war in at least two ways. It laid the basis for what growth, even in stable dollars, could not possi- was to become the National Science Foundation bly permit adequate exploration. This requires in 1950. Also, it set forth a paradigm that would more time, more men, more equipment than over time influence the way that policy makers and ever before in industry. 164 Audretsch et al.

5. National Security. The unsettled international During the Reagan administration, expenditures situation requires that our military research on defense R&D increased dramatically as part of and development expenditures be maintained his Star Wars system. President Bush (no relation- at a high level for the immediate future. Such shipto Vannevar Bush) set forth this nation’s first expenditures may be expected to decrease in technology policy (see below) and increased the time, but they will have to remain large for sev- scope of the National Institute of Standards and eral years, at least. Technology (NIST, see below). President Clinton established important links between science and An important element of the Steelman report technology policy, championing programs to trans- was the recommended creation of a National Sci- fer public technology to the private sector. ence Foundation, similar in focus to the National Research Foundation outlined by Bush. And, Congress passed the National Science Foundation 4. Economic growth and technological change Act in 1950. In the previous section, we described Vannevar Renewed post-war attention toward science and Bush’s paradigm of research and development technology came with the success of the Soviet leading to economic growth. As a practical mat- Union’s space program and the orbit of its Sputnik ter, economic growth is generally defined at the I in October 1957. In response, President Eisen- macroeconomic level in terms of Gross Domes- hower championed a number of committees and tic Product (GDP).17 Figure 2 shows GDP-related agencies to ensure that the United States could growth for a number of countries. GDP per capita soon be at the forefront of this new frontier. Note- is greater in the United States than in any of the worthy was the National Defense Education Act other industrial nation shown in the figure. of 1958, which authorized $1 billion in federal An inspection of Figure 2 raises a number of moneys for support of science, mathematics, and questions, two of which are: Why do economies technology graduate education. This proposal is grow? and, Why has the U.S. economy outper- precisely the type of support that Bush recom- formed that of other industrial nations even after mended in his report. controlling for national size? As the post-World War II period came to a close, there was a well-established national and industrial infrastructure to support the advance- Theories of economic growth ment of science and technology. But, more The early literature on economic growth is formu- important than the infrastructure, there was an lated analytically using what economists refer to imbedded belief that scientific and technologi- as a production function. Simply put, a produc- cal advancements are fundamental for economic tion function represents the relationship between growth, and that the government has an impor- the output of an economic unit (a firm, industry, tant supporting role—both direct and indirect—to or economy) and the factors of production—or ensure such growth. inputs or resources—used to produce that output. Every president since Eisenhower has initiated major science policy initiatives.16 Kennedy set the 35 goal of sending a man to the moon by the end 30 of the 1960s and funded the needed programs 25 20 to make this a reality. Johnson emphasized the 15 use of scientific knowledge to solve social prob- Real GDP 10 lems through, for example, his War on Poverty. 5 0 Nixon dramatically increased federal funding for 1960 1965 1970 1975 1980 1985 1990 1995 2000 biomedical research as part of his War on Cancer. Year Ford created the Office of Science and Technol- U.S. Japan France Germany U.K. ogy Policy (OSTP) within the Executive Branch. Figure 2. Real GDP per capita for selected countries Carter initiated research programs for renewable 1960–1999 (U.S. $1000). Source: Science and Engineering energy sources such as solar energy and fission. Indicators—2000, Appendix Table 7-3. The Economics of Science and Technology 165

If output, Q, can be defined in terms of the be attributable to something else.18 Solow specu- two most basic factors of production, the stock of lated that what was captured in his residual calcu- capital (plant and equipment), K, and the stock of lation may reflect technology advance over time. labor, L, then a basic production function can be Changes in A from Equation (3) measure what is written as: called total factor productivity growth, or techno- Q = f K L (1) logical advancement. Other researchers, using alternative frame- Equation (1) denotes that a firm’s, industry’s, works, reached similar conclusions. Abramovitz or economy’s output will change in response to (1956), for example, referred to the unexplained changes in either the quantity or quality of capi- portion of growth more cautiously as a measure tal or the quantity or quality of labor. However, of our “ignorance.” This implied that while to account for other influences on output such economists were able to calculate unexplained as new technology, the production relationship growth, they were unable to provide a con- in Equation (1) can be modified most simply to clusive explanation for what caused improve- include a catch-all variable, A, as: ments in economic performance. Unlike Solow, Q = AF K L (2) Abramovitz speculated in some detail that growth not attributable to capital and labor was likely where, to be more specific, A is a shift factor to due to improvements in education and increases account for exogenous technological factors (as in research and development activity (R&D). opposed to conventional factor inputs such as K The academic literature is replete with theo- and L) that affect production, and where, because ries to explain growth over time. The so-called of the inclusion of A in Equation (2), FKL is “old growth” theory literature (Nelson and Phelps, distinct from fKL in Equation (1). 1966) is based on more sophisticated versions of By dividing both sides of Equation (2) by the Equation (1). That is, this literature emphasizes combination of K and L inputs (i.e., by total fac- additional inputs aside from K and L such as tors) denoted by FKL, variable A can be inter- investments in R&D and education. As well, it preted as an index of output per unit of all inputs emphasizes the greater specificity by which inputs or of total factor productivity. are measured including consideration for the het- A = Q/F K L (3) erogeneity of K and L (e.g., new vintages of K Early on, Robert Solow (1957), who was sub- embody others’ technological investments). sequently awarded the Nobel Prize in economics, The so-called “new growth” theory (Romer, estimated a variation of Equation (3) using aggre- 1986, 1994) emphasizes the influence of other fac- gate U.S. data. He calculated changes in the tors on growth that are not directly specified in an expanded version of Equation (1). These fac- value of A between 1909 and 1949. His analysis tors include, for example, technologies or efficien- showed that more than 87 percent of the growth cies that spillover into a firm’s production function in the U.S. economy could not be explained by either from other firms or from general advances the growth in capital and labor, and hence the in the economy (such as information technology) residual or unexpected portion of growth must or that spillover into a nation’s production function from trade policies. New growth theory is also 110 based on careful, explicit analytical modeling of 100 the incentives of agents to invest in new technol- 90 ogy. Figure 4 expands upon Figure 1 to incorpo- 80 TFP rate these ideas. In particular, two new elements 70 are included in Figure 4 that were not in Figure 1. 60 First, in-house or private investment in R&D is 50 1948 1953 1958 1963 1968 1973 1978 1983 1988 1993 1998 shown to have an influence on how the firm Year responds to the interaction of its strategic direc- Figure 3. Private nonfarm TFP, 1948–1997. Source: U.S. tion and its competitive environment. Second, we Bureau of Labor Statistics. have capsulated the essence of new growth theory 166 Audretsch et al.

There are also other important aspects of Strategic direction of the firm R&D. One such dimension is related to the size Entrepreneurial response Innovation Value Added Competitive market of firms that conduct R&D, a second to the geo- conditions In-house R&D External influences on graphical distribution of R&D, and a third to the economic performance relationshipbetween R&D and total factor productivity growth. We now consider each of these Figure 4. The entrepreneurial process: a second look. dimensions in turn. by simply acknowledging that external influences affect firm performance directly as well as indi- Sources of funding of R&D rectly through innovation. Regardless of whether one adheres to the more The toprow of Table II shows the sources for the narrow old theories or the broader new theo- $247.0 billion of R&D expenditures in the United ries, the evidence is overwhelming that technology States in 1999. Industry accounted for nearly 69 drives economic growth. There was renewed inter- percent of those expenditures; the federal govern- est in promoting economic growth in the postwar ment another 27 percent; and all other sources, aftermath of the destruction of the industrial base including state and local governments, universi- of many nations. Thus, it is not surprising that ties and colleges, and other nonprofit institutions, greater attention was devoted to the analysis of about 5 percent. R&D, which was hypothesized early on to be an The primacy of industry in funding R&D has important determinant of economic growth. not always held, as shown in Figure 5. In the aftermath of World War II upthrough the early 1980s, the federal government was the leading 5. Dimensions of R&D provider of R&D funds in the Nation. Although For purposes of measurement, there are three fun- a federal R&D presence existed before then, dur- damental dimensions of R&D. The first relates ing the war the federal government dramatically to the source of funding of R&D (who finances expanded its R&D effort by establishing a network R&D), the second to the performance of R&D of federal laboratories, including atomic weapons (who actually does the work), and the third to laboratories. It was at that time that the federal the character of use of R&D (whether the work government also greatly increased its support to is basic research, applied research, or develop- extramural R&D performers, especially to a select ment). These three fundamental dimensions are groupof universities and large industrial firms. not mutually exclusive. After the war (along with the widespread influence

Table II National R&D expenditures, by performer and funding source: 1999 ($millions)

Source of R&D funds

Other Percent Federal Universities Non-federal nonproﬁt distribution, Performer government Industry and colleges government institutions Total by performer

Total R&D 65853 169312 5838 2085 3912 247000 100 0% Federal government 17362 — — — — 17362 7 0% Industry 19937 165955 — — — 185892 75 3% Industry FFRDCs 2166 — — — — 2166 0 9% Universities and colleges 16137 2163 5838 2085 2032 28255 11 4% University FFRDCs 6169 — — — — 6169 2 5% Other nonproﬁt institutions 3246 1194 — — 1880 6320 2 6% Nonproﬁt FFRDCs 836 — — — — 836 0 3% Percent distribution, by source 26.7% 68.5% 2.4% 0.8% 1.6% 100%

Source: National Science Foundation. The Economics of Science and Technology 167

250 slow growth from the industrial sector, the fed- 200 eral government and industry accounted for about

150 equal shares by the early 1980s.

$R&D 100 Since then the federal government’s share of

50 R&D decreased to about 40 percent of total

0 in 1990 to its current share, somewhat below 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 30 percent. Initially, the decreasing federal share Year came about even though federal dollar support Total Federal Industry for R&D—in absolute terms—was increasing. Figure 5. U.S. R&D funding by source, 1953–1998 (bil- Between 1980 and 1987, federal R&D rose about lions $1992). Source: Science and Engineering Indicators—2000, 40 percent after adjusting for inflation. Most of Appendix Table 2-6. this growth, however, was in support of defense activities so that by 1987, the defense R&D share of the Bush and Steelman reports), federal R&D had grown to two-thirds of the federal R&D total support continued to expand for both defense and (its highest share since 1963). After the break-up non-defense purposes, including health R&D in of the Soviet Union, the imperative for continual the National Institutes of Health and—after the growth in federal defense R&D support was not establishment of the National Science Founda- as strong and the federal R&D total once again tion in 1950—a broad portfolio of fundamental slowed (and even fell in constant dollars). research activities. As a result of a post-Sputnik In terms of which agencies provide the R&D national commitment to catch upto the Soviet funds, federal sources are highly concentrated space successes, federal support for space-related among just a few agencies. According to the latest R&D mushroomed in the late 1950s and early data provided by the agencies themselves, of the 1960s. By 1960, the federal government accounted $74 billion obligated for R&D and R&D plant in for 65 percent of the nation’s total investment fiscal year 1998, just five accounted for 94 per- (80 percent of which was for defense), and indus- cent of all funds: Department of Defense (48%), try accounted for 32 percent of the total. Department of Health and Human Services, pri- Over the next twenty years the federal govern- marily the National Institutes of Health (19%), ment continued to be the leading source of R&D National Aeronautics and Space Administration (13%), Department of Energy (9%), and National funding, although the direction shifted over time. Science Foundation (3%). In the early 1960s, the relative defense share of Concurrent with recent reductions in federal federal R&D funding dropped precipitously from R&D spending, major changes have also occurred 80 percent in 1960 to about 50 percent in 1965, in industrial R&D spending patterns. After lack- where it fluctuated narrowly until 1980. Early on, luster funding in the early 1990s (reflecting the R&D for space exploration was the primary non- impact of mild recessions on its R&D activities) defense recipient of federal R&D funding. Indeed industry R&D support has grown rapidly since more than three-fourths of federal non-defense 1994 and now accounts for almost 70 percent of R&D funds were in support of NASA’s mission the national R&D total. As a result, and compared activities by 1965. By 1970, however, after the suc- with the funding patterns of the mid-1960s, indus- cess of several lunar landings, support for other try and government have reversed positions. non-defense purposes began to claim an increasingly larger share of the federal R&D totals, and continued to do so throughout the 1970s; notably R&D performers growth in federal energy R&D occurred as a R&D is performed in what has been termed the response to the several oil embargoes. Also by U.S. national innovation system. The system is, 1970, R&D support from industry was on the rise, according to Crow and Bozeman (1998, p. 42): and it accounted for just over 40 percent of the the complex network of agents, policies, and institu- total national R&D effort. As a result of relatively tions supporting the process of technical advance in an flat federal funding in the 1970s and continual economy. 168 Audretsch et al.

The performers of R&D within the system are States. Of the nearly $186 billion in R&D per- research laboratories. The laboratory performers formed by industry in 1998 (the latest year for of R&D correspond to the sectors that finance which the foreign-performed data are available), R&D, but not all R&D funded by a sector is about $16 million, or about 9 percent, was con- performed in that sector. For example, industry ducted in other countries. Foreign investments performed approximately $186 billion of R&D in in R&D are not unique to U.S. firms; the out- 1999, of which $166 billion came from industry flow of U.S. industrial R&D into other countries itself. The additional amount of R&D performed is approximately offset by an inflow of others’ by industry came from the federal government. R&D to be performed in the United States. Almost one-third of the R&D funded by the Most (68 percent) of U.S.-funded R&D abroad federal government is performed in industry, and was performed in Europe—primarily in Germany, more than one-half of those dollars are spent in the United Kingdom, and France. The current the aircraft, missiles, and transportation equip- European share of U.S. industry’s offshore R&D ment industries. Universities and colleges fund activity, however, is somewhat less than the 75 per- only about 20 percent of the R&D they perform. cent share reported for 1982 (peak year). Overall, Fifty-seven percent of the R&D performed in U.S. R&D investments abroad have generally universities and colleges comes from the federal shifted away from the larger European coun- government and the rest equally from industry, tries and Canada, and toward Japan, several of nonprofit institutions, and nonfederal government the smaller European countries (notably Sweden sources. and the Netherlands), Australia, and Brazil. Phar- As shown in Figure 5, since the late-1980s maceutical companies accounted for the largest the federal government has decreased its fund- industry share (18 percent of U.S. 1997 over- ing of national R&D. The lion’s share of that seas R&D), which was equivalent to 21 per- decrease has come in the form of federal alloca- cent of their domestically-financed R&D. Much tions for R&D performed in industry, for which of this pharmaceutical R&D took place in the the R&D level of support displays a somewhat United Kingdom. roller-coaster-like pattern. The latest peak in fed- Foreign firms in the United States make sub- eral support for industrial R&D was a result of stantial R&D investments. From 1987 to 1996, major defense-related funding increases for Presi- inflation-adjusted R&D growth from majority- dent Reagan’s Strategic Defense Initiative prior to owned affiliates of foreign firms averaged 10.9 per- the collapse of the Soviet Union. By contrast, fed- cent per year, and are now roughly equivalent to eral funding to universities and colleges, adjusted U.S. companies’ R&D investment abroad. Affili- for inflation, has increased slightly each year since ates of firms domiciled in Germany, Switzerland, at least the late 1970s. the United Kingdom, France, and Japan collec- There are other important dimensions to the tively account for 72 percent of this foreign fund- performance of industrial R&D. About three- ing. Foreign-funded R&D in the United States fourths of industrial R&D is performed in manu- in 1996 was concentrated in drugs and medicines facturing industries. The dominant manufacturing (mostly from Swiss, German, and British firms), industries in terms of dollars of R&D performed industrial chemicals (funded predominantly by are chemicals and allied products, electrical equip- German and Dutch firms), and electrical equipment (including computers), and transportation equipment. The remaining one-fourth is per- ment (one-third of which came from French formed in the non-manufacturing sector, including affiliates). services. Computer-related services are the leaders therein. The steepgrowth in R&D performed R&D by character of use in the services is a relatively recent phenomenon. As recently as 15 years ago, manufacturers still Vannevar Bush is credited for first using the accounted for more than 90 percent of the indus- term “basic research,” which he defined to mean trial R&D total. research performed without thought of practical Also, not all industry-performed R&D occurs ends in his 1945 report to President Roosevelt, within the geographical boundaries of the United Science—The Endless Frontier. Since that time, The Economics of Science and Technology 169 policy makers have been concerned about defini- and performance of these R&D component cate- tions that appropriately characterize the various gories. Applied research and development activ- aspects of scientific inquiry that broadly fall under ities are primarily funded by industry and per- the label of R&D and that relate to the linear formed by industry. Basic research, however, is model that Bush proffered. primarily funded by the federal government and Definitions are important to the National Sci- generally performed in universities and colleges. ence Foundation because it collects expenditure The decline in federal support of R&D over the data on R&D. For those data to accurately reflect past decade has primarily come at the expense of industrial and academic investments in technolog- applied research and development performed in ical advancement, and for those data to be com- industry. parable over time, there must be a consistent set of reporting definitions. The classification scheme used by the National R&D activity in large and small firms Science Foundation for reporting purposes was Table III shows the level of R&D expendi- developed for its first industrial survey in 1953– tures from all sources for several of the larger 1954.19 While minor definitional changes were R&D-performing firms in the United States made in the early years, namely to modify the in 1997. These data come from a variety of pub- category originally referred to as “basic or funda- lic sources: several patterns can be seen from the mental research” to simply “basic research,” the data in the table: concepts of basic research, applied research, and • Microsoft (the tenth-largest R&D-active com- development have remained much as was implic- pany) invests about one-fifth of the amount itly contained in Bush’s 1945 linear model. of R&D invested by General Motors (the The objective of basic research is to gain more largest R&D-active company). Thus, even comprehensive knowledge or understanding of the among the R&D giants, R&D expenditures subject under study, without specific applications vary dramatically. in mind. Basic research is defined as research that advances scientific knowledge but does not Table III have specific immediate commercial objectives, Largest R&D-active U.S. companies although it may be in fields of present or potential commercial interest. Much of the scientific Rank $R&D $R&D/ in 1997 Company (millions) $sales (%) research that takes place at universities is basic research. Applied research is aimed at gaining 1 General Motors 8200 04 9 the knowledge or understanding to meet a spe- 2 Ford Motor Company 6327 04 1 cific recognized need. Applied research includes 3 IBM 4307 05 2 4 Lucent Technologies 3100 611 8 investigations oriented to discovering new sci- 5 Hewlett-Packard 3078 07 2 entific knowledge that has specific commercial 6 Motorola 2748 08 6 objectives with respect to products, processes, 7 Intel 2347 09 4 or services. Development is the systematic use 8 Johnson & Johnson 2140 09 5 9 Pfizer 1928 015 4 of the knowledge or understanding gained from 10 Microsoft 1925 016 9 research directed toward the production of useful materials, devices, systems, or methods, including the design and development of prototypes and 95 Imation 194 98 9 20 96 Dana 193 02 2 processes. 97 Thermo Electron 191 65 4 Approximately 61 percent of national R&D is 98 Eastman Chemical 191 04 1 development, with 23 percent of R&D being allo- 99 Cabletron Systems 181 813 2 cated to applied research and 16 percent being 100 Whirlpool 181 02 1 allocated to basic research. Different sectors con- Source: Science & Engineering Indicators—2000, Appendix tribute disproportionately to the Nation’s funding Table 2-58. 170 Audretsch et al.

• Microsoft is, however, more than three times • Small firms are just as innovative as large firms, as R&D intensive as General Motors, meaning in general. But, in some industries, large firms that its invests nearly three times the amount have the innovative advantage (pharmaceuti- as General Motors relative to its sales. cals, aircraft), while in other industries small • In general (with notable exceptions such as firms have the innovative advantage (software, Cabletron Systems), larger R&D performers biotechnology). also spend more on R&D relative to their size • Small-firm and large-firm innovative activities than do the lower-ranked firms in the list of are complementary. the top100. • The level of R&D expenditures is not unre- Table IV provides a selected summary of the find- lated to the industry of the R&D performer. ings from the literature on innovation and firm For example, first-ranked General Motors and size. Ford Motor Company are in the transporta- The economic importance of small firms, tion industry. Second-ranked IBM, Lucent including the innovative differences between small Technologies, Hewlett-Packard, Motorola, and firms and large firms, requires an explanation Intel are in information systems. since the share of overall economic activity Company-specific data on R&D expenditures attributable to small firms is small and it did not for small-sized firms are not readily available. increase during the 1990s. The explanation rel- However, aggregate NSF data show that $98 bil- evant to the focus of this primer begins with a lion of the $169 billion that industry spent on model of the knowledge production function.21 R&D in 1998 was performed in firms with 10,000 or more employees. Thirty billion dollars (18% of the industry total) was performed in firms with Table IV fewer than 500 employees. Similarly, 70 percent of Selected studies of the relationshipbetween innovation and firm size the funds expended were part of company R&D budgets that exceeded $100 million. Innovation Stylized facts aside, there is a more subtle measure Findings Authors and perhaps more important R&D-related issue. R&D R&D spending in Mueller (1967) Since the early 1980s, policy makers have been positively related Grabowski (1968) concerned that critical American industries were to firm size Mansfield (1968) losing their competitive dominance of world mar- Patents Patenting is Scherer kets. During the 1990s, these same industries positively or (1965, 1983) seem to have reemerged as major international proportionally Pakes and competitors. While some of this resurgence is a related to Griliches (1980) firm size Hall et al. (1986) response to purposive policies, a portion of it Schwalback and can also be attributable to small firms, many of Zimmermann which were not in existence in the early 1980s (1991) to be affected by policy and many of which do New product Parity across firm Acs and Audretsch not even conduct R&D. Still, during the 1990s, innovations size, although (1990) small firms were a driving engine of growth, there are Audretsch (1995) differences job creation, and renewed global competitiveness according to through innovation. industry There is a rich literature related to the per- Adoption of Positive relationship Romeo (1975) formance of R&D in small firms as compared advanced between firm size Dunne (1994) to large firms. Some of the conclusions from this manufacturing and the probability Siegel (1999) research are: technologies of adopting an advanced manufacturing • Large firms have a greater propensity to patent technology than do small firms. The Economics of Science and Technology 171

The simplified production function in Equa- implement new ideas that otherwise would be tion (1) above can be expanded conceptually and rejected or would remain unexploited in an orga- analytically to include the stock of knowledge as a nizationally rigid firm. New firms thus serve as discrete input along with K and L. One investment agents of change. In a global economy where com- in knowledge that many firms make is in R&D. parative advantage is based in large part on inno- However, there are other key factors that generate vation, small firms are a critical resource. Public knowledge for the firm besides R&D, and in fact policies to enhance innovation in small firms are many small firms do not even conduct R&D yet discussed below. they are very innovative. Some such firms rely on knowledge that spills over from external sources including universities, and small firms are rela- R&D activity by geographic location tively more adept at absorbing knowledge from R&D activities in the United States are highly external sources than large firms. Table V pro- concentrated in a small number of states. In 1997, vides a brief summary of this spillover literature the 20 highest-ranking states in R&D accounted with respect to small firms. for about 86 percent of the U.S. total; the lowest Included in Table V under the source category 20 states accounted for only 4 percent. Califor- of individual spillovers are new employees. Why nia, at nearly $42 billion, had the highest level of are, for example, small firms able to exploit knowl- R&D expenditures; it alone accounted for approx- edge embodied in new employees to a greater imately one-fifth of the $199 billion U.S. total. extent than large firms? New and small firms pro- The six states with the highest levels of R&D vide the opportunity for creative individuals to expenditures—California, Michigan, New York, New Jersey, Massachusetts, and Texas (in decreas- Table V ing order of magnitude)—accounted for nearly Selected studies on knowledge spillovers two-thirds of the national effort. Among these Spillover topten states, California’s R&D effort exceeded, source Findings Authors by nearly a factor of three, the next-highest state, Michigan, with $14 billion in R&D expen- Industry Spillovers vary across Jaffe (1989) spillovers industries; greater Saxenien (1990) ditures. After Michigan, R&D levels declined rel- spillovers in Acs et al. (1992) atively smoothly to approximately $7 billion for knowledge-intensive Trajtenberg and Maryland. industries Henderson (1993) States that are national leaders in total R&D Audretsch and performance are usually ranked among the lead- Feldman (1996) ing sites in industrial and academic R&D per- University University spillovers Link and spillovers more important to Rees (1990) formance. For industrial R&D, nine of the top small firms than Audretsch and ten states were among the topten for total large firms Feldman (1996) R&D, with Ohio of the topindustrial R&D states Firm Firm spillovers Acs et al. (1994) replacing Maryland. For academic R&D, North spillovers more important to Feldman (1994) Carolina and Georgia replaced New Jersey and large firms than Eden et al. (1997) Washington. There was less commonality with the small firms topten for total R&D among those states that City spillovers Diversity generates Glaeser et. al performed the most federal intramural research. more spillovers than (1992) specialization; Almeida and Only four states were found in both top-ten lists: localized Kogut (1997) Maryland, California, Texas, and New Jersey. competition more Feldman and Competition for resources is a fundamental than monopoly Audretsch (1999) explanation for the skewed distribution of R&D Individual Spillovers shaped Audretsch and and science resources within the United States. spillovers by role and Stephan (1996) States that lack such resources lag others in inno- mobility of Prevezer (1997) knowledge workers vation, scholarship, graduate education, and overall economic growth.22 172 Audretsch et al.

The relationship between R&D and Table VI productivity growth Selected studies of the relationshipbetween R&D and productivity growth As previously noted, Robert Solow’s seminal article in 1957 established that an extremely large Findings Authors percentage of U.S. economic growth (over 87 87.5% of the increase in Solow (1957) percent) could not be explained by growth in aggregate output between 1909 conventional inputs, i.e., capital and labor. Since and 1949 can be attributed then, researchers have searched for statistical to technical change correlates of this unexplained growth, which is R&D has positive impact on total Terleckyj (1974) commonly referred to as technological advance- factor productivity growth as Scherer (1983) ment or change. Hence, what these investigators does R&D embodied in purchased Siegel (1997) intermediate and capital goods have done is to posit that technological change, measured as total factor productivity growth, is Rate of return to privately-financed Mansfield (1980) basic research greater than for Link (1981) causally related to increased investments in R&D. applied research or development Griliches (1986) Using manufacturing sector, industry, and firm- Lichtenberg and level data, researchers have examined the strength Siegel (1991) of the statistical relationshipbetween R&D and Small direct impact of Link (1981) total factor productivity growth. These analyses federally-financed R&D on total Griliches (1986) are based on models that correlate estimates of factor productivity growth the growth of A from Equation (3) with measures of R&D investment undertaken by firms, industries, or aggregate sectors (depending on the unit of return studies clearly indicate that the spillover of analysis). Mathematics aside, the extent of the benefits to society were somewhere in the 50 per- correlation can be shown to be a measure of the cent to 100 percent range. rate of return to R&D. This literature is consistent Because of these findings, namely that the pri- in terms of the following findings: vate and social rate of return to R&D is rela- • the rate of return to privately-funded R&D is tively high, policy makers have remained focused relatively large, ranging on average between 30 on R&D investments in the private sector as a tar- percent and 50 percent; get variable for stimulating economic growth. The • the rate of return to privately-funded basic research is significantly greater than to argument underlying such a focus is that, through privately-funded development, the differences incentives, firms will continue to invest in addi- being over 100 percent to basic research com- tional R&D projects and thus continue to stim- pared to about 15 percent to 20 percent for ulate economic growth and enhance standards of applied research plus development; and living through additional spillover effects. • the rate of return to federally-funded research Most of the academic studies associated with performed in industry varies by character this line of research were funded by the National of use; the returns to federally-funded basic Science Foundation during the late 1970s and research performed in industry is over 100 per- early 1980s, motivated in large part by a slowdown cent, while federally-funded development has a in industrial productivity growth that began in the negligible return on productivity growth. early 1970s23 and increased in the late 1970s and See summary Table VI. early 1980s (see Figure 3) and by the fact that U.S. Related studies have attempted to evaluate the industries were losing their competitive advantage social benefits, that is the spillover benefits to soci- in global markets.24 It is not surprising then that in ety, from industrial R&D. The rates of return to applied research and development described just the early 1980s, given the findings that the private above are for the most part private rates of return. and social rates of return to R&D were very high, More limited in number than the private rate of that there were several important policy initiatives return studies, the findings from the social rate designed specifically to stimulate industrial R&D. The Economics of Science and Technology 173

6. Government’s role in innovation President Clinton took a major stepforward in his 1994 Economic Report of the President by first The government should have an important role articulating principles about why the government to play in fostering innovation, especially private- has a role in innovation and in the overall techno- sector innovation. The following reasoning has logical process (p. 191): been used to justify government intervention in the innovation process: Technological progress fuels economic growth The Administration’s technology initiatives aim to promote • Innovation results in technological advance. the domestic development and diffusion of growth- and • Technological advance is the prime driver of productivity-enhancing technologies. They seek to correct economic growth. market failures that would otherwise generate too little investment in R&D The goal of technology policy is • Government has a responsibility to encourage not to substitute the government’s judgment for that of pri- economic growth. vate industry in deciding which potential “winners” to back. Rather the point is to correct [for] market failure . However, the economic underpinnings of government’s role in innovation are more complex Although the 1994 Economic Report of the than might first appear. From an economic per- President did not expand on how to correct for spective, the justification for the role of govern- market failure much less discuss appropriate pol- ment in innovation rests on a comparison of the icy mechanisms for doing so, it did for the first efficiency of market resources with and without time posit an economic rationale for government government intervention. involvement in the innovation process. Even more recently, the 1998 so-called Ehlers report, Unlock- ing Our Future, fails to articulate a clear rationale for government’s involvement although it vastly Economic rationale for government involvement departs in a positive way from the linear model Even today, many policy makers and academics of technology development proffered by Bush. point to Science—The Endless Frontier to date the origins of U.S. science and technology policy.25 Vannevar Bush did not articulate a science and Barriers to technology and market failure technology policy, and he did not articulate an Market failure refers to conditions under which economic rationale for government’s role. Rather, the market, including those performing R&D Bush like his predecessors and contemporaries and those adopting the R&D outputs of others, simply assumed that the government had a role underinvests from society’s perspective in any par- to play in the innovation process, and then he set ticular technology. Such underinvestment occurs out to describe that role (as opposed to a rationale because of conditions that prevent firms from fully for that role). realizing or appropriating the benefits expected The first U.S. policy statement on science and from their investments. Stated alternatively, firms technology was issued in 1990 by President Bush, underinvest in R&D when they determine, based U.S. Technology Policy. As with any initial policy on their expectations of post-innovation activity, effort, this was an important general document. that their private return is less than their private However, precedent aside, it too failed to articu- hurdle rate (minimum acceptable rate of return late a rationale for government’s intervention into on their R&D investment). This is of public con- the private sector’s innovation process. Rather, cern when the R&D investment not undertaken is like Science—The Endless Frontier and subsequent socially desirable. reports, it implicitly assumed that government had There are a number of factors that explain why such a role, and it then set forth a rather general a firm might perceive its private return to be less goal (1990, p. 2): than its hurdle rate. These factors are what we call barriers to technology and they relate to technical The goal of U.S. technology policy is to make the best use of technology in achieving the national goals of improved and market risk, where risk is defined to mea- quality of life for all Americans, continued economic sure the possibility that an actual outcome will growth, and national security. deviate from an expected outcome. Also, a firm 174 Audretsch et al. may believe that it cannot appropriate a sufficient Social Rate of Return return on its R&D investment, even if it can over- Private Hurdle Rate come technical uncertainty. This may be due to A B C a perceived inability to maintain proprietary control of the technology, thus enabling rivals to imitate their invention and reducing any resulting profitability. Individually or in combination, the Social Hurdle Rate following factors contribute to why a firm may perceive a private rate of return as being less than its hurdle rate26:

• Because of high technical risk, the outcome of 450 Private Rate of Return the R&D may not solve the technical problem adequately to meet perceived needs. Figure 6. Gapbetween social and privaterates of return to R&D projects. • Even if the R&D is technically successful, the market may not accept the technology because of competing alternatives or interoperability society from the private R&D investment. How- concerns. ever, the inability of the private sector to appropri- • Even absent technical and market risk, it may ate all benefits from its investment is not so great be difficult to assign intellectual property rights as to prevent the project from being adequately to the technology, and it might be quickly imi- tated so that the innovator may not receive funded by the private firm. adequate return on the R&D investment. In general, then, any R&D project that is expected to lead to a technology with a private These factors create barriers to investing in tech- rate of return to the right of the private hurdle nology, which thus lead to market failure. rate and on or above the 45-degree line is not From an economic perspective, the role of gov- a candidate for public support, and hence government is to correct for these market failures in ernment does not have an intervening role from those instances where society will benefit from the an economic perspective. Even in the presence of technology. This latter situation occurs when the spillover benefits, project C will be funded by the rate of return to society is greater than the social private firm. hurdle rate (minimum accepted rate of return For projects A and B, the gap between the for society from investments of resources with social and private returns is larger than in the case alternative social uses), as illustrated in Figure 6.27 of project C; neither project A nor B will be ade- The social rate of return is measured on the quately funded by the private firm as evidenced by vertical axis along with society’s hurdle rate on the private rate of return being less than the pri- investments in R&D. The private rate of return vate hurdle rate. To address this market failure, is measured on the horizontal axis along with the the government has several policy mechanisms. private hurdle rate on investments in R&D. A 45- Before examining specific policy mechanisms, degree line (long dashed line) is imposed on the a critical question is: Do private-sector firms, or figure. This is the point at which the social rate of does industry, underinvest in R&D? Alternatively return from an R&D investment equals the pri- stated: Are there many projects like A and B that vate rate of return from that same investment. The the private sector considers but rejects for some area above the 45-degree line and to the left of the reason? private hurdle rate is the area of policy interest. Private firms may not pursue promising techni- Three R&D projects are labeled in the figure. cal opportunities for the following reasons: As drawn, the private rate of return exceeds the private hurdle rate for project C, and the social • R&D scientific and technical frontiers are risky rate of return exceeds the social hurdle rate. The and the chances of failure are high. gap, measured vertically, between the social and • An individual firm may not have the capa- private returns reflects the spillover benefits to bilities required to developthe technology. The Economics of Science and Technology 175

Complex new technologies may require col- The Patent and Trademark Office issues laboration and information sharing; however, patents for inventions. The patent term is 20 years, the cost of establishing research and develop- and it grants exclusive property rights to the inven- ment partnership and making then work pro- tor over that period of time. Patents are effective ductively may provide disincentives to under- only within the United States and its territories taking the project. and possessions. • Private incentives may not be sufficient to U.S. law is clear about what can be patented. induce a firm to undertake the project in the Any person who: face of difficulties in appropriating the result- invents or discovers any new or useful process, machine, ing benefits (i.e., the resulting knowledge may manufacture, or composition of matter, or any new and follow to others who may benefit from the useful improvement thereof, may obtain a patent. R&D without sharing the cost). It is important to note the word “useful,” recalling Government has at its disposal at least four pol- too that Franklin created the American Philosoph- icy mechanisms to reduce risk and market fail- ical Society of Philadelphia in 1742 to promote ure and thus overcome an underinvestment in “useful knowledge.” One cannot patent an idea.30 R&D, where underinvestment refers to the situ- While the U.S. Code applies to patents in the ation where the private sector invests less in R&D United States, and in the territories and posses- than society would like it to invest (such as for sions of the United States, treaties have been pro- project A and project B). These policy mecha- mulgated to extend protection beyond national nisms include: boundaries. The Paris Convention for the Pro-

• Patent laws tection of Intellectual Property of 1883 provided • Tax incentives that each of the 140 signatory nations recognizes • Improved environment for collaborative re- the patent rights of other countries. Subsequent search treaties have extended such coverage and made • Subsidies to fund the research filings in other countries more efficient. Figure 7 illustrates the economics of patenting Each mechanism is discussed in the sections that from the perspective of the firm. The marginal follow. private rate of return to R&D is measured on The U.S. patent system is more of an institu- the vertical axis and the level of R&D spending tion than a policy mechanism. Thus, patent laws, is measured on the horizontal axis. The marginal in a sense, characterize the overall innovative envi- private return schedules are downward sloping ronment in which firms operate. In contrast, tax reflecting diminishing returns to R&D in any given incentives, efforts to stimulate collaborative R&D, time period, and for simplicity we assume the and direct and indirect government subsidies are marginal private return schedule to be linear. in a sense policy levers.

Marginal Private Rate 7. The patent system of Return The history of the U.S. patent system dates to the authority given to Congress in the Constitution of the United States.28 Article I, section 8 states:

Congress shall have power to promote the progress Marginal Private Cost of science and useful arts, by securing for limited times to authors and inventors the exclusive right to their respective Marginal Private Return writings and discoveries. with a patent

Marginal Private Return Based on this authority, Congress initiated a num- without a patent 29 ber of patent laws. The version of law that is now R&D Expenditures in effect was enacted on July 19, 1952 (to be effec- RD 0 RD1 tive January 1, 1953). It is Title 35 in the United Figure 7. Economics of patenting: increasing marginal private States Code. return for the ﬁrm. 176 Audretsch et al.

Absent the patent system, the firm will choose to • Since the enactment of the Bayh-Dole Act of invest RD0 in R&D. This is an optimal invest- 1980, which transferred ownershipof intellec- ment for the firm; it invests to the point where tual property from federal agencies to univer- its marginal cost of R&D (assumed to be constant sities, there has been a rapid rise in university for simplicity) equals its expected marginal return. patenting (see Henderson et al., (1998). One might think of RD as a level of R&D that 0 Researchers have investigated a number of the firm is willing to invest. Is it, however, at a aspects related to patenting activity, and the eco- level insufficient to generate a socially desirable nomic role of patents in the innovation pro- technology? If so, we then have the case of project cess. Some significant findings from this body B in Figure 6: the R&D expenditure is insufficient of research are summarized in Table VII. One for the project to be undertaken. Level RD1 in key result is that the value of patents is highly Figure 7 is sufficient for project B to be under- skewed, when citations (see Trajtenberg, 1990a; taken, but the firm does not have an incentive to Jaffe et al., 1993; and Henderson et al., 1998) are invest in R&D at that level. The key point is that used as an indicator of value. This result sug- the existence of the patent, or more precisely, the gests that the use of counts of patents as an indi- expectation that the firm will be awarded a patent, cator of innovative output can be misleading if can induce the firm to devote additional resources they are not properly deflated. Another critical to R&D (to reach level RD1.) finding is the existence of a strong positive cor- Receipt of monopoly power for 20 years relation between R&D expenditure and patents. through a patent increases the firm’s marginal pri- Finally, most studies report a positive correla- vate return from its investments in R&D, thus tion between patent activity and various measures shifting the marginal private return schedule to of economic performance, including productivity, the right. Several trends in patenting activity in the Table VII United States are noteworthy: Selected literature on patent activity

• In the early 1980s, the number of patents Findings Authors awarded to U.S. inventors began to decline Strong positive correlation Scherer (1965) and the number of U.S. patents awarded to between R&D expenditure Schmookler (1966) foreign inventors began to rise, thus causing (or employment) and patents Scherer (1983) some policy makers to question the inventive Bound et al. (1984) environment in U.S. firms. This trend was yet Pakes and another indicator that U.S. global competitive- Griliches (1984) Hall et al. (1986) ness was declining. Acs and Audretsch • During the 1980s, the number of U.S. patents (1989) awarded to foreign inventors was greatest for Positive correlation between patents Griliches (1981) Japanese inventors. In fact, in 1995, over and market value (stock market Hirschey (1982) 20,000 patents were awarded to Japanese rate of return and Tobin’s Q) Ben-Zion (1984) inventors compared to about 7,000 for the next Pakes (1985) highest represented country, Germany. Cockburn and Griliches (1988) • The share of total patents awarded to foreign Austin (1993) inventors is low in the United States compared Value of patents is highly skewed, Trajtenberg (1990a) to other countries. It is highest in Italy and where value is determined by Jaffe et al. (1993) Canada and lowest in Japan and Russia. citations Henderson et al. • During the past decade, Japanese inven- (1998) tors have more international patents in three Jaffe et al. (1998) important technologies than any country, with Citation-weighted measures of Hall et al. (2000) the United States being second. These tech- patents are more highly correlated with market value than nologies are: robotics, genetic engineering, and unweighted measures advanced ceramics. The Economics of Science and Technology 177 measures of accounting profitability, Tobin’s Q, The economics of tax credits and stock prices. Figure 8 illustrates the economics of a tax credit. A recent paper by Hall et al. (2000) pro- The marginal rate of return is measured on vides an important extension of this literature the vertical axis and the level of R&D spend- on estimation of the private returns (the returns ing is measured on the horizontal axis. Both the that accrue to firms) to patenting by attempting marginal social return and the marginal private to link citation-weighted patents to the market return schedules are downward sloping reflect- value of firms. Their preliminary results suggest ing diminishing returns to R&D investments in a that citation-weighted measures of patents are given time period. The social return schedule is more highly correlated with market value than drawn greater than the private return schedule for unweighted measures. all levels of R&D because firms cannot appropriate all the benefits from conducting R&D; some of those benefits spillover to other firms in the current time period and in the post-innovation time 8. Tax incentives as a policy tool period thus generating additional benefits to soci- Tax incentives are another mechanism that gov- ety. The marginal cost to the firm to undertake ernment uses to increase private sector R&D. R&D is shown to be constant (horizontal). Like any policy tool, tax incentives have advan- As drawn, the firm will equate its marginal tages and disadvantages.31 Advantages include the cost to conduct R&D with its marginal return, following: and the firm will invest RD0. One might think of RD0 as corresponding to a partial level of funding • Tax incentives entail less interference in the for project B in Figure 6, much like the case of marketplace than do other mechanisms, thus the patent illustration in Figure 7. Society, given affording private-sector recipients the ability the firm’s marginal cost schedule, would like the to retain autonomy regarding the use of the firm to invest in R&D to maximize social benefits. incentives. Hence, the optimal credit is one that provides an • Tax incentives require less paperwork than incentive to the firm to increase its R&D to point other programs. RD1. Receipt of a tax credit can be thought of • Tax incentives obviate the need to directly tar- as a reduction in the marginal cost of undertak- get individual firms in need of assistance. ing additional R&D, and the firm will re-equate • Tax incentives have the psychological advan- its new marginal cost with its marginal return tage of achieving a favorable industry reaction. and invest at RD1. Unlike patents, the research • Tax incentives may be permanent and thus do and experimentation (R&E) credit does not cor- not require annual budget review. rect for market failure. It simply increases the firm’s private return on marginal R&D projects • Tax incentives have a high degree of political feasibility. Marginal Rate Some disadvantages of tax incentives are: of Return

• Tax incentives may bring about unintended windfalls by rewarding ﬁrms for what they would have done in the absence of the incentive. Marginal Private Cost • Tax incentives often result in undesirable in-

equities. Marginal Social Return • Tax incentives raid the federal treasury. Marginal Private Return • Tax incentives frequently undermine public R&D Expenditures accountability. RD0 RD1 • The effectiveness of tax incentives often varies Figure 8. Economics of a tax credit: decreasing marginal pri- over the business cycle. vate cost. 178 Audretsch et al. by reducing its marginal private cost to undertake In 1996, the Congressional Ofﬁce of Technol- such projects. Thus, tax incentives will increase the ogy Assessment released a report on the effec-

firm’s level of R&D from RD0 to RD1 but will not tiveness of the R&E tax credit. The report con- alleviate technical or market risk. cluded: However, because R&D is not a homogeneous • There is not sufficient information available activity and because the research (R) portion of to conduct a complete benefit-cost analysis of R&D has a greater impact on productivity growth the effectiveness of the R&E tax credit on the and hence economic growth than does develop- economy. ment (D), any uniform tax incentive that treats • The econometric studies that have been done R&D as if it were a homogeneous activity will to date conclude that the credit has been effec- likely encourage more of the same mix of R&D. tive in the sense that for every dollar lost in That may not be bad in the case of R&D since federal revenue there is an increase of a dollar economic studies suggest that the marginal private in private sector R&D spending. These studies return from R&D is greater than the marginal conclude that the credit would be more effec- cost of conducting R&D. However, a tax credit on tive if it were made permanent.33 research as opposed to development, while con- • The R&E tax credit represents a small fraction ceptually more desirable, could be cumbersome to of federal R&D expenditures, about 2.6 per- administer. cent of total federal R&D funding and about 6.4 percent of federal R&D for industry. R&E tax credit The R&E tax credit is not unique to the United The adoption of Section 174 of the Internal States. Japan’s tax credit is marginal, and it was Revenue Code in 1954 codified and expanded initiated in 1966. Canada also initiated a pro- tax laws pertaining to the R&D expenditures gram in the 1960s, but their program is a flat tax of firms. This provision permitted businesses to program. deduct fully R&E expenditures but not devel- Economic theory predicts that firms that coop- opment or research application expenditures in erate in a research joint venture type of arrange- the year incurred.32 Businesses under Section 174 ment have an incentive to cooperate at the are not allowed to expense R&E related equip- research end of the R&D spectrum rather than at ment. Such equipment must be depreciated. How- the development end. Thus, some have proposed ever, the Economic Recovery and Tax Act of a tax credit for cooperative research involvement 34 1981 (ERTA) provided for a faster depreciation as a viable alternative to the R&E tax credit. of R&E capital assets than other business capital assets. 9. Research collaborations ERTA also included a 25 percent tax credit for qualified R&E expenditures in excess of the Research partnerships, meaning formal or infor- average amount spent during the previous three mal collaboration among firms in the conduct of taxable years or 50 percent of the current year’s research, are an organizational form that may expenditures (the R&E base). The initial R&E overcome elements of market failure by reduc- tax credit had several limitations including the ing technical risk to the R&D-conducting firms by fact that it did not cover expenses related to the enlarging the underlying knowledge base. These administration of R&D or to research conducted partnerships may also limit market risk by helping outside of the United States. The Tax Reform Act to ensure that particular elements of the technol- of 1986 modified these limitations, but reduced ogy are standardized and thus interoperate in a the marginal rate from 25 percent to 20 percent. system. As well, to the extent that collaboration Over the years the credit had been modified, pri- reduces redundant research, there may be cost marily in terms of the definition of the R&E base, savings to each partner, reduced time to market, but the credit has never been made permanent. It and better appropriability of R&D results. has expired a number of times, only to be renewed Research partnerships assume many forms in retroactively. practice. Partners may aim to develop or refine The Economics of Science and Technology 179 a new product, improve production processes, set IBM described to the industry the nature of the standards, or developtechnology to meet environ- growing competition with Japan and proposed the mental regulations. The collaboration may take creation of a “semi-conductor research coopera- place between partners who compete in the mar- tive” to assure continued U.S. technology leader- ketplace, or between partners who produce com- ship. This event witnessed the birth of the SRC. In plementary products. Some partnerships include December 1981, Robert Noyce, then SIA chair- government or university partners as well. man and vice-chairman of Intel, announced the Research partnerships are certainly not a new establishment of the SRC for the purpose of stim- organizational form, but since the mid-1980s gov- ulating joint research in advanced semiconduc- ernment has provided a favorable environment for tor technology by industry and U.S. universities them. Before reviewing the related policies, we and to reverse the declining trend in semiconduc- describe one research collaboration in detail as an tor research investments. The SRC was formally illustration of the types of collaborations that are incorporated in February 1982 with a stated pur- prevalent and that have had a major impact on pose to36: research. • Provide clearer view of technology needs. • Fund research to address technology needs. Semiconductor research corporation • Focus attention on competition. • Reduce research redundancy. One of the first formal research collaborations in the United States was the Semiconductor Policy makers soon noticed the virtues of coop- Research Corporation (SRC). A brief history of erative research in part because such organiza- the SRC will serve to illustrate that many research tional structures had worked well in Japan and collaborations or partnerships are formed to in part because the organizational success of the address industry-wide technological issues, or at SRC demonstrated that cooperation among com- least issues that affect a sizeable segment of the petitive firms at the fundamental research level industry. This history is also interesting because it was feasible. illustrates a purposeful entrepreneurial response to competitive market conditions.35 In the late 1950s, an integrated circuit (IC) Public policy toward research collaborations industry emerged in the United States. The To place the activities surrounding the SRC’s for- fledgling industry took form in the 1960s and expe- mation in a broader perspective, recall that in the rienced rapid growth throughout the 1970s. In early 1980s there was growing concern about the 1979, when Japanese companies captured 42 per- persistent slowdown in productivity growth that cent of the U.S. market for 16 kbit DRAMs first began to plague the U.S. industrial sector (memory devices) and converted Japan’s inte- in the mid-1970s and about industry’s apparent grated circuit trade balance with the United States loss of its competitive advantage in world markets, from a negative $122 million in 1979 to a posi- especially firms in the semiconductor industry.37 tive $40 million in 1980, the U.S. industry became As noted in a November 18, 1983 House report painfully aware that its dominance of the IC indus- about the proposed Research and Development try was being seriously challenged. It was clear to Joint Ventures Act of 1983: all in the industry that it was in their collective best interest to invest in an organizational struc- A number of indicators strongly suggest that the position of world technology leadershiponce firmly held by the United ture that would strengthen the industry’s position States is declining. The United States, only a decade ago, in the global semiconductor marketplace. with only five percent of the world’s population was gener- The Semiconductor Industry Association (SIA) ating about 75 percent of the world’s technology. Now, the was formed in 1977 to collect and assemble reli- U.S. share has declined to about 50 percent and in another able information on the industry and to develop ten years, without fundamental changes in our Nation’s mechanisms for addressing industry issues with technological policy the past trend would suggest that it may be down to only 30 percent. [In hearings,] many dis- the federal government. In a presentation at an tinguished scientific and industry panels had recommended SIA Board Meeting in June 1981, Erich Bloch of the need for some relaxation of current antitrust laws to 180 Audretsch et al.

encourage the formation of R&D joint ventures. The facility where member companies could improve encouragement and fostering of joint research and devel- their semiconductor manufacturing process tech- opment ventures are needed responses to the problem of nology. Its establishment came after the Defense declining U.S. productivity and international competitive- Science Board recommended direct government ness. According to the testimony received during the Com- mittee hearings, this legislation will provide for a significant subsidy to the industry in a 1986 report com- increase in the efficiency associated with firms doing similar missioned by the Department of Defense (there- research and development and will also provide for more fore SEMATECH is discussed below under the effective use of scarce technically trained personnel in the broader heading of a public/private technology United States. partnership). It was thought that SEMATECH In an April 6, 1984 House report on competing would be the U.S. semiconductor industry’s/U.S. legislation, the Joint Research and Development government’s response to the Japanese govern- Act of 1984, the supposed benefits—and recall ment’s targeting of their semiconductor indus- that at this time it was still too soon for there to try for global domination. Since its inception, be visible benefits coming from the SRC’s activi- SEMATECH’s stated mission has evolved and ties on behalf of the IC industry—of joint research become more general. The consortium currently and development were for the first time clearly defines its mission around solving the technical articulated: challenges presented in order to sustain a lead- Joint research and development, as our foreign competitors ershippositionfor the United States in the global have learned, can be procompetitive. It can reduce duplica- semiconductor industry. tion, promote the efficient use of scarce technical personnel, and helpto achieve desirable economies of scale. [W]e must ensure to our U.S. industries the same eco- Trends in RJVs nomic opportunities as our competitors, to engage in joint research and development, if we are to compete in the To date, there have been over 800 formal RJVs world market and retain jobs in this country. filed under the NCRA. Certainly, this number is The National Cooperative Research Act a lower bound on the total number of research (NCRA) of 1984, after additional revisions in the partnerships in the United States, even since 1984. initiating legislation, was passed on October 11, Not all are as publicly visible as SEMATECH. 198438: Some are quite small, with only two or three members, and others are quite large with hundreds to promote research and development, encourage innovation, stimulate trade, and make necessary and appropri- of members. On average, a joint ventures has 14 ate modifications in the operation of the antitrust laws. members.39 While informal cooperation in research may The NCRA created a registration process, later have been prevalent in the United States for expanded by the National Cooperative Research decades, formal RJV relationships are new and it and Production Act (NCRPA) of 1993, under will take longer than a decade and a half to detect which research joint ventures (RJVs) can disclose meaningful trends. their research intentions to the Department of Albeit that research partnerships as formal Justice. RJVs gain two significant benefits from entities are relatively new to the technology strat- such voluntary filings: if subjected to criminal or egy arena, the literature concludes that there civil action they are evaluated under a rule of rea- are both benefits and costs to members of the son that determines whether the venture improves venture.40 The benefits include: social welfare; and if found to fail a rule-of-reason analysis they are subject to actual rather than • the opportunity for participants to capture treble damages. knowledge spillovers from other members, One of the more notable RJVs formed and • reduced research costs due to a reduction in made public through the NCRA disclosure pro- duplicative research, cess was SEMATECH (SEmiconductor MAnufac- • faster commercialization since the fundamen- turing TECHnology). It was established in 1987 tal research stage is shortened, and as a not-for-profit research consortium with an • the opportunity to form, in some cases, original mission to provide a pilot manufacturing industry-wide competitive vision. The Economics of Science and Technology 181

The costs include: And (Council on Competitiveness, 1996, p. 11):

• a lack of appropriability since research results For universities, cutbacks in defense spending have resulted are shared among the participants, and in a de facto reallocation of funding away from the physical sciences and engineering and shifted the focus of defense • managerial tension, in some cases, as partic- research away from the frontiers of knowledge [e.g., basic ipants learn to trust each other and to work science] to more applied efforts. Although defense together. spending is clearly not the only viable mechanism to support frontier research and advanced technology, the United Research partnerships are correctly viewed as States has yet to find an alternative innovation paradigm to a complementary source of technical knowledge replace it. and technical efficiency for the firm. Thus, firms that participate in a research partnership enhance Given this spending trend, and the increasing their own R&D process through interactions. ease of global technology transfer, it is conceiv- able, at least according to the Council, that the United States may lose its technological leader- shipin some importantareas such as health and Universities as research partners advanced materials since innovation in these fields Especially noticeable in the RJVs filed with the is closely linked to improvements in basic science. Department of Justice are the presence of univer- There is some indication that scholars are sities as research partners and that the number beginning to think more deeply and more broadly of RJVs with at least one university partner has about the social, economic, and technological con- increased over the past 15 years.41 On average, sequences of university involvement in private sec- about 15 percent of RJVs have at least one uni- tor research partnerships. This thinking reflects versity research partner, and of these over 90 per- some major concerns about the impact of these cent are U.S. universities. RJVs with universities relationships on the research university’s mission as research partners have, on average, five univer- to conduct basic research. Unfortunately, more 42 sity partners. information is necessary in order for researchers The literature on universities as research part- to examine the ramifications of this trend from a ners is sparse. However, some stylized conclusions wide variety of disciplinary perspectives. 43 can be drawn from the limited investigations : It is likely that the increasing trend toward university private-sector research partnerships will • Firms that interact with universities generally have greater R&D productivity and greater continue. A 1993 national survey of U.S. biology, patenting activity. chemistry, and physics faculty members revealed • One key motive for firms to maintain joint that many academic scientists desired more such research relationships with universities is to involvement. An earlier survey of engineering fac- have access to key university personnel—facul- ulty members reached the same conclusion. How- ty as well as students as potential employees. ever, one of the authors (Morgan, 1998, p. 169) of the surveys was quick to point out one area of Many speculate that university participation major concern: may increase in importance and frequency in the future. According to the Council on Competitive- a diminution of the role of the university as an indepen- ness (1996, pp. 3–4): dent voice to helplook out for the broader societal good and to guard against industrial as well as other excesses. Over the next several years, participation in the U.S. R&D An independent science, engineering and public policy role enterprise will have to continue experimenting with differ- is essential to ensure an adequate supply of well educated ent types of partnerships to respond to the economic con- scientists and engineers prepared to work with the public straints, competitive pressures and technological demands sector in public interest groups. Having industry assume that are forcing adjustment across the board. [and in a more central role as customer and client for university- response] industry is increasingly relying on partnerships based scientific and engineering research, while in some with universities, while the focus of these partnerships is way a natural and desirable step, needs to be balanced shifting progressively toward involvement in shorter-term against the need for independence, oversight and service research. to society and the larger public good. 182 Audretsch et al.

Government laboratories as a research partner led to the development and marketing of a new product, a product development rate comparing The federal government also enters directly into favorably to firms operating independently. Those research partnerships with firms through the fed- companies most likely to develop products from eral laboratory system. This relationshipcan take their partnerships with federal laboratories had various forms ranging from informal relationships the following characteristics: (1) smaller than the whereby a firm(s) interacts with a federal lab- average for all companies in the data base, (2) oratory scientist, or more formal relationships high levels of R&D intensity (R&D employees whereby a firm(s) utilizes federal laboratory facili- as a percentage of total employees), (3) more ties and is jointly involved with the laboratory sci- recently established firms (Bozeman and Wittmer, entists in the research. Or, the relationshipcan 2002). Interestingly, while 89 percent of partici- be nothing more than an informational transfer pating companies reported a high degree of sat- whereby the firm utilizes public information that isfaction with their federal laboratory partnership, was generated within a government agency. those developing products were actually somewhat While very few studies have systematically less satisfied, perhaps because the federal labora- looked at the economics of federal laboratories tories are generally a better source for upstream as research partners, three generalizations can be basic and applied research than for technology made44: development (Rogers and Bozeman, 1997). This • Federal laboratories are generally associated interpretation comports with a finding from a sep- with research joint ventures that are large in arate study using a different data set (Roessner terms of other member companies. and Bean, 1991). • One key advantage to partnering with a federal laboratory is access to specialized technical equipment. Comparing universities and government • Firms’ research with laboratories tends to be laboratories as research partners nearer the basic research end of the R&D In recent times, much science and technology spectrum and, related, firms generally view policy has striven to determine the respective access to the laboratories’ research scientists as advantages and disadvantages of government lab- a more important benefit than the technologies oratories and universities in cooperative research available for licensing. and technology transfer. Absent such knowledge, One of the few broad-based research pro- it is difficult to make decisions about the alloca- grams examining federal laboratories as research tions of resources among institutional actors. partners is work performed by Bozeman and According to the directors of the federal labo- colleagues (summarized in Crow and Bozeman, ratories, a major comparative advantage of federal 1998). According to a study by Bozeman and laboratories is their ability to perform interdis- Papadakis (1995), companies’ objectives for inter- ciplinary team research. Generally, universities, acting with federal laboratories include engaging organized on the disciplinary lines, have difficulty in strategic pre-commercial research (51 percent), performing research that cuts across disciplines interest in access to the unique resources of the and traditional academic departments. This has lab (45 percent) and a desire to develop new prod- changed somewhat in recent years, however, has ucts and services (42 percent). Their decisions to universities have developed a wide array of inter- work with the federal laboratories are particularly disciplinary centers, often in response to National related to the skills and knowledge of the federal Science Foundation initiatives for creating science lab’s scientists and engineers (61 percent). Many centers or engineering research centers. A second companies are “repeat customers” and 42 percent major advantage of the federal laboratories, espe- indicate that prior experience with the lab is one cially the national labs, is that extremely expensive, of the major reasons for choosing to work with the often unique, scientific equipment and facilities lab in the more recent project. are located on their premises. The “user facilities” Regarding the incidence of commercial out- at federal laboratories are designed explicitly to comes, more than 22 percent of the projects share resources and these user facilities can be an The Economics of Science and Technology 183 important instrument for technology transfer. Few developcolleges to offer curricula in agriculture universities, or even university consortia, have the and mechanical arts. Then in 1887, the Hatch Act resources to build facilities national in scope. provided resources for a system of state agricul- The most obvious advantage of universities ture experiment stations that would be under the over federal laboratories is a vitally important auspices of land-grant colleges and universities. A one—students. The presence of students makes a partnership among the various levels of govern- remarkable difference in the output, culture and ment was established by the Smith-Level Act of utility of research. In recommending that federal 1914. The Cooperative Agriculture Extension ser- funding for science and technology give strongest vice was charged to deliver the practical benefits of emphasis to academic institutions, the National research to citizens through an extension service. Academy of Sciences’ Committee on Criteria for According to Carr (1995, p. 11): Federal Support of Research and Development Until the end of the 1970s, the philosophy behind the dis- (National Academy of Sciences, 1995, p. 20) con- semination of federally-funded research was that if the pub- cluded that university R&D funding supports pro- lic paid for the research, the resulting intellectual property duction of “well prepared scientists and engineers should be made equally available to all interested parties. who not only will be the next generation of fac- Because the 1970s and 1980s witnessed many ulty, but who will also work productively in, and foreign competitors beginning to successfully chal- transfer technology to, industry and government.” lenge the long-standing dominance of the United Students are a reservoir of cheaplabor supporting university research, bartering their below States both in world and domestic markets, it market wage rate for training. More important became clear to public and private policymakers for present purposes is that students are a means that a change in the philosophy of federal R&D of technology transfer (through postgraduate job support was needed. The Office of Technology placements) and they often provide enduring links Policy (1996) reflected on the motivation for this as the social glue holding together many fac- change in policy mindset as: ulty scientists and the companies they work with. • Global competitors were better able to Roessner and Bean (1991) found that the single appropriate the output of U.S. basic and most important benefit to industry from partic- mission-oriented research as their technical ipation in the NSF Engineering Research Cen- sophistication grew. ters, according to the industrial participants them- • Traditional public-sector mechanisms of tech- selves, is the ability to hire ERC students and nology development, transfer, and develop- graduates. In some cases, the vast benefits accru- ment took too long in an era of accelerating ing from students are enjoyed by government private sector development. laboratories, but chiefly at such institutions as • U.S. federal R&D represented a declin- Lawrence Berkeley Lab or Ames Laboratory, ing share of the world R&D as globally- those actually located on university campuses. We shall return to this issue subsequently in a discus- competitive nations increased their public sion of the role of “scientific and technical human funding; hence the marginal benefits to indus- capital” (Rogers and Bozeman, 1997). try from additional public moneys (allocated in the same historical manner) declined.

10. Public/private technology partnerships45 Stated alternatively, but maintaining the same general theme, Link and Tassey (1987, pp. 4, 131) One can trace the origins of public/private reﬂected on the changing competitive environ- partnerships—federal grants assistance technology ment of U.S. industry in the early 1980s: partnerships—in the United States at least as far back as the Lincoln Administration. In 1862, the there is a new order of competition in the world. An Morrill Act established what was known as the inescapable element of the competition is technology land-grant college system. The Act created a part- With the advent of technology-based economies [throughout the world], the increase in the number of world com- nershipbetween the federal and state govern- petitors has been greater than the increase in the size of the ments to cooperate with the private sector in tech- world market. What has resulted from this is a signiﬁcant nology development. The Act charged states to shortening of technology life cycles As such, effective 184 Audretsch et al.

long-run competitive strategies will have to deal explicitly share to grow at the expense of that of the United with technology [C]ompetitive survival will depend on States. In January 1997, President Reagan recom- technology-based strategies. These strategies will have to mended $50 million in matching federal funding evolve from new philosophies about interdependence for R&D related to semiconductor manufactur- The importance of interdependence arises from the need of a domestic industry to rapidly and efficiently develop ing, and this was to be part of the Department of complex technological elements from which specific appli- Defense’s 1988 budget. Soon thereafter, the SIA cations (innovations) are drawn for competitive activity approved the formation of SEMATECH and the [Accordingly,] government must expand and adapt its construction of a world-class research facility.47 role with industry for more effective joint planning in In September 1987, Congress authorized $100 mil- research. lion in matching funding for SEMATECH. Beginning with legislation in the 1980s, as sum- SEMATECH and its members have a mission marized in Table VIII, a new era in federal tech- to: nology policy began. This new era was based create a shared competitive advantage by working on the belief that the global competitiveness of together to achieve and strengthen manufacturing technol- U.S. firms can be enhanced through legislation to ogy leadership. bolster the commercial impact of federal R&D This shared vision is accomplished by joint spon- investments. As such, using the terminology of the sorshipof leading edge technology development Office of Technology Policy (1996, p. 26), “a new in equipment supplier companies. As these com- paradigm for public/private technology partner- panies become world-class manufacturers, so will ships emerged.” This new paradigm viewed indus- the members of SEMATECH. try as a partner in the formation and execution of By 1988, Japan’s world market share reached technology programs rather than a passive recipi- over 50 percent, and that of the United States ent of the output from federal research. fell to about 37 percent. The U.S. share declined Several public/private partnerships are over- again in 1989 and then it began to increase at viewed below from an institutional perspective. the expense of that of Japan. Early in 1992, the SEMATECH is discussed in detail because it was United States was again at parity with Japan at one of the earliest public/private partnerships and about 42 percent, and stayed slightly ahead of for years was heralded as the model organizational Japan until 1995 when the gap began to widen. form for other public/private partnerships to fol- The mid-1990s saw increasing cooperation low. The Small Business Innovation Research Pro- between U.S. and Japanese semiconductor com- gram and the Advanced Technology Program are panies, and in fact, in 1998 International SEMAT- also singled out for discussion because these are ECH began operations with Hyundai (Japan) and two of the programs that have conducted in-depth Philips (Amsterdam) as important members. Also, program evaluations and hence more is known beginning in 1998, all funding came only from about their economic impacts than about those member companies. associated with other partnerships.

Small Business Innovation Research Program48 SEMATECH46 The Small Business Innovation Research (SBIR) In 1986 when the Semiconductor Industry Associ- program began at the National Science Founda- ation (SIA) and the Semiconductor Research Cor- tion (NSF) in 1977. At that time the goal of poration (SRC) began to explore the possibility the program was to encourage small businesses, of joint industry/government cooperation, the U.S. long believed to be engines of innovation in the semiconductor industry was not in a favorable eco- U.S. economy, to participate in NSF-sponsored nomic position. During 1986, Japan overtook the research, especially research that had commercial United States for the ﬁrst time in terms of their potential. Because of the early success of the pro- share of the world semiconductor market. Japan gram at NSF, Congress passed the Small Business had about 45 percent of the world market com- Innovation Development Act of 1982. The Act pared to about 42 percent for the United States. required all government departments and agencies The U.S. semiconductor industry expected Japan’s with external research programs of greater than The Economics of Science and Technology 185

Table VIII Selected public/private technology partnership legislation

Enabling legislation Characteristics of the program

Stevenson-Wylder Technology The Act was predicated on the premise that federal laboratories embody important and Innovation Act of 1980 industrially-useful technology. Accordingly, each federal laboratory was mandated to establish an Office of Research and Technology Application to facilitate the transfer of public technology to the private sector. University and Small Business This Act is also known as the Bayh-Dole Act. It reformed federal patent policy by pro- Patent Procedure Act of 1980 viding increased incentives for the diffusion of federally-funded innovation results. Universities, non-profit organizations, and small businesses were permitted to obtain titles to innovations they developed with the use of governmental financial support, and federal agencies were allowed to grant exclusive licenses to their technology to industry. Small Business Innovation This Act required federal agencies to provide special funds to support small business Development Act of 1982 R&D that complemented the agency’s mission. These programs are called Small Business Innovation Research (SBIR) programs. The Act was reauthorized in 1992. National Cooperative Research NCRA was legislated in an effort to reduce antitrust barriers and thus encourage the Act of 1984 formation of joint research venture among U.S. firms. This Act was amended by the National Cooperative Research and Production Act of 1993, thereby expanding antitrust protection to joint production ventures. Trademark Clarification This Act set forth new licensing and royalty regulations to take technology Act of 1984 from federally-funded facilities into the private sector. It specifically permitted government-owned, contractor-operated (GOCO) laboratories to make decisions regarding which patents to license to the private sector, and contractors could receive royalties on such patents. Federal Technology Transfer This Act was amended by the Stevenson-Wylder Act. It made technology transfer an Act of 1986 explicit responsibility of all federal laboratory scientists and engineers. It authorized cooperative research and development agreements (CRADAs). This Act was amended by the National Competitiveness Technology Transfer Act of 1989 which expanded the definition of a federal laboratory to include those that are contractor operated. Omnibus Trade and Competitiveness This Act established two competitiveness programs, the Advanced Technology Program Act of 1988 (ATP) and the Manufacturing Extension Partnership(MEP) within the re-named National Institute of Standards and Technology (NIST). Defense Conversion, Reinvestment, This Act created an infrastructure for dual-use partnerships. Through Technology Rein- and Transition Assistance Act vestment Project partnerships the Department of Defense was given the ability to of 1992 leverage the potential advantages of advanced commercial technologies to meet departmental needs.

$100 billion to establish their own SBIR programs 3. to foster and encourage participation by minor- and to set aside funds equal to 0.2 percent of ity and disadvantaged persons in technological the external research budget.49 Currently, agencies innovation must allocate 2.5 percent of the external research 4. to increase private sector commercialization of budget to SBIR. innovations derived from federal research and The 1982 Act states that the objectives of the development. program are: The Act was reauthorized in 1992. SBIR awards are of three types. Phase I awards 1. to stimulate technological innovation are small, generally less than $100,000. The pur- 2. to use small business to meet Federal research pose of these awards is to assist firms to assess and development needs the feasibility of the research they propose to 186 Audretsch et al. undertake for the agency in response to the Appropriations to ATP increased from $10 mil- agency’s objectives. Phase II awards can range lion in 1990 to a peak of $341 million in 1995. upto $750,000. These awards are for the firm Funding decreased in 1996 to $221 million, and to undertake and complete its proposed research, has averaged about $200 million per year until hopefully leading to a commercializable product 2000 when it fell to just under $150 million. To or process. date, ATP has funded through competitive pro- The Department of Defense’s (DoD’s) SBIR cesses approximately 450 research projects. program has been studied in some detail. It can be Much like the case of DoD’s SBIR program, concluded that DoD’s SBIR program in encour- ATP has provided incentives to firms to under- aging commercialization from research that would take research that would not otherwise have been not have been undertaken without SBIR support. pursued—such as projects A or B in Figure 6. And moreover, the structure of DoD’s SBIR program is overcoming reasons for market failure that previously have caused small firms to underinvest 11. Infrastructure technology in R&D.50 Infrastructure technology is a term that refers to the technological environment in which firms innovate. Two important parts of the techno- Advanced Technology Program logical environment of firms are patent laws The Advanced Technology Program (ATP) was (previously discussed) and the federal laboratory established within the National Institute of system (publicly funded). The federal laboratory Standards and Technology (NIST) through the system’s output is infrastructure technology, which Omnibus Trade and Competitiveness Act of 1988, by its nature is a public good, freely accessible by and modified by the American Technology Preem- firms. inence Act of 1991. The goals of the ATP, as stated in its enabling legislation, are to assist U.S. businesses in creating and applying the generic tech- Federal laboratory system nology and research results necessary to: As shown above in Table II, the federal govern- 1. commercialize significant new scientific discov- ment allocated $65.8 billion to R&D in 1999, of eries and technologies rapidly which $17.4 billion was performed by the fed- 2. refine manufacturing technologies. eral government. Most of this research occurred in the more than 700 federally-funded R&D lab- These same goals were restated in the Federal oratories in the country.51 The Department of Register on July 24, 1990: Defense’s laboratories account for almost one- The ATP will assist U.S. businesses to improve their half of this intramural research, and the Depart- competitive position and promote U.S. economic growth ment of Health and Human Services accounts for by accelerating the development of a variety of pre- about 15 percent. Table IX provides some sum- competitive generic technologies by means of grants and mary information about a few of the laboratories cooperative agreements. in the federal laboratory system. The ATP received its first appropriation from The governmental agency that provides a sig- Congress in FY 1990. The program funds nificant amount of direct infrastructure technology research, not product development. Commercial- to the industrial economy is the National Institute ization of the technology resulting from a project of Standards and Technology (NIST) within the might overlapthe research effort at a nascent Department of Commerce.52 NIST is discussed level, but generally full translation of the tech- herein in more detail than any other federal lab- nology into products and processes may take a oratory for two reasons. One, its research mis- number of additional years. ATP, through cost sion is varied by field of science, as opposed to sharing with industry, invests in risky technologies being focused on one major scientific area such as that have the potential for spillover benefits to the energy; and two, the Program Office within NIST economy. has a long history of evaluation of the outputs The Economics of Science and Technology 187

Table IX Overview of selected national laboratories

Laboratory Year established Research Ownership

Ames Laboratory 1942 Energy Department of Energy operated by Iowa State University Brookhaven National Laboratory 1947 Energy Department of Energy operated by SUNY Stony Brook Los Alamos National Laboratory 1943 National security Department of Energy and energy National Institute of 1901 Standards Department of Commerce Standards and Technology Rome Laboratory 1951 Communications Department of Defense

Source: Crow and Bozeman (1998). from its research programs, and as such its con- On July 20, 1866, Congress and President tribution to economic growth is more readily doc- Andrew Johnson authorized the use of the metric umented. system in the United States. This was formalized in the Act of 28 July 1866—An Act to Authorize the Use of the Metric System of Weights and Mea- National Institute of Standards and Technology sures: Historical overview of NIST.53 A standard is a pre- Be it enacted , That from and after the passage of this scribed set of rules, conditions, or requirements act it shall be lawful throughout the United States of Amer- ica to employ the weights and measures of the metric sys- concerning: tem; and no contract or dealing, or pleading in any court, shall be deemed invalid or liable to objection because the • Definitions of terms weights or measures expressed or referred to therein are • Classification of components weights and measures of the metric system. • Specification of materials, their performance, As background to this Act, the origins of the and their operations metric system can be traced to the research of • Delineation of procedures Gabriel Mouton, a French vicar, in the late 1600s. • Measurement of quantity and quality in His standard unit was based on the length of an describing materials, products, systems, ser- arc of 1 minute of a great circle of the earth. Given vices, or practices. the controversy of the day over this measurement, the National Assembly of France decreed To understand the current activities that take on May 8, 1790, that the French Academy of Sci- place at NIST, its public good mission must be ences along with the Royal Society of London placed in an historical perspective. The concept of deduced an invariable standard for all the mea- the government’s involvement in standards traces sures and all the weights. Within a year, a stan- to the Articles of Confederation signed on July 9, dardized measurement plan was adopted based 1778. In Article 9, §4: on terrestrial arcs, and the term mètre (meter), The United States, in Congress assembled, shall also have from the Greek metron meaning to measure, was the sole and exclusive right and power of regulating the assigned by the Academy of Sciences. alloy and value of coin struck by their own authority, or by Because of the growing use of the metric system that of the respective States; fixing the standard of weights and measures throughout the United States in scientific work rather than commercial activity, the French government held an international This responsibility was reiterated in Article 1, §8 conference in 1872, which included the participa- of the Constitution of the United States: tion of the United States, to settle on procedures The Congress shall have power To coin money, regulate for the preparation of prototype metric standards. the value thereof, and of foreign coin, and fix the standard Then, on May 20, 1875, the United States partici- of weights and measures pated in the Convention of the Meter in Paris and 188 Audretsch et al. was one of the eighteen signatory nations to the level of funding for such a laboratory, the impor- Treaty of the Meter. tance of the laboratory was not debated. Finally, In a Joint Resolution before Congress on the Act of 3 March 1901, also known as the March 3, 1881, it was resolved that: Organic Act, established the National Bureau of The Secretary of the Treasury be, and he is hereby directed Standards within the Department of the Treasury, to cause a complete set of all the weights and measures where the Office of Standard Weights and Mea- adopted as standards to be delivered to the governor of sures was administratively located: each State in the Union, for the use of agricultural colleges in the States, respectively, which have received a grant of Be it enacted by the Senate and House of Representatives of lands from the United States, and also one set of the same the United States of America in Congress assembled, That the for the use of the Smithsonian Institution. Office of Standard Weights and Measures shall hereafter be known as the National Bureau of Standards That the Then, the Act of 11 July 1890 gave authority to functions of the bureau shall consist in the custody of the the Office of Construction of Standard Weights standards; the comparison of the standards used in scien- and Measures (or Office of Standard Weights and tific investigations, engineering, manufacturing, commerce, and educational institutions with the standards adopted or Measures), which had been established in 1836 recognized by the Government; the construction, when nec- within the Treasury’s Coast and Geodetic Survey: essary, of standards, their multiples and subdivisions; the testing and calibration of standard measuring apparatus; For construction and verification of standard weights and the solution of problems which arise in connection with measures, including metric standards, for the custom- standards; the determination of physical constants and the houses, and other offices of the United States, and for the properties of materials, when such data are of great impor- several States tance to scientific or manufacturing interests and are not to The Act of 12 July 1894 established standard be obtained of sufficient accuracy elsewhere. units of electrical measure: The Act of 14 February 1903, established the Be it enacted , That from and after the passage of this Department of Commerce and Labor, and in that Act the legal units of electrical measure in the United Act it was stated that the National Bureau of Stan- States shall be as follows: That it shall be the duty dards be moved from the Department of the Trea- of the National Academy of Sciences [established in 1863] to prescribe and publish, as soon as possible after the pas- sury to the Department of Commerce and Labor. sage of this Act, such specifications of detail as shall be Then, in 1913, when the Department of Labor was necessary for the practical application of the definitions of established as a separate entity, the Bureau was the ampere and volt hereinbefore given, and such specifica- formally housed in the Department of Commerce. tions shall be the standard specifications herein mentioned. In the post World War I years, the Bureau’s research focused on assisting in the growth of Following from a long history of our Nation’s industry. Research was conducted on ways to leaders calling for uniformity in science, trace- increase the operating efficiency of automobile able at least to the several formal proposals for and aircraft engines, electrical batteries, and gas a Department of Science in the early 1880s, and appliances. Also, work was begun on improv- coupled with the growing inability of the Office ing methods for measuring electrical losses in of Weights and Measures to handle the explosion response to public utility needs. This latter of arbitrary standards in all aspects of federal and research was not independent of international state activity, it was inevitable that a standards lab- efforts to establish electrical standards similar to oratory would need to be established. The politi- those established over 50 years earlier for weights cal force for this laboratory came in 1900 through and measures. Lyman Gage, then Secretary of the Treasury under After World War II, significant attention and President William McKinley. Gage’s original plan resources were devoted to the activities of the was for the Office of Standard Weights and Mea- Bureau. In particular, the Act of 21 July 1950 sures to be recognized as a separate agency called established standards for electrical and photomet- the National Standardizing Bureau. This Bureau ric measurements: would maintain custody of standards, compare Be it enacted by the Senate and House of Representatives of standards, construct standards, test standards, and the United States of America in Congress assembled, That resolve problems in connection with standards. from and after the date this Act is approved, the legal units Although Congress at that time wrestled with the of electrical and photometric measurements in the United The Economics of Science and Technology 189

States of America shall be those defined and established as The economics of standards.54 An industry stan- provided in the following sections. The unit of electri- dard is a set of specifications to which all ele- cal resistance shall be the ohm The unit of electrical ments of products, processes, formats, or proce- current shall be the ampere The unit of electromotive force and of electrical potential shall be the volt The dures under its jurisdiction must conform. The unit of electrical quantity shall be the coulomb The process of standardization is the pursuit of this unit of electrical capacity shall be the farad The unit conformity, with the objective of increasing the of electrical inductance shall be the henry The unit of efficiency of economic activity. power shall be the watt The units of energy shall be The complexity of modern technology, espe- the (a) joule and (b) the kilowatt-hour The unit of intensity shall be the candle The unit of flux light cially its system character, has lead to an increase shall be the lumen It shall be the duty of the Secretary in the number and variety of standards that affect of Commerce to establish the values of the primary elec- a single industry or market. Standards affect the tric and photometric units in absolute measure, and the R&D, production, and market penetration stages legal values for these units shall be those represented by, of economic activity and therefore have a signif- or derived from, national reference standards maintained by the Department of Commerce. icant collective effect on innovation, productivity, and market structure. Thus, a concern of govern- Then, as a part of the Act of 20 June 1956, the ment policy is the evolutionary path by which a Bureau moved from Washington, DC to Gaithers- new technology or, more accurately, certain ele- burg, Maryland. The responsibilities listed in the ments of a new technology become standardized. Act of 21 July 1950, and many others, were Standardization can and does occur without transferred to the National Institute of Stan- formal promulgation as a standard. This distinc- dards and Technology (NIST) when the National tion between de facto and promulgated standards Bureau of Standards was renamed under the is important more from an institutional process guidelines of the Omnibus Trade and Competi- than an economic impact perspective. In one tiveness Act of 1988: sense, standardization is a form rather than a type The National Institute of Standards and Technology of infrastructure because it represents a codifi- [shall] enhance the competitiveness of American industry cation of an element of an industry’s technology while maintaining its traditional function as lead national or simply information relevant to the conduct of laboratory for providing the measurement, calibrations, economic activity. And, because the selection of and quality assurance techniques which underpin United one of several available forms of a technology ele- States commerce, technological progress, improved product reliability and manufacturing processes, and public ment as the standard has potentially important safety [and it shall] advance, through cooperative economic effect, the process of standardization is efforts among industries, universities, and government lab- important. oratories, promising research and development projects, While economics is increasingly concerned with which can be optimized by the private sector for commer- standards due to their proliferation and perva- cial and industrial applications [More specifically, NIST is to] prepare, certify, and sell standard reference materials siveness in many new high-technology industries, for use in ensuring the accuracy of chemical analyses and the economic roles of standards are unfortunately measurements of physical and other properties of materials poorly understood. Standards can be grouped into two basic categories: product-element standards and nonproduct-element standards. This distinc- The organizational structure at NIST. NIST’s mis- tion is important because the economic role of sion is to promote U.S. economic growth by work- each type is different. ing with industry to develop and apply technology, Product-element standards typically involve one measurements, and standards. It carries out this of the key attributes or elements of a product, mission through four major programs including as opposed to the entire product. In most cases, ATP, but its centerpiece program is the measure- market dynamics determine product-element stan- ment and standards laboratories program. It pro- dards. Alternative technologies compete intensely vides technical leadershipfor vital componentsof until a dominant version gains sufficient market the nation’s technology infrastructure needed by share to become the single de facto standard. Mar- U.S. industry to continually improve its products ket control by one firm can truncate this com- and services. petitive process. Conversely, nonproduct-element 190 Audretsch et al. standards tend to be competitively neutral within Strategic direction of the firm the context of an industry. This type of stan- Entrepreneurial response Innovation Value Added Competitive market dard can impact an entire industry’s efficiency conditions and its overall market penetration rate. Industry In-house R&D organizations often set nonproduct-element stan- Research Spillovers to society Partnerships dards using consensus processes. The technical Infrastructure bases (infratechnologies) for these standards have Technology

Science a large public good content. Examples include Base measurement and test methods, interface standards, and standard reference materials. External influences on economic performance From both the positions of a strategically- focused firm as well as a public policy maker, stan- Figure 9. The entrepreneurial process: an integrated look. dardization is not an all-or-nothing proposition. In complimented system technologies, such as dis- many innovations diffuse into society and generate tributed data processing, telecommunications, or spillover benefits to other firms in outside indus- factory automation, standardization typically pro- tries. Finally, the science base also influences the ceeds in an evolutionary matter in lock stepwith level of R&D activity. The science base conceptu- the evolution of the technology. Complete stan- alizes the stock of knowledge generated from basic dardization too early in the technology’s life cycle research. The science base resides in the public can constrain innovation. domain—and the public domain is international The overall economic value of a standard is in scope—generally in the form of scientific jour- determined by its functionality (interaction with nals but also it is in part embodied in university other standards at the systems level) and the cost scientists. of implementation (compliance costs). Standards Two internal feedback mechanisms, depicted by should be competitively neutral, which means dashed lines, are also added to Figure 9. The adaptable to alternative applications of a generic first feedback in the model flows from innovation technology over that technology’s life cycle. to the science base. Once an innovation exists, knowledge has been created and it too will reside in the public domain. 12. Toward a more integrated The second internal feedback extends from entrepreneurial process innovation to competitive market conditions. It An integrated entrepreneurial process is illus- reflects the extent to which innovation can alter trated in Figure 9. As in the earlier schematics the competitive landscape. This can be seen, for in Figures 1 and 4, the strategic direction of the example, in the evolution of industries, where new firm and the competitive pressures that it faces technologies eclipse old.55 Although profit oppor- motivate an entrepreneurial response. R&D activ- tunities create an incentive for innovation, inno- ity is the primary resource that the firm relies upon vation can subsequently alter the structure of a to investigate the appropriate response and to act market and in so doing, the profit opportunities upon it, but Figure 9 includes additions to the ear- in that marketplace. Generally speaking, market lier schematics that reflect the complexity of the structure and market behavior are jointly endoge- entrepreneurial process. nous. More recent literature on industry structure, One addition to Figure 9 is the inclusion of applied to issues such as innovation, often uses several factors that influence the level of in-house a game-theory approach to separate out endoge- R&D. The first of these is the infrastructure tech- nous effects from true exogenous conditions.56 nology that comes from federal laboratories, such Forward-looking entrepreneurial strategy takes as NIST, or from the environment created by being seriously the feedback from innovation to mar- located in a science park. The second is involve- ket conditions. Such strategy shapes the nature ment in research partnerships, with other firms or of rivalry in the innovation process. For example, perhaps with either a university or a federal lab- in one scenario, a significant innovation can cre- oratory. The third addition is the recognition that ate a monopoly for the successful entrepreneur, The Economics of Science and Technology 191 either due to a patent award or from the fail- The majority of doctoral scientists trained in ure of rivals to quickly imitate. In this case, the United States are employed in institutions absent other market failures such as spillovers, the of higher education although over time, indus- profit incentive likely leads to intense competition try is employing more and more scientists. While between competing entrepreneurs in their “race” academe once employed 60 percent of all Ph.D. to innovate. It follows that the initial competi- scientists, this percentage has been below 50 since tive landscape ultimately changes to a monopoly. 1990. In contrast, industry, which used to employ Interestingly, game-theory modeling shows that fewer than one in four scientists, now employs excessive duplicative R&D may occur in this sce- approximately one in three. nario. However, as discussed in previous sections, The production or supply of new doctorates and confirmed by game theoretic models, an oppo- in the U.S. can be summarized in terms of the site result can occur if significant spillovers of ratio of Ph.D.s granted to U.S. citizens and perma- knowledge flow from one firm to another. That nent residents to the U.S.-population aged 25–34. is, strategic R&D competition may yield too lit- The proportion of 25–34 aged individuals receiv- tle innovative activity because the flow of knowl- ing a Ph.D. in both the physical and life sciences edge to competitors has a negative effect on the increased throughout the 1960s, declined in the entrepreneur’s position.57 1970s, was fairly stable in the 1980s, and again The usefulness of Figure 9 is not only as a sum- increased during the 1990s. mary device but also it is a means to highlight the myriad sources of scientific and technical information that firms rely on to support their innovative activity. Certainly, not all firms rely on each source Foreign-born scientists and engineers to the same degree. Larger firms in competitive Approximately 18 percent of the highly-skilled environments generally rely more heavily on their in-house R&D than smaller firms. Small firms rely (doctoral or medical degree) scientists in the U.S. 59 more on external sources of technical expertise. in 1980 were foreign-born. The percentage of Figure 9 is also a useful device for summa- foreign-born was highest among physical scien- rizing public policies toward R&D. The patent tists (20.4 percent) and lowest among life sci- system and the R&E tax credit provide direct entists (15.4 percent). By 1990, the proportion incentives to the firm to increase its level of foreign-born increased to nearly 25 percent. More R&D. The NCRA and the NCRPA affect the effi- than one in four physical scientists and math ciency of in-house R&D by reducing duplicative and computer scientists working in the U.S. were research costs and shortening the fundamental born abroad. For life scientists, the proportion research stage. Government-provided infrastruc- had increased from approximately one in seven ture technology through standards reduces trans- to one in five. The proportion of engineers who action costs in the market, thus lowering the are foreign born is substantially smaller than that marginal cost of R&D. Finally, federal support of highly-trained (bachelor’s degree) scientists. In of university research continually enriches the sci- 1980 approximately 14 percent were foreign-born; ence base thus facilitating the in-house R&D this had crept up to about 16 percent by 1990. process. A large number of immigrants who receive their doctoral training abroad initially come to the United States to take postdoctoral positions. For 13. Labor market for R&D scientists example, of the 14,918 postdoctoral appointees and engineers training in the life sciences in doctorate-training Key inputs into the R&D process are scientists institutions in the United States in 1996, NSF esti- and engineers (S&E) working in R&D laborato- mates that 7,425—almost exactly half—were not ries. Because of the need for these individuals to U.S. citizens.60 In the physical sciences the propor- be highly trained, the relevant segment of that tion was even higher, at 57 percent. While some workforce consists of those who have received a of these are permanent non-residents, a number doctorate or are studying for it.58 come as temporary residents. 192 Audretsch et al.

The effect of policy to stimulate R&D on S&E workforce are less than efﬁcient. Possible solutions labor markets to these inefﬁciencies include:

The federal government has several policy mech- • The provision of information concerning em- anisms to stimulate R&D activity. In industry, ployment opportunities in industry to aspiring research is stimulated through the use of the students as well as to students enrolled in pro- R&E tax credit, as well as through direct subsi- grams. dies. University research is encouraged through • The creation of training grants which fund the the provision of a large funding pool for fac- student, instead of the faculty member, thereby ulty grants. Depending on the sector targeted, the providing students greater independence. labor market consequences of these policies vary • The reshaping of graduate education along the considerably. lines followed by professional schools, includ- The government’s policy intent to stimulate ing the provision of information about career R&D in the private sector increases the demand prospects and the provision of opportunities to for R&D processes and, by extension, the demand work in industry while in graduate school.62 for highly-trained scientists and engineers. This increase in demand, however, only results in an Forecasting scientific labor markets increase in the number of S&E workers in industry if the supply of S&E workers is not fixed but Although economists’ models of scientific labor instead is responsive to higher wage rates. If the markets have been somewhat successful in provid- supply of S&E workers is not wage responsive, ing insight into factors affecting demand and sup- then public policies to stimulate R&D will only ply, reliable forecasts of scientific labor markets have a compensation effect rather than an employ- do not exist, partly because of the unavailability ment effect. of reliable predictions of variables affecting sup- The government policy to support R&D in the ply and demand.63 While this problem is endemic university sector, however, can affect S&E labor to forecasting in general, the ups and downs of markets quite differently. This is because gradu- federal funding make forecasts of scientific labor ate students are especially responsive to the level markets particularly unreliable. and availability of research support, and a signif- Despite these problems, forecasts of scientific icant portion of the R&D funding that supports labor markets are somewhat common, in part university research provides support for graduate because they are mandated by Congress, suppos- research assistants. edly in an effort to keepthe United States’ innova- There are a number of barriers to attracting tion process healthy and its industry competitive. S&E doctorates to industry:61 14. Public accountability64 • A lack of good information to graduate students concerning the rewards to working in The concept of public accountability can be industry. traced as far back as President Woodrow Wil- • Attitudes among faculty that employment in son’s reforms, and in particular to the Budget industry makes the student a second-class sci- and Accounting Act of 1921. This Act of June 10, entist. 1921 not only required the President to transmit to • Opportunities for recent Ph.D.s to remain in Congress a detailed budget on the first day of each academe and work as a post doctorate fellows. regular session, but also it established the General • Funding patterns that tie students and post doc- Accounting Office (GAO) to settle and adjust all torates to their mentor’s laboratory, thereby accounts of the government. We note this fiscal decreasing the opportunity to have differ- accountability origin because the GAO has had a ent research experiences and potentially learn significant role in the evolution of accountability- more about positions in industry. related legislation during the past decade. What follows is a review the legislative history The existence of such barriers suggests that of legislation that falls broadly under the rubric of government policies as they relate to the S&E public accountability. As Collins (1997, p. 7) notes: The Economics of Science and Technology 193

As public attention has increasingly focused on improv- and in each major executive agency in the Fed- ing the performance and accountability of Federal pro- eral Government. grams, bipartisan efforts in Congress and the White House 2. Provide for improvement, in each agency of the have produced new legislative mandates for management Federal Government, of systems of accounting, reform. These laws and the associated Administration and Congressional policies call for a multifaceted approach— financial management, and internal controls to including the provision of better financial and performance assure the issuance of reliable financial infor- information for managers, Congress, and the public and mation and to deter fraud, waste, and abuse of the adoption of integrated processes for planning, manage- Government resources. ment, and assessment of results. 3. Provide for the production of complete, reli- Fundamental to any evaluation of a public insti- able, timely, and consistent financial informa- tution is the recognition that the institution is tion for use by the executive branch of the accountable to the public, that is to taxpayers, Government and the Congress in the financing, for its activities. With regards to technology-based management, and evaluation of Federal pro- institutions, this accountability refers to being able grams. to document and evaluate research performance The key phrase in these stated purposes is in using metrics that are meaningful to the institu- point (3) above, “evaluation of Federal programs.” tions’ stakeholders, meaning to the public. Toward this end, the Act calls for the establishment of agency Chief Financial Officers, where agency is defined to include each of the Fed- Performance accountability eral Departments. And, the agency Chief Finan- Chief Financial Officers Act of 1990. The GAO cial Officer shall, among other things, “develop has a long-standing interest and a well-docu- and maintain an integrated agency accounting and mented history of efforts to improve governmen- financial management system, including financial tal agency management through performance mea- reporting and internal controls,” which, among surement. For example, in February 1985 the other things, “provides for the systematic mea- GAO issued a report entitled “Managing the Cost surement of performance.” of Government—Building An Effective Financial While the Act does outline the many fiscal Management Structure” which emphasized the responsibilities of agency Chief Financial Officers, importance of systematically measuring perfor- and their associated auditing process, the Act’s mance as a key area to ensure a well-developed only clarification of “evaluation of Federal pro- financial management structure. grams” is in the above phrase, “systematic mea- On November 15, 1990, the 101st Congress surement of performance.” However, neither a passed the Chief Financial Officers Act of 1990. definition of “performance” nor guidance on “sys- As stated in the legislation as background for this tematic measurement” is provided in the Act. Still, Act: these are the seeds for the growth of attention to performance accountability. The Federal Government is in great need of fundamental reform in financial management requirements and practices Government Performance and Results Act of 1993. as financial management systems are obsolete and ineffi- Legislative history is clear that the Government cient, and do not provide complete, consistent, reliable, and timely information. Performance and Results Act (GPRA) of 1993 builds upon the February 1985 GAO report The stated purposes of the Act are to: and the Chief Financial Officers Act of 1990. The 103rd Congress stated in the August 3, 1993 1. Bring more effective general and financial man- legislation that it finds, based on over a year of agement practices to the Federal Government committee study, that: through statutory provisions which would establish in the Office of Management and Bud- 1. waste and inefficiency in Federal programs get a Deputy Director for Management, estab- undermine the confidence of the American lish an Office of Federal Financial Management people in the Government and reduce the Fed- headed by a Controller, and designate a Chief eral Government’s ability to address adequately Financial Officer in each executive department vital public needs; 194 Audretsch et al.

2. Federal managers are seriously disadvantaged prepare an annual performance plan [beginning with in their efforts to improve program efficiency fiscal year 1999] covering each program activity set forth and effectiveness, because of insufficient artic- in the budget of such agency. Such plan shall establish ulation of program goals and inadequate infor- performance indicators to be used in measuring or assessing the relevant outputs, service levels, and outcomes of mation on program performance; and each program activity; 3. congressional policymaking, spending decisions and program oversight are seriously handi- where “performance indicator means a particular capped by insufficient attention to program per- value or characteristic used to measure output or formance and results. outcome.” Cozzens (1995) correctly notes that one fear Accordingly, the purposes of GPRA are to: about GPRA is that it will encourage agencies 1. improve the confidence of the American peo- to ignore what is difficult to measure, no mat- ple in the capability of the Federal Govern- ter how relevant. Alternatively, one could wear ment, by systematically holding Federal agen- a more pessimistic hat and state that GPRA will cies accountable for achieving program results; encourage agencies to emphasize what is easy to 2. initiate program performance reform with a measure, no matter how irrelevant. series of pilot projects in setting program goals, measuring program performance against those goals, and reporting publicly on their progress; Fiscal accountability 3. improve Federal program effectiveness and Legislation following GPRA emphasizes fiscal public accountability by promoting a new focus accountability more than performance account- on results, service quality, and customer satis- ability. While it is not our intent to suggest that faction; 4. helpFederal managers improveservice deliv- performance accountability is more or less impor- ery, by requiring that they plan for meeting tant than fiscal accountability, for we believe that program objectives and by providing them with both aspects of public accountability are impor- information about program results and service tant, the emphasis in the case studies conducted at quality; NIST that are summarized in this paper is on per- 5. improve congressional decisionmaking by pro- formance accountability. Nevertheless, our discus- viding more objective information on achieving sion would not be complete in this chapter with- statutory objectives, and on the relative effec- out references to the Government Management tiveness and efficiency of Federal programs and Reform Act of 1994 and the Federal Financial spending; and Management Improvement Act of 1996. 6. improve internal management of the Federal Government. Government Management Reform Act of 1994. The Act requires that the head of each agency The Government Management Reform Act of submit to the Director of the Office of Manage- 1994 builds on the Chief Financial Officers Act ment and Budget (OMB): of 1990. Its purpose is to improve the management of the federal government though reforms to no later than September 30, 1997 a strategic plan for program activities. Such plan shall contain a the management of federal human resources and description of the program evaluations used in establishing financial management. Motivating the Act is the or revising general goals and objectives, with a schedule for belief that federal agencies must streamline their future program evaluations. operations and must rationalize their resources to And, quite appropriately, the Act defines pro- better match a growing demand on their services. gram evaluation to mean “an assessment, through Government, like the private sector, must adopt objective measurement and systematic analysis, of modern management methods, utilize meaningful the manner and extent to which federal programs program performance measures, increase work- achieve intended objectives.” In addition, each force incentives without sacrificing accountability, agency is required to: and strengthen the overall delivery of services. The Economics of Science and Technology 195

Federal Financial Management Improvement Act people in public sector research be strengthened of 1996. The Federal Financial Management by simply comparing benefits to costs? Improvement Act of 1996 follows from the belief We now consider two different approaches that federal accounting standards have not been to program evaluation. It appears that the best implemented uniformly through federal agen- evaluation technique for publicly-funded, publicly- cies. Accordingly, this Act establishes a uni- performed research is based on a counterfactual form accounting reporting system in the federal method. In contrast, we conjecture that the best government. evaluation method for publicly-funded, privately- This overview of what we call public account- performed research consists of an analysis of ability legislation makes clear that government spillovers. These techniques are described in the agencies are becoming more and more account- following sections. able for their fiscal and performance actions. And, these agencies are being required to a greater Traditional evaluation methods. Griliches (1958) degree than ever before to account for their activities through a process of systematic measure- and Mansfield et al. (1977) pioneered the applica- ment. For technology-based institutions in particu- tion of fundamental economic insight to the devel- lar, internal difficulties are arising as organizations opment of measurements of private and social learn about this process. rates of return to innovative investments. Streams Compliance with these guidelines is causing of investment outlays through time—the costs— increased planning and impact assessment activ- generate streams of economic surplus through ity and is also stimulating greater attention to time—the benefits. Once identified and measured, methodology. Perhaps there is no greater vali- these streams of costs and benefits are used to dation of this observation than the diversity of calculate rates of return, benefit-to-cost ratios, or response being seen among public agencies, in other related metrics. general, and technology-based public institutions, In the Griliches/Mansfield model, the innova- in particular, as they grope toward an understand- tions evaluated are conceptualized as reducing ing of the process of documenting and assess- the cost of producing a good sold in a coming their public accountability. Activities in recent petitive market at constant unit cost. For any years have ranged from interagency discussion period, there is a demand curve for the good, meetings to a reinvention of the assessment wheel, representing its marginal benefit to consumers, so to speak, in the National Science and Tech- and a horizontal supply curve. Innovation low- nology Council’s (1996) report, “Assessing Funda- ers the unit cost of production, shifting down- mental Science.” ward the horizontal supply curve and thereby, at the new lower equilibrium price, resulting in greater consumer surplus (economists’ measure Systematic approaches to the evaluation of of the value of the difference between the price technology-based programs consumers are willing to pay and the price they GPRA is directionally, as opposed to methodolog- actually pay, summed over all purchases). Addi- ically, clear about the evaluation process; pub- tionally, the Griliches/Mansfield model accounts lic institutions/research programs will identify out- for producer surplus, measured as the difference puts and quantify the economic benefits of the between the price the producers receive and the outcomes associated with such outputs. In our actual marginal cost, summed over the output opinion, agencies should attempt to quantify out- sold, minus any fixed costs. Social benefits are then come benefits and then compare those quanti- the streams of new consumer and producer sur- fied benefits to the public costs to achieve the pluses, while private benefits are the streams of benefits. Although these are GPRA’s directions, producer surplus, not all of which are necessarily the methodological hurdle that has been plagu- new because the surplus gained by one producer ing most public agencies is how to quantify ben- may be cannibalized from the pre-innovation sur- efits. And even with an acceptable quantification plus of another producer. Social and private costs of benefits, will the confidence of the American will, in general, also be divergent. 196 Audretsch et al.

The Griliches/Mansfield model for calculating The spillover evaluation method. There are impor- economic social rates of return add the public tant projects where economic performance can and the private investments through time to deter- be improved with public funding of privately- mine social investment costs, and then the stream performed research. Public funding is needed of new economic surplus generated from those when socially valuable projects would not be investments is the benefit. Thus, the evaluation undertaken without it. If the expected private rate question that can be answered from such an eval- of return from a research project falls short of uation analysis is: What is the social rate of return the required rate called the hurdle rate, then the to the innovation, and how does that compare to private sector firm will not invest in the project. the private rate of return? We argue that this is Nonetheless, if the benefits of the research spill not the most appropriate question to ask from a over to consumers and to firms other than the public accountability perspective. The fact that the ones investing in the research, the social rate of social rate of return is greater than the private return may exceed the appropriate hurdle rate. It would then be socially valuable to have the invest- rate of return may validate the role of government ments made, but since the private investor will in innovation if the private sector would not have not make them, the public sector should. By pro- undertaken the research; but it ignores, for exam- viding some public funding, thereby reducing the ple, consideration of the cost effectiveness of the investment amount needed from the private firm public sector undertaking the research as opposed or firms doing the research, the expected private to the private sector. rate of return can be increased above the hurdle rate. Thus, because of this subsidy, the private firm The counterfactual evaluation method. A dif- is willing to perform the research, which is socially ferent question should be considered when desirable because much of its output spills over to publicly-funded, publicly-performed investments other firms and sectors in the economy. are evaluated. Holding constant the very stream The question asked in the spillover method is of economic surplus that the Griliches/Mansfield one that facilitates an economic understanding of model seeks to measure, and making no attempt whether the public sector should be underwriting a to measure that stream, one should ask the coun- portion of private-sector firms’ research, namely: terfactual question: What would the private sector What proportion of the total profit stream gen- have had to invest to achieve those benefits in the erated by the private firm’s R&D and innovation absence of the public sector’s investments? The does the private firm expect to capture; and hence, answer to this question gives the benefits of the what proportion is not appropriated but is instead public’s investments—namely, the costs avoided captured by other firms that imitate the innovation or use knowledge generated by the R&D to by the private sector. With those benefits— produce competing products for the social good? obtained in practice through extensive interviews The part of the stream of expected profits cap- with administrators, federal research scientists, tured by the innovator is its private return, while and those in the private sector who would have the entire stream is the lower bound on the social to duplicate the research in the absence of public rate of return. In essence, this method weighs the performance—counterfactual rates of return and private return, estimated through extensive inter- benefit-to-cost ratios can be calculated to answer views with firms receiving public support about the fundamental evaluation question: Are the pub- their expectations of future patterns of events and lic investments a more efficient way of generat- future abilities to appropriate R&D-based knowl- ing the technology than private sector investments edge, against private investments. The social rate would have been? The answer to this question is of return weights the social returns against the more in line with the public accountability issues social investments. implicit in GPRA, and certainly is more in line The application of the spillovers model to the with the thinking of public sector stakeholders, evaluation of public funding/private performance who may doubt the appropriateness of govern- of research is appropriate since the output of the ment’s having a role in the innovation process in research is only partially appropriable by the pri- the first place. vate firm with the rest spilling over to society. The The Economics of Science and Technology 197 extent of the spillover of such knowledge with pub- patents and intellectual property included rela- lic good characteristics determines whether or not tively few important contributions among applied the public sector should fund or partially fund the economists until fairly recently. Patent databases research. and measurement techniques have greatly facili- tated the field of study, but equally important has been the influence of public policy. With changes Program evaluation in the laws and statutes pertaining to patents, and particularly the ownership of intellectual property Many public technology-based institutions have among university and government researchers, the conducted program evaluations both as part of topic has become manifestly more important dur- their overall management of the program and in ing recent years. Similarly, the study of domestic response to GPRA. technology transfer could barely be considered a The U.S. General Accounting Office (GAO) field even twenty years ago. But the convergence monitors the performance evaluation progress of research techniques and policy initiative has led of federal government agencies. Based on their to a growth of research on this topic. assessment, “agencies’ fiscal year 2000 perfor- We believe that there are several issues that mance plans show moderate improvements over require further exploration. The first of these is the the fiscal year 1999 plans [h]owever, key relationshipbetween globalization and the devel- weaknesses remain ” (GAO, 1999, p. 3). One opment and ownership of science and technol- weakness is that agencies’ performance data lack ogy. Although there is a long-standing literature credibility. on international technology transfer, the typical Not only is GAO expected to continue to mon- model employed in these studies focuses on under- itor the performance of research programs, but standing relations between a technology or knowl- also GPRA-like frameworks are beginning to be edge donor nation and a recipient nation. Unfor- used at the state level. tunately, this type of model is not as relevant or useful for understanding the complexity of the current state of technology transfer. That is because 15. Conclusions there are now numerous instances where the firms’ Our primary purpose in writing this primer was nation of charter is almost inconsequential, where to survey the landscape of topics that are related capital flows in a dizzying array of vehicles, insti- to the field of economics of science and technol- tutions and multinational forums, and where tech- ogy. We have provided this survey based on histor- nology is simultaneously marketed along different ical and applied perspectives, all within the context channels with different firms in a great number of of what we call the entrepreneurial process. We countries. The complexity of this process does not believe that this material is useful because most lend itself easily to current models. students and policymakers are not well informed Another area that should prove fruitful for stu- about the economics of science and technology, dents of the economics of science and technology despite the fact that the field’s knowledge includes is e-commerce. Indeed, there is already an increase many issues fundamental to economic growth. in interest in electronic commerce and funding After reading this survey, we hope that readers agencies such as the National Science Founda- will reach many of the same conclusions as we did tion are girding themselves to support programs of concerning the state of the field, its emphases and, study for this topic. But thus far e-commerce has just as important, the gaps in knowledge. In our not been a topic of much interest to more than judgment, and from the perspective of researchers a handful of researchers using the theories and in this field, the study of the economics of science tools of the economics of science and technology. and technology has made important advances in We suspect that this will change in the next few the past two decades or so. In some instances, years. the areas of study we examined here were essen- A third area that requires more attention is tially unknown as formal research topics just a the intersection between the economics of science few short years ago. For example, the study of and technology and distributional and social equity 198 Audretsch et al. issues (beyond the question of how new technol- 9. One could make an argument that Jefferson’s funding of ogy affects the workforce (Siegel, 1999). One of Lewis and Clark was the first instance of public support for the most interesting questions of the economics pure research, whereas Morse was funded to conduct applied research. There are other historical examples of governmental of science and technology is who “wins” and who support to individuals for research that has the potential to “loses” with innovation and the introduction of benefit society, such as the Longitude Act of 1714. The British new technology. We have, for example, begun to Parliament offered a prize (equal to several million dollars in talk about the “digital divide,” but there are also today’s terms) for a practicable solution for sailing vessels to health care technology divides, among others. If determine longitude (Sobel, 1995). 10. This presumption has been the genesis for the formation we understand the economic forces that allow us of many science parks around the world. See Link (1995). efficiently to produce and market health care tech- 11. An excellent history of the growth of U.S. industrial nologies and pharmaceuticals but do not under- research organizations is in Hounshell (1996). standing the distributional issues relating to access 12. The Allison Comission failed in 1884 to formulate an technology, it is possible to encourage, at the same infrastructure to undertake this task. See above. 13. Historical data on the number of industrial research labo- time, more and more innovation and greater divi- ratories is less than precise. Crow and Bozeman (1998) report, sions in society. based on secondary sources, that in 1920 there were about 500 These are some of the “big questions.” Unfor- research laboratories, and just over 1,000 in 1930. Regard- tunately, the field is not yet prepared to provide less of the precise number that existed in 1920, the direc- much evidence on these questions. But as needs tion of growth in industrial research laboratories since then is clear. change and resources shift, tomorrow’s economics 14. The term “basic research” is credited to Vannevar Bush. of science and technology will almost certainly look He proffered the definition: “Basic research is performed with- very different than today’s. out thought of practical ends.” “Basic research” thus is equivalent to “science.” 15. See National Science Board (2000) for background information. Acknowledgments 16. See National Science Board (2000). We are grateful for comments and suggestions by 17. Gross Domestic Product (GDP) is the value of all the goods and services produced in an economy. In order to make John Jankowski, Jamie Link, and John Scott on cross-country comparisons we normalize GDP by dividing for earlier versions of this paper. All remaining short- the size of the country’s population. This adjustment allows a comings are our own. comparison of the standard of living for the average person. Economic growth is often thought of in terms of increases in GDP per capita over time. 18. In a sense, A in Equation (3) is a residual. It captures Notes changes in output over time that are not explained by changes in inputs. 1. We do not emphasize the analytical development of the 19. See Link (1996b) for a detailed history of this classifica- academic literature herein, but rather we summarize salient tion scheme. conclusions from the literature so as to provide a broad con- 20. These descriptions come from Science and Engineering text for understanding attendant policies. Indicators—2000, and they are published in other National 2. When the innovation is itself the final marketable result, Science Foundation documents. it is sometimes referred to as a product innovation. When the 21. This model was first formalized by Griliches (1979). innovation is applied in a subsequent production process it is 22. See Kerr (1994) and Clark (1995). sometimes referred to as a process innovation (meaning that 23. Researchers at that time were quick to point out that its application affects a production process). the slowdown in total factor productivity in the early 1970s 3. It is not uncommon to see this process referred to as the was preceded by several years of flat and even declining R&D innovation process. spending in the economy. 4. See Hébert and Link (1988) for a complete history of the 24. Lichtenberg and Siegel (1991) reported that the returns concept of the entrepreneur. to company-funded R&D remained high during the produc- 5. See Cohen and Levinthal (1989) for a detailed develop- tivity slowdown of the 1970s. ment of this idea. 25. While the historical overview presented above indicates 6. This section draws directly from UNESCO (1968) and that the roots of our national science and technology efforts National Science Board (2000). pre-date Bush’s report, Science—The Endless Frontier remains 7. There is a strong similarity between this early public/ the frequently heralded origin. private partnership and the establishment of the land-grant 26. See Tassey (1999) and National Science Board (2000). college system under the Morrill Act in 1862. 27. See Tassey (1997), and Link and Scott (2001) for a com- 8. Congress passed the first patent act in 1790. plete discussion of variations of this graphical model. The Economics of Science and Technology 199

28. Kaufer (1989) provides an excellent historical perspective 41. See Hall et al. (2001) for a discussion of barriers that on patents throughout the history of modern civilization. prevent firms from partnering with universities. 29. The first U.S. patent was issued in 1790. 42. However, the range is large. The number of universities 30. To be granted a patent, three criteria must hold: utility, in those joint ventures with universities as research partners novelty, and non-obviousness. Utility means that the invention ranges from one to over 100. should be useful; novelty means that the invention be new and 43. See Hall et al. (2000). not merely a copy or repetition of another invention; and, 44. See Leyden and Link (1999). non-obviousness—the most difficult criterion—means that the 45. This section draws from Link (1999). invention is neither suggested by previous work nor totally 46. For more information about SEMATECH, see www. anticipated given existing practices. sematech.org. 31. This section draws from Bozeman and Link (1984). 47. The thirteen charter members of SEMATECH were: 32. There is a slight distinction between R&D expenditures Advanced Micro Devices, AT&T, Digital Equipment Cor- from a NSF-reporting perspective and R&E expenditures from poration, Harris Corporation, Hewlett-Packard Company, a tax perspective. R&E expenditures are somewhat more nar- IBM Corporation, Intel Corporation, LSI Logic Corporation, rowly defined to include all costs incident to development. Micron Technology, Inc., Motorola, Inc., National Semicon- R&E does not include ordinary testing or inspection of mate- ductor Corporation, Rockwell International Corporation, and rials or products for quality control of those for efficiency Texas Instruments, Inc. studies, etc. R&E, in a sense, is the experimental portion of 48. This section draws from Tibbetts (1999). R&D. That said, in practice it is often difficult to distinguish 49. As a set aside program, the SBIR program redirects exist- one category from the other. ing R&D rather than appropriating new monies for R&D. 33. More recently, Hall and van Reenen (2000) conclude 50. See Audretsch et al. (2002). from their review of the literature that the tax elasticity of 51. Some of these laboratories are government owned but R&D is about unity, meaning that a 1 percent increase in the contractor operated (GOCO) and others are government credit will increase industry R&D by about 1 percent. owned and government operated (GOGO). 52. Emphasis is placed on direct technology infrastructure 34. See Bozeman and Link (1984). because a significant amount of indirect technology infrastruc- 35. This historical information draws from http://www.src.org. ture comes from all departments and agencies as technological 36. The eleven founding members were Advanced Micro knowledge spillover and is transferred to industry in one form Devices, Control Data Corporation, Digital Equipment Cor- or another. poration, General Instrument, Honeywell, Hewlett-Packard, 53. This historical overview draws from Link and Scott IBM, Intel, Monolithic Memories, Motorola, National Semi- (1998). conductor, and Silicon Systems. 54. This section draws from Tassey (2000). 37. The declining U.S. position in the semiconductor industry 55. See Audretsch (1995). was well known and in other industries there was widespread 56. The market structure and behavior relationshipis concern although the empirical evidence about the competi- described in general in Tirole (1988), pp. 1–4, and in terms of tive position of the United States in international markets was innovative activity in Kamien and Schwartz (1982), Chapter 6. incomplete. However, when the U.S. Department of Com- 57. See Tirole (1988), Chapter 10, for analysis of both inno- merce (1990) released its 1990 report on emerging technolo- vation races and spillovers. gies, it was apparent to all that the concerns expressed in the 58. There are certain fields in S&E where doctoral training early 1980s were quite valid. is not a necessary condition for work in R&D. Engineering is 38. This purpose is stated as a preamble to the Act. a good case in point. Moreover, as innovation becomes more 39. As an illustration of the research activity that can success- and more imbedded in non-R&D functions of the firm, the fully occur through a small, less visible research partnership, doctoral trained workforce may become less key to under- consider the Southwest Research Institute Clean Heavy Diesel standing patterns of innovation. Engine II joint venture, noticed in the Federal Register in 59. If they immigrate prior to receiving their doctoral train- early-1996. The eleven member companies, from six countries ing, and if they indicate at the time that they receive their including the United States, joined together to solve a com- degree that they plan to remain in the U.S., they are mon set of technical problems. Diesel engine manufacturers captured in the National Science Foundation’s Survey of were having difficulties, on their own, meeting desired emis- Doctorate Recipients (the database from where the above sion control levels. The eleven companies were coordinated by statistics came). If, however, they immigrate after receiving Southwest Research Institute, an independent, non-profit con- their doctoral training, they are not included in this database tract research organization in San Antonio, Texas, to col- and are only captured in a sampling frame created every laborate on the reduction of exhaust emissions. The joint ten years, based on the decennial census and known as the research was successful, and each member company took with National Survey of College Graduates (NSCG). See Stephan it fundamental process technology to use in their individ- and Levin (2000) for a discussion. ual manufacturing facilities to meet desired emission con- 60. See National Science Foundation (1998). trol levels. The joint venture was formally disbanded in mid- 61. See Romer (2000) and Stephan and Levin (2000). 1999. 62. See Romer (2000). 40. For a review of the academic literature on RJVs see 63. The variables usually found to affect the supply of Hagedoorn et al. (2000). enrollees (or the number of graduates) in a specific field are 200 Audretsch et al. the salary paid in that field, salary in an alternative occupa- Bozeman, B. and A.N. Link, 1984, ‘Tax Incentives for R&D: tion, such as law or business and (for men) the draft deferment A Critical Evaluation’, Research Policy 13, 21–31. policy. Bozeman, B. and M. Papadakis, 1995, ‘Firms’ Objectives 64. This section draws directly from Link and Scott (1998). in Industry-Federal Laboratory Technology Development Partnerships’, Journal of Technology Transfer 20, 64–74. Bozeman, B. and D. Wittmer, 2002, ‘Technical Roles and Suc- References cess of Federal Laboratory-Industry Partnerships’, Science Abramovitz, M., 1956, ‘Resource and Output Trends in the and Public Policy, in press. United States Since 1870’, American Economic Review Brod, A.C. and A.N. Link, 2001, ‘Trends in Cooperative 46 (2), 5–23. Research Activity: Has the National Cooperative Research Acs, Z. and D.B. Audretsch, 1989, ‘Patents as a Measure of Act Been Successful?’, in M. Feldman and A. Link Innovative Activity’, Kyklos 42 (2), 171–180. (eds.), Innovation Policy in the Knowledge-Based Economy, Acs, Z. and D.B. Audretsch, 1990, Innovation and Small Firms, Norwell, MA: Kluwer Academic Publishers, pp. 105–119. Cambridge, MA: MIT Press. Bush, V., 1945, Science—The Endless Frontier, Washington, Acs, Z., D.B. Audretsch, and M.P. Feldman, 1992, ‘Real DC: U.S. Government Printing Office. Effects of Academic Research: Comment’, American Eco- Carr, R.K., 1995, ‘U.S. Federal Laboratories and Technology nomic Review 82, 363–367. Transfer’, mimeograph. Acs, Z., D.B. Audretsch, and M.P. Feldman, 1994, ‘R&D Clark, B., 1995, Places of Inquiry, Berkeley: University of Spillovers and Recipient Firm Size’, Review of Economics California Press. and Statistics 100, 336–340. Cockburn, I. and Z. Griliches, 1988, ‘Industry Effects and Almeida, P. and B. Kogut, 1997, ‘The Exploitation of Tech- Appropriability Measures in the Stock Market’s Valuation nological Diversity and the Geographical Localization of of R&D and Patents’, American Economic Review Papers Innovation’, Small Business Economics 9, 21–31. and Proceedings 78 (2), 419–423. Association for University Related Research Parks (AURRP), Cohen, W.M. and D.A. Levinthal, 1989, ‘Innovation and 1998, Worldwide Research & Science Park Directory, Wash- Learning: The Two Faces of R&D’, Economic Journal ington, DC: Association for University Related Research 569–596. Parks. Collins, E., 1997, ‘Performance Reporting in Federal Man- Audretsch, D.B., 1995, Innovation and Industry Evolution, agement Reform’, National Science Foundation Special Cambridge, MA: MIT Press. Report, mimeographed. Audretsch, D.B. and M.P. Feldman, 1996, ‘R&D Spillovers and Council on Competitiveness, 1996, Endless Frontiers, Limited the Geography of Innovation and Production’, American Resources: U.S. R&D Policy for Competitiveness, Washing- Economic Review 86 (3), 630–640. ton, DC: Council on Competitiveness. Audretsch, D.B., A.N. Link, and J.T. Scott, 2002, ‘Pub- Cozzens, S.E., 1995, ‘Assessment of Fundamental Science Pro- lic/Private Partnerships: Evaluating SBIR-Supported grams in the Context of the Government Performance and Research’, Research Policy 31 (1), 145–158. Results Act (GPRA)’, Washington, DC: Critical Technolo- Audretsch, D.B. and P.E. Stephan, 1996, ‘Company-Scientist gies Institute. Locational Links: The Case of Biotechnology’, American Crow, M. and B. Bozeman, 1998, Limited by Design: R&D Economic Review 86 (3), 641–652. Laboratories in the U.S. National Innovation System, New Austin, D.H., 1993, ‘An Event Study Approach to Measuring York: Columbia University Press. Innovative Output: The Case of Biotechnology’, American Dunne, T., 1994, ‘Plant Age and Technology Usage in U.S. Economic Review Papers and Proceedings 83 (2), 253–258. Manufacturing Industries’, Rand Journal of Economics 25 Barnett, H.G., 1953, Innovation: The Basis of Cultural Change, New York: W.W. Norton. (3), 488–499. Baudeau, N., 1910 [originally 1767]; in A. Dubois (ed.), Economic Report of the President, 1994, Washington, DC: Gov- Premier Introduction à la Philosophie Économique, Paris: ernment Printing Office. P. Geuthner. Eden, L., E. Levitas, and R.J. Martinez, 1997, ‘The Production, Ben-Zion, U., 1984, ‘The R&D and Investment Decision Transfer and Spillover of Technology: Comparing Large and Its Relationshipto the Firm’s Market Value: Some and Small Multinationals as Technology Producers’, Small Preliminary Results’, in Z. Griliches (ed.), R&D, Patents, Business Economics 9, 53–66. and Productivity, Chicago: University of Chicago Press, pp. Feldman, M.P., 1994, ‘Knowledge Complementarity and Inno- 134–162. vation’, Small Business Economics 6, 363–372. Bound, J., C. Cummins, Z. Griliches, B. Hall, and A. Jaffe, Glaeser, E., H. Kallal, J. Scheinkman, and A. Shleifer, 1992, 1984, ‘Who Does R&D and Who Patents?’, in Z. Griliches ‘Growth of Cities’, Journal of Political Economy 100, (ed.), R&D, Patents, and Productivity, Chicago: University 1126–1152. of Chicago Press, pp. 21–54. Grabowski, H.G., 1968, ‘The Determinants of Industrial Bozeman, B. and A.N. Link, 1983, Investments in Research and Development: A Study of the Chemical, Technology: Corporate Strategies and Public Policy Alterna- Drug, and Petroleum Industries’, Journal of Political Econ- tives, New York: Praeger Publishers. omy 76, 156–159. The Economics of Science and Technology 201

Griliches, Z., 1958, ‘Research Costs and Social Returns: Jaffe, A., 1989, ‘Real Effects of Academic Research’, American Hybrid Corn and Related Innovations’, Journal of Politi- Economic Review 79, 957–970. cal Economy 66 (5), 501–522. Jaffe, A.B., M.S. Fogarty, and B.A. Banks, 1998, ‘Evidence Griliches, Z., 1979, ‘Issues in Assessing the Contribution of from Patents and Patent Citations on the Impact of NASA R&D to Productivity Growth’, Bell Journal of Economics and Other Federal Labs on Commercial Innovation’, Jour- 10, 92–116. nal of Industrial Economics 46 (2), 183–206. Griliches, Z., 1981, ‘Market Value, R&D, and Patents’, Eco- Jaffe, A., M. Trajtenberg, and R. Henderson, 1993, ‘Geograph- nomics Letters 7, 183–187. ical Localization of Knowledge Spillovers as Evidenced Griliches, Z. (ed.), 1984, R&D, Patents, and Productivity, by Patent Citations’, Quarterly Journal of Economics 63, National Bureau of Economic Research for the University 577–598. of Chicago Press, Chicago: University of Chicago Press. Kamien, M.I. and N.L. Schwartz, 1982, Market Structure and Griliches, Z., 1986, ‘Productivity, R&D, and Basic Research at Innovation, Cambridge, UK: Cambridge University Press. the Firm Level in the 1970s’, American Economic Review Kaufer, E., 1989, The Economics of the Patent System, Chur, 76 (1), 141–154. UK: Harwood Academic Publishers. Griliches, Z., 1990, ‘Patent Statistics as Economic Indicators: Kerr, C., 1994, Troubled Times in American Higher Education, A Survey’, Journal of Economic Literature 28, 1661–1707. Albany: State University of New York Press. Griliches, Z., 1998, R&D and Productivity: The Economet- Leyden, D.P. and A.N. Link, 1999, ‘Federal Laboratories as ric Evidence, National Bureau of Economic Research for Research Partnerships’, International Journal of Industrial the University of Chicago Press, Chicago: University of Organization 17, 575–592. Chicago Press. Lichtenberg, F.R. and D. Siegel, 1991 ‘The Impact of R&D Hagedoorn, J., A.N. Link, and N.S. Vonortas, 2000, ‘Research Investment of Productivity—New Evidence Using Linked Partnerships’, Research Policy 29, 567–586. R&D-LRD Data’, Economic Inquiry 29 (2), 203–228. Hall, B.H., Z. Griliches, and J.A. Hausman, 1986, ‘Patents and Link, A.N., 1981, ‘Basic Research and Productivity Increase in Manufacturing’, American Economic Review 71 (5), 1111– R&D: Is There a Lag?’, International Economic Review 27, 1112. 265–302. Link, A.N., 1983, ‘Market Structure and Voluntary Product Hall, B.H., A. Jaffe, and M. Trajtenberg, 2000, ‘Market Value Standards’, Applied Economics 15, 393–401. and Patent Citations: A First Look’, National Bureau of Link, A.N., 1990, ‘Firm Size, University-Based Research, and Economic Research Working Paper No. 7741. the Returns to R&D’, Small Business Economics 2, 25–32. Hall, B.H., A.N. Link, and J.T. Scott, 2000, ‘Universities as Link, A.N., 1995, A Generosity of Spirit: The Early History Research Partners’, NBER Working Paper No. 7643. of Research Triangle Park, Research Triangle Park, NC: Hall, B.H., A.N. Link, and J.T. Scott, 2001, ‘Barriers Inhibiting Research Triangle Foundation of North Carolina. Industry from Partnering with Universities: Evidence from Link, A.N., 1996a. Evaluating Public Sector Research and Devel- the Advanced Technology Program’, Journal of Technology opment, New York: Praeger Publishers. Transfer 26, 87–98. Link, A.N., 1996b, ‘On the Classification of Industrial R&D’, Hall, B.H. and J. van Reenen, 2000, ‘How Effective Are Fiscal Research Policy 25, 397–401. Incentives for R&D? A Review of the Evidence’, Research Link, A.N., 1999, ‘Public/Private Partnerships in the United Policy 449–469. States’, Industry and Innovation 6, 191–217. Hébert, R.F. and A.N. Link, 1988, The Entrepreneur: Main- Link, A.N., 2000, ‘An Assessment of the SBIR Fast Track stream Views & Radical Critiques, 2nd ed., New York: Program in Southeastern United States’, in C. Wessner, Praeger Publishers. (ed.), The Small Business Innovation Research Program: Henderson, R., A. Jaffe, and M. Trajtenberg, 1998, ‘Universi- An Assessment of Fast Track, Washington, DC: National ties as a Source of Commercial Technology: A Detailed Academy of Sciences. Analysis of University Patenting, 1965–1988’, Review of Link, A.N. and J. Rees, 1990, ‘Firm Size, University Based Economics and Statistics, 80 (1), 119–127. Research and the Returns to R&D’, Small Business Eco- Hirschey, M., 1982, ‘Intangible Capital Assets of Advertising nomics 2, 25–32. and R&D Expenditures’, Journal of Industrial Economics Link, A.N. and J.T. Scott, 1998, Public Accountability: Evalu- 30, 370–390. ating Technology-Based Institutions, Norwell, MA: Kluwer Hounshell, D.A., 1996, ‘The Evolution of Industrial Research Academic Publishers. in the United States’, in R. Rosenbloom and W. Spencer Link, A.N. and J.T. Scott, 2000, ‘Evaluating Federal Research (eds.), Engines of Innovation: U.S. Industrial Research at Programs’, mimeographed. the End of an Era, Boston: Harvard Business School Press, Link, A.N. and J.T. Scott, 2001, ‘Public/Private Partnerships: pp. 13–86. Stimulating Competition in a Dynamic Economy’, Interna- House Committee on Science, 1998, Unlocking Our Future: tional Journal of Industrial Organization 19 (5), 763–794. Toward a New National Science Policy, Report to Congress, Link, A.N. and G. Tassey, 1987, Strategies for Technology-based September 24. Competition: Meeting the New Global Challenge, Lexington, Jaffe, A.B., 1986, ‘Technological Opportunity and Spillovers of MA: Lexington Books. R&D: Evidence from Firm’s Patents, Profits, and Market Mansfield, E., 1968, Industrial Research and Technological Value’, American Economic Review 76, 984–1001. Change, New York: W.W. Norton. 202 Audretsch et al.

Mansfield, E., 1971, Technological Change, New York: W.W. Pakes, A. and Z. Griliches, 1984, ‘Patents and R&D at the Norton. Firm Level: A First Look’, in Z. Griliches (ed.), R&D, Mansfield, E., 1980, ‘Basic Research and Productivity Increase Patents, and Productivity, Chicago: University of Chicago in Manufacturing’, American Economic Review 70 (5), 863– Press, pp. 55–72. 873. Pakes, A. and M. Schankerman, 1984, ‘The Rate of Obsoles- Mansfield, E., J. Rapoport, A. Romeo, S. Wagner, and cence of Patents, Research Gestation Lags, and the Private G. Beardsley, 1977, ‘Social and Private Rates of Return Rate of Return to Research Outcomes’, in Zvi Griliches, from Industrial Innovations’, Quarterly Journal of Eco- (ed.), R&D, Patents, and Productivity, National Bureau of nomics 91 (2), 221–240. Economic Research for the University of Chicago Press, Morgan, R.P., 1998, ‘University Research Contributions to Chicago: University of Chicago Press, pp. 73–88. Industry: The Faculty View’, in P. Blain and R. Frosch Paul, C.J. Morrison and D.S. Siegel, 2001, ‘The Impacts of (eds.), Trends in Industrial Innovation: Industry Perspectives Technology, Trade, and Outsourcing on Employment and & Policy Implications, Research Triangle Park: Sigma Xi, Labor Composition’, Scandinavian Journal of Economics The Scientific Research Society, pp. 163–170. 103 (2), 241–264. Morrison, C. and D. Siegel, 1997, ‘External Capital Factors Prevezer, M., 1997, ‘The Dynamics of Industrial Clustering in and Increasing Returns in U.S. Manufacturing’, Review of Biotechnology,’ Small Business Economics 9, 255–271. Economics and Statistics 79 (4), 647–654. Reis, P. and D.H. Thurgood, 1991, Summary Report 1991: Doc- Mowery, D.C. and D.J. Teece, 1996, ‘Strategic Alliances and toral Recipients from United States Universities, Washington Industrial Research’, in R. Rosenbloom and W. Spencer DC: National Academy Press. (eds.), Engines of Innovation: U.S. Industrial Research at Roessner, J.D. and A. Bean, 1991, ‘How Industry Interacts the End of an Era, Boston: Harvard Business School Press, with Federal Laboratories’, Research and Technology Man- pp. 111–129. agement 21, 22–26. Mueller, D.C., 1967, ‘The Firm Decision Process: An Econo- Rogers, J. and B. Bozeman, 1997, ‘Basic Research and Tech- metric Investigation’, Journal of Political Economy 81, nology Transfer in Federal Laboratories’, Journal of Tech- 58–87. nology Transfer 22, 37–48. National Resources Committee, 1938, Research—A National Romeo, A., 1975, ‘Interindustry and Interfirm Differences in Resource, Washington, DC: Science Committee of the the Rate of Diffusion of an Innovation’, Review of Eco- National Resources Committee. nomics and Statistics 57, 311–319. National Science and Technology Council, 1996, Assessing Fun- Romer, P.M., 1986, ‘Increasing Returns and Long-Run damental Science, Washington, DC: National Science and Growth,’ Journal of Political Economy 94 (5), 1002–1037. Technology Council. Romer, P.M., 1994, ‘The Origins of Endogenous Growth,’ Jour- National Science Board, 2000, Science and Engineering nal of Economic Perspectives 8, 3–22. Indicators—2000, Washington, DC: National Science Foun- Romer, P.M., 2000, ‘Should the Government Subsidize Supply dation. or Demand in the Market of Scientists and Engineers?’, National Science Foundation, 1998, Graduate Studies and Post- NBER Working Paper No. 7723. doctorates in Science and Engineering: 1996, NSF 98–301, Washington, DC: National Science Foundation. Saxenian, A., 1990, ‘Regional Networks and the Resurgence of Nelson, R.R. and E.S. Phelps, 1966. ‘Investment in Humans, Silicon Valley’, California Management Review 33 89–111. Technological Diffusion, and Economic Growth’, American Schankerman, M. and A. Pakes, 1985, ‘The Rate of Economic Review 56, 69–75. Obsolescence and the Distribution of Patent Values: Nightingale, P., 1998, ‘A Cognitive Model of Innovation’, Some Evidence from European Patent Renewals’, Revue Research Policy 27, 689–709. Economique 36 (5), 917–941. Office of the President, 1990, U.S. Technology Policy, Schankerman, M. and A. Pakes, 1986, ‘Estimates of the Value Washington, DC: Executive Office of the President. of Patent Rights in European Countries During the Post- Office of Technology Policy, 1996, Effective Partnering: A 1950 Period’, Economic Journal 96, 1052–1076. Report to Congress on Federal Technology Partnerships, Scherer, F.M., 1965, ‘Firm Size, Market Structure, Opportunity Washington, DC: U.S. Department of Commerce. and the Output of Patented Inventions’, American Eco- Olson, K., 1998, ‘Total Science and Engineering Gradu- nomic Review 55, 1097–1125. ate Enrollment Falls for Fourth Consecutive Year’, Data Scherer, F.M., 1983, ‘The Propensity to Patent’, International Brief, Arlington, VA: National Science Foundation, Decem- Journal of Industrial Organization 1, 107–128. ber 17. Schmookler, J., 1966, Invention and Economic Growth, Cam- Pakes, A., 1985, ‘On Patents, R&D, and the Stock Market bridge, MA: Harvard University Press. Rate of Return’, Journal of Political Economy 93 (2), 390– Schwalbach, J. and K.F. Zimmermann, 1991, ‘A Poisson Model 409. of Patenting and Firm Structure in Germany’, in Z. Acs Pakes, A., 1986, ‘Patents as Options: Some Estimates of the and D.B. Audretsch (eds.), in Innovation and Technological Value of Holding European Patent Stocks’, Econometrica Change: An International Comparison, Ann Arbor: Univer- 54 (4), 755–784. sity of Michigan Press, pp. 109–120. Pakes, A. and Z. Griliches, 1980, ‘Patents and R&D at the Siegel, D., 1994, ‘Errors in Output Deflators Revisited: Unit Firm Level: A First Report’, Economics Letters 5, 377–381. Values and the PPI’, Economic Inquiry, 32 (1), 11–32. The Economics of Science and Technology 203

Siegel, D., 1997, ‘The Impact of Computers on Manufactur- Stephan, P. and S.G. Levin, 2002, ‘Exceptional Contribu- ing Productivity Growth: A Multiple-Indicators, Multiple- tions to U.S. Science by the Foreign-Born and Foreign- Causes Approach’, Review of Economics and Statistics 79 Educated’, Population Policy and Research Review,in (1), 68–78. press. Siegel, D.S., 1999, Skill-Biased Technological Change: Evidence Tassey, G., 1999, ‘R&D Trends in the U.S. Economy: Strate- from A Firm-Level Survey, Kalamazoo, MI: W.E. Upjohn gies and Policy Implications’, Gaithersburg, MD: National Institute Press. Institute of Standards and Technology. Siegel, D. and Z. Griliches, 1992, ‘Purchased Services, Out- Tassey, G., 1997, The Economics of R&D Policy, New York: sourcing, Computers, and Productivity in Manufacturing’, Quorum Books. in Output Measurement in the Service Sector, Chicago: Uni- Tassey, G., 2000, ‘Standardization in Technology-Based Mar- versity of Chicago Press, pp. 429–458. kets’, Research Policy 29, 587–602. Siegel, D., J.G. Thursby, M.C. Thursby, and A.A. Ziedonis, Terleckyj, N.E., 1974, Effects of R&D on the Productivity 2001, ‘Organizational Issues in University-Industry Tech- Growth of Industries: An Explanatory Study, Washington, nology Transfer: An Overview of the Symposium Issue’, DC: National Planning Association. Journal of Technology Transfer 26 (1–2), 5–11. Tibbetts, R., 1999, ‘The Small Business Innovation Research Siegel, D., D. Waldman, and A.N. Link, 1999, ‘Assessing the Program and NSF SBIR Commercialization Results’, Impact of Organizational Practices on the Productivity of mimeograph. University Technology Transfer Offices: An Exploratory Tirole, J., 1988, The Theory of Industrial Organization, Cam- Study’, NBER Working Paper No. 7256 and Research Pol- bridge, MA: The MIT Press. icy, in press. Trajtenberg, M., 1990a, ‘A Penny for Your Quotes: Patent Cita- Siegel, D., D. Waldman, L. Atwater, and A.N. Link, 2000, tions and the Value of Innovations’, Rand Journal of Eco- ‘Transferring Knowledge from Academicians to Prac- nomics 21 (1), 172–187. titioners: Qualitative Evidence from University Tech- nology Transfer Offices’, mimeograph, University of Trajtenberg, M., 1990b, Economic Analysis of Product Innova- Nottingham. tion: The Case of CT Scanners, Cambridge, MA: Harvard Siegel, D., P. Westhead, and M. Wright, 2002, ‘Assessing the University Press. Impact of Science Parks on the Research Productivity of UNESCO, 1968, National Science Policies of the U.S.A.: Ori- Firms: Evidence from the U.K.’, Small Business Economics, gins, Development and Present Status, Paris: Unesco. in press. U.S. Congress, Office of Technology Assessment, 1996, The Sobel, D., 1995, Longitude, New York: Penguin Books. Effectiveness of Research and Experimentation Tax Credits, Solow, R.M., 1957, ‘Technical Change and the Aggregate Pro- Washington, DC: Office of Technology Assessment. duction Function’, Review of Economics and Statistics, 39, U.S. Department of Commerce, 1990, Emerging Technolo- 312–320. gies: A Survey of Technical and Economic Opportunities, Steelman, J.R., 1947, Science and Public Policy: A Program for Washington, DC: Technology Administration. the Nation. Washington, DC: The U.S. Presidents Scientific U.S. General Accounting Office (GAO), 1999, Managing Research Board, August 27. for Results: Opportunities for Continued Improvements in Stephan, P., 1996, ‘The Economics of Science’, Journal of Eco- Agencies’ Performance Plans, Washington, DC: General nomic Literature 34, 1199–1235. Accounting Office.