NIST GCR 02–841 Between Invention and Innovation

NIST GCR 02–841
Between Invention and Innovation
An Analysis of Funding for Early-Stage Technology Development

INTRODUCTION: MOTIVATION and APPROACH

The ultimate limits to growth may lie not as much in our ability to generate new ideas, so much as in our ability to process an abundance of potentially new seed ideas into usable forms.
—Martin L. Weitzman (1998: p. 333)

1. MOTIVATION

In the field of economics, fewer relationships are more broadly supported by both theory and empirical evidence than the relationship between technological innovation and long-term growth.⁽²⁾ Yet prior to the mid-1980s, economists undertook little detailed study of the process by which ideas are transformed into new goods and services, or how new industries and sectors of economic activity arise.⁽³⁾ As Nobel Laureate Kenneth Arrow observed in 1988: “Innovations, almost by definition, are one of the least analyzed parts of economics, in spite of the verifiable fact that they have contributed more to per capita economic growth than any other factor” (Arrow 1988: p. 281). Similarly, public policies aimed at enhancing science and technology-based economic growth were based on the assumption that leadership in basic science and military R&D would automatically and indefinitely translate into broad economic benefit.⁽⁴⁾ Strong political support for funding of basic research allowed U.S. research universities to achieve international pre-eminence and supported the establishment and growth of the National Institutes of Health (NIH) and the National Science Foundation (NSF).

During the 1980s, however, this linear model of innovation with its associated laissez-faire policy implications came into question. Japanese firms successfully challenged formerly dominant U.S. companies in a range of high-tech industries. Industry in the United States was widely perceived to have failed to give high enough priority to manufacturing efficiency and consumer satisfaction.⁽⁵⁾ Leading U.S. technology companies were embarrassed by widely publicized failures of commercial development of market-transforming technologies invented in their own research laboratories.⁽⁶⁾ At the same time, the “new growth” or “endogenous growth” theories associated with Romer (1986, 1990), Grossman and Helpman (1991), and Aghion and Howitt (1993) refocused economists’ attention on the manner in which micro-economic incentives affect the transformation of ideas into long-term growth. Work by Young (1991) and Lucas (1993) in particular emphasized that, although incremental technical change accounts for most observed increases in productivity, sustained long-term growth requires the continuous introduction of new goods and services. For this reason, national investment into the conversion of inventions into radically new goods and services, although small in absolute terms when compared to total industrial R&D, significantly affects long-term economic growth by converting the nation’s portfolio of science and engineering knowledge into innovations generating new markets and industries.

In the 1990s, the U.S. economy experienced a remarkable resurgence driven by dramatic gains in industrial productivity. Scholars have produced a solid body of knowledge about innovation systems;⁽⁷⁾ economic behavior in the face of technological risk, uncertainty, and incomplete information;⁽⁸⁾ and social capital, regional agglomeration, and industry clustering.⁽⁹⁾ We now know that the early development of a novel technology depends on academic science, which generates the ideas that drive the innovation process;⁽¹⁰⁾ on the magnitude and geographical localization of knowledge spillovers;⁽¹¹⁾ and on the social returns from investments in R&D, including those made by the federal government.⁽¹²⁾ The roles played by large corporations, new firms (in particular, those backed by private-equity financing), research and development alliances, and partnerships of various types, and federal and state governments have been described. ⁽¹³⁾

As investors stampeded first into, then out of, the public market for equity in technology-based “new economy” firms, the assertion that U.S. economic growth is led by entrepreneurial, venture capital-backed technology firms became almost an article of faith among politicians, pundits, policy makers, and the general public. The volume of traffic along the path from Palo Alto’s Sand Hill Road⁽¹⁴⁾ to Wall Street came increasingly to represent not merely an indicator of the vibrancy of a single economic sector, but a scorecard for the economy as a whole. The number and diversity of institutions specializing in supporting the commercial development and marketing of new technologies have expanded dramatically, in a manner unlikely to be reversed. These include venture capital firms, corporate venture funds, incubators of various types, niche law firms, university and government offices of technology transfer, and networks of individual private-equity angel investors.⁽¹⁵⁾

Funds available to high-growth technology ventures appear at first glance to have grown accordingly. In particular, the overall growth in venture capital suggests to many observers of the U.S. innovation system that private funding is available for high-technology projects. Indeed, at present, by some measures, the supply of such funds seems to exceed the demand. Venture capital funds disbursed to firms reached a peak of over $100 billion in the year 2000, before dropping off to $37 billion in 2001. As of February 2002, the magnitude of commitments from the limited partners that invest in venture capital funds (such as pension funds, banks, endowments, and wealthy individuals) exceeded industry-wide disbursements by a total of $75 billion—more than the cumulative total of venture capital investments from 1990 to 1998.

Yet, even in such an environment, practitioners report that the process of translating a basic science invention into a commercially viable innovation is extremely difficult and getting more so.⁽¹⁶⁾ The economic and technological factors driving this trend are not new. Four years ago, then Undersecretary of Commerce Mary Good testified before the Senate Committee on Governmental Affairs: “As the competitive pressures of the global marketplace have forced American firms to shift more of their R&D into shorter term product and process improvements, an ‘innovation gap’ has developed.... Sit down with a group of venture capitalists. The funding for higher risk ventures ... is extraordinarily difficult to come by.”⁽¹⁷⁾ Entrepreneurs in many settings consistently report difficulty in raising funds in the range of $200,000 to $2 million.⁽¹⁸⁾ The current environment was summed up by Bill Joy of Sun Microsystems, who observed in July 2001, “A couple of years ago, even the bad ideas were getting capital. Now we have gone too far in the opposite direction, shutting down investment in good ideas.”⁽¹⁹⁾

Markets, technologies, and their interrelation are becoming increasingly complex, further complicating the challenge of converting inventions into innovations. The rapid advance of the scientific frontier and the increasing breadth and depth of knowledge available across all scientific fields have contributed to the acceleration of technological complexity. Today, even the largest corporations and the most deep-pocketed venture capital firms have difficulty putting together all the elements required for in-house development and commercialization of truly novel science-based technologies.

2. PROJECT OBJECTIVES

Investments in basic and applied research support the development of both science-based inventions and entrepreneurial talent, the dual prerequisites for commercial innovation. The purpose of this study is to provide comprehensive analysis of investments into early-stage, high-technology ventures to support informed design of public policies regarding invention, technology entrepreneurship, and innovation.

The focus of the project is on early-stage technology development—the difficult transition from (science-based) invention to (commercial) innovation. We use the term “early-stage technology development” (with the abbreviation ESTD) to describe the technical and business activities required to develop a nascent technology into a clearly defined product or service whose specifications and business plan are matched to a particular market. ESTD and invention-to-innovation transition are equivalent in our usage. The premise of the project is that some degree of quantification of the magnitude and distribution of investments in early-stage technology development is a prerequisite to determining the appropriate role of government in supporting science-based innovation and technology entrepreneurship.

As noted above, many technology entrepreneurs, investors, and policy makers have noted what is called a funding gap that faces science-based, new enterprises seeking seed-stage funding. On an anecdotal level, assertions of the existence of a funding gap have been remarkably persistent.⁽²⁰⁾ How can we directly test the validity of these assertions? A series of rigorous studies at the level of the U.S. economy have consistently estimated that social returns to research and development investments substantially exceed private returns.⁽²¹⁾ Unfortunately, because such studies characteristically lump into undifferentiated R&D all basic research expenditures together with all phases of investment in technology development, they neither support nor refute the notion of a funding gap between a basic science breakthrough and the development technological prototype linked to a market. The existence of a funding gap, in the textbook economics sense of a shortfall from a social optimum, would be extremely difficult to establish empirically; to do so would require at minimum not only reliable data on both the demand and supply for ESTD funding in particular (as opposed to all R&D expenditures), but also computation of project-level marginal social benefits of such funding.

In a similar vein, boilerplate statements regarding the existence of market failure in the context of ESTD have little content without elaboration regarding specifics. As argued first by Arrow (1962), market failures—rigorously defined—abound in the market for new ideas and technological information. Generically, perfect competition may fail to achieve optimal resource allocation whenever products are indivisible (marginal cost pricing rules apply imperfectly), economic actors are unable to appropriate the full returns from their activities (social and private benefits diverge), and/or outcomes are uncertain (future states of nature are unknown). Clearly, all three of these attributes characterize basic research as well as ESTD projects. Of the three, it is instructive to note that the discussion in Arrow (1962) begins not with inadequate incentives to innovate due to imperfect appropriability, but rather with contracting problems due to uncertainty. In particular, Arrow points out that the activity of invention has particular characteristics that complicate the ability of economic actors to relieve themselves of risks due to uncertainty. Arrow notes that success in “highly risky business activities, including invention” depends on “an inextricable tangle of objective uncertainties and decisions of the entrepreneurs and is certainly uninsurable. On the other hand, such activities should be undertaken if the expected return exceeds the market rate of return, no matter what the variance is. The existence of common stocks would seem to solve the allocation problem... But then again the actual managers no longer receive the full reward of their decisions; the shifting of risks is again accompanied by a weakening of incentives to efficiency.”⁽²²⁾

Elaborating upon Arrow (1962) and Zeckhauser (1996), we observe that every high-technology innovation, by its nature, calls for specialized technical knowledge; and every radical innovation that expects to create a market that does not yet exist, can only be evaluated by someone with experience in new market creation in that segment of the business world. Talent at this level to assess both technologies and markets is scarce. Furthermore, the value of a technical idea close to commercial application depreciates rapidly. Consequently, as Zeckhauser (1996) argues, technological information (TI) is not, as is widely assumed in the economics literature, a public good. Indeed, “excessive focus” on the public good character of technological information has led economists “to slight the major class of market failures associated with TI that stems from its amorphous quality.”⁽²³⁾

From the standpoint of data, our objective thus is not to test directly the hypothesis of a funding gap. Rather, more modestly, our objective is to estimate the sources of funding for early-stage technology development. In this sense we seek not to offer conclusive results regarding the appropriate distribution of input, but instead to suggest some underlying parameters and definitions to set the context for debate over public policy and for future academic research.

We begin by articulating a set of definitions of the ESTD process that are focused on the technology project (driven by a specific champion and team). We then employ these definitions to develop useful qualitative and quantitative comparisons of ESTD projects and investments across institutional settings, including new firms, corporations, universities, and government labs.

Investments by government, corporations, institutions, and individuals in basic and applied research support the development of both science-based inventions and entrepreneurial talent, the prerequisites for commercial innovation. Corporate and venture capital investors are effective in exploiting scientific and technological advances when such advances are embodied in new products and services whose specifications and costs match well-defined market opportunities. However, this conversion of inventions into commercial innovations is a process fraught with obstacles and risks. Despite the apparent abundance of funds available for the marketing of readily commercializable technologies, many technologists, investors, and public- and private-sector decision makers argue that significant institutional and behavioral barriers continue to impede technology development after invention.

A huge amount of academic research effort has been dedicated to understanding the U.S. research and invention enterprise, which includes non-profit universities, government laboratories, and those companies that engage in the more basic end of the research spectrum.⁽²⁴⁾ At the other end of the innovation spectrum, a considerable amount of effort (most of it centered in business schools) has been dedicated to understanding how businesses are managed once they have a core set of products and are incrementally innovating. Considerably less effort has been devoted to understanding what goes on between the point at which research has defined an economic opportunity and the later stage when a champion can make the business case that the opportunity will be a predictable source of revenue. This is our focus. More specifically, our inquiry is organized around two sets of questions:

We emphasize the inputs into the innovation process, rather than outputs or outcomes. From a public policy perspective, inputs are only interesting to the extent that they relate to socially desired outcomes. However, before one can begin to discuss the relationship of inputs to outcomes, one must first arrive at a coherent picture of the process, the institutional participants, and the basic definitions that allow for comparison of roles and contributions.

3. APPROACH

We pursued two approaches in parallel: first, learning from the observations of practitioners in the context of a series of workshops held in the U.S., and second, collecting the largely fragmentary data available on ESTD investments from other studies and from public statistical sources. These studies were supplemented by four case studies and a set of thirty-nine interviews of corporate technology managers, CEOs and venture capitalists conducted by Booz Allen & Hamilton.

Two practitioner workshops were held: at the Carnegie Endowment for International Peace in Washington, D.C., on January 25, 2001; and at the Xerox Palo Alto Research Center (PARC) in Palo Alto, California, on February 2, 2001. An analytic workshop was held at the Kennedy School of Government (Cambridge, Massachusetts) on May 2, 2001. The workshops brought together representatives from the following groups (see Annex III for a workshop agendas and biographies of participants):

The workshops were particularly helpful in two ways: refining our operational definition of the invention-to-innovation transition, and providing guidance in the interpretations of and, where necessary, extrapolations from the available data. Participants in the practitioner workshops included past ATP awardees; participants in the International Business Forum (IBF) Early-Stage Investing Conference;⁽²⁵⁾ firms and individuals nominated by leading angel investors and venture capitalists engaged in seed-stage funding of technology-based firms; firms and individuals affiliated with the MIT Entrepreneurship Center; and the investigators’ personal contacts. The three workshops are the source of all direct quotations in the document, unless otherwise noted.

The two practitioner workshops included methodological, data, and case-based panel discussions. Participants in the methodological and data panels were asked to describe the organizational and institutional context underlying their publicly reported figures on R&D investments. Of particular interest were panelists’ estimates of the distribution of firm expenditures at each stage of the innovation process. The case-based panel discussions focused on two technology areas: amorphous silicon technology and bioinformatics. Each of the case-based panels traced the history of the development of the technology, highlighting the role of different funding sources at each stage and the particular challenges encountered. The separately published case studies on Caliper Technologies and GE Medical Devices further explore the process of early-stage technology development in the context of these two technology areas.

Our analysis is based upon examination of both published and unpublished data sources, and on insights from extensive conversations with survey managers, industry analysts, and practitioners. Our purpose is to achieve a new perspective on the level of funding that is applied to ESTD. We focus on the six most important sources of funding for ESTD identified in Part I: corporate, venture capital, angels, federal government, state governments, and universities. Beginning with an aggregate figure for support of scientific and technological innovation from each funding source, we develop a rationale for more realistic estimates of the fraction of funding flows to research and development that are directed into ESTD.

Given the lack of rigor in the definitions used in much of the available data and its fragmentary character, we present our findings in the form of two models. One model takes a very restricted view of what constitutes ESTD, so that the inferences based on this model are almost certainly lower than the most realistic value. The second model takes a more expansive view, using source data that almost certainly overestimate investments in ESTD. Both models are defined by a set of assumptions that are in some cases subjective, but are based on the insights of informed practitioners. It must be recognized, however, that a more accurate model might represent different choices, some from the model based on less restrictive definitions and others from the model based on more restrictive definitions. Our intent here is to create plausible upper and lower estimates for ESTD funding.

The results from the use of these two models are summarized in Table 1 in Part II (page 62 of this report). Table 1 is then used to calculate the relative magnitude of ESTD expenditures from the different sources considered. This method does not allow the precise determination of a best or most probable estimate. Furthermore, the range between the upper and lower range estimates is very large—a factor of six in the total ESTD flows. However, by combining those percentages from the two models, we see that the relative importance of different sources of investment is similar. This finding is more relevant for supporting some of the public policy conclusions that we seek. As Figure 1 indicates, the distribution of ESTD funding is relatively independent of the model used.

The observation year for this study is 1998, except as noted in a couple of cases. Due to reporting lags inherent in most large-scale surveys, we have relied upon 1998 as the most recent year for which comprehensive and reliable data are available. While the selection of this observation year was motivated chiefly by the absence of more recent data, it is also a sensible choice for other reasons. Given the size of market fluctuations affecting the technology sector from 1999 through 2001, 1998 is a more reliable benchmark of innovation funding activities than 2000, when markets were at their historic peaks.⁽²⁶⁾

Limitations inherent in the data and the magnitude of the extrapolations and subtractions we carry out demand that our findings be interpreted with caution.

Our results are in the form of two sets of estimates, based on the upper and lower models and on assumptions that are broadly consistent with the full range of data sources available to us. The funding range we present for each category is large, but as a first approximation, these initial estimates provide valuable insight into the overall scale and composition of ESTD funding patterns and allow at least a preliminary comparison of the relative level of federal, state, and private investments.

To build our lower estimates, we applied a narrowly defined lens to develop a conservative estimate of innovation activities in different institutional settings. Our aim was to develop a baseline minimum amount of funding that sets a reasonable and defensible floor for estimated total ESTD funding in the United States.

To derive our upper estimates, we attributed basic and applied research funding more generously to ESTD. These allocation estimates varied by institutional setting and were significantly informed by conversations with practitioners and analysts in the field. We deliberately aimed to choose allocations that were as large as reasonable in order to determine an upper limit on the nation’s potential ESTD funding.

We have made extensive use of large-scale R&D surveys conducted by the National Science Foundation (NSF). These survey results rely heavily on respondents’ judgement for such crucial items as the classification of R&D projects across industries, geography, and institutional settings. NSF surveys provide the best data of its kind and scope, but because they were not crafted specifically to help track activities in the invention-to-innovation divide, our interpolations of ESTD funding flows from this data depend on our analysis of these survey results and our best guesses—informed by the perspectives of both practitioners and the data-gatherers—as to how categorizations of ESTD activities into basic research, applied research, and development categories vary across institutional settings. These are described in Part II.

Time series data would be helpful in tracking trends in funding flows and identifying relationships with business cycles, but the scope of the present study provides only a point estimate for the given observation year. Some insights into trends over time that developed during the course of this project are presented later in this report. Regional and sectoral concentrations of resources are largely ignored in Table 1, though the importance of these patterns is well recognized.

The core report contains an executive summary, this chapter on motivation and approach, and two additional chapters—the first (Part I) summarizing qualitative findings, drawing upon the insights offered at the practitioner workshops, and the second (Part II) presenting the methods behind our analysis of funding for ESTD from different institutional categories. The core report also includes several

The case studies and the full report from Booz Allen & Hamilton are available as publications of the Advanced Technology Program which can be found on the program’s website: <http://www.atp.nist.gov>.

A team of researchers at the Belfer Center for International Affairs, Kennedy School of Government, Harvard University carried out the project, which was funded by the Advanced Technology Program of the U.S. Department of Commerce. Professor Lewis Branscomb (Aetna Professor of Public Policy and Corporate Management, emeritus, Kennedy School of Government, Harvard University) and Dr. Philip Auerswald (Deputy Director of the Science, Technology and Public Policy Program and Adjunct Lecturer, Kennedy School of Government, Harvard University) led the project team. Brian Min contributed substantially to the research and writing of Part II of the report. Independent research on corporate support for ESTD was carried out in support of this project by a team at Booz Allen & Hamilton (BAH) led by Nicholas Demos (Vice President, Strategy Practice), Gerald Adolph (Senior Vice President), Rhonda Germany (Vice President, Consumer and Health Practice), and Raman Muralidharan (Vice President, Consumer and Health Practice.). A memorandum summarizing the BAH finding is attached as Annex I. Dr. Mona Ashiya, Robert Kolasky, Thomas Livesey, and Jonathan Westrup authored supporting case studies of technology projects and institutional innovations; these are published separately from this report. Livesey additionally contributed research support.

____________________ [Click on image to go back to text.]

2. Lively debates do exist over the effects of specific innovations on human and environmental welfare, but the central role of technological innovation as a driver of conventionally measured (GDP) growth is undisputed. Jones and Williams (1998) provide a survey of both models and evidence.

3. Important exceptions include Griliches (1963, 1979), Arrow (1962), Shell (1966, 1967), and Nelson and Phelps (1966).

7. See Nelson (1993), Branscomb and Keller (1998), and Branscomb, Kodama, and Florida (1999).

8. See Aghion and Tirole (1994), Dixit and Pindyck (1994), Zeckhauser (1996), and Branscomb and Auerswald (2001).

9. See Krugman (1991) Glaeser et al. (1992), Branscomb (1996), Gaspar and Glaeser (1997), Fountain (1998), Glaeser et al. (2000).

10. See Rosenberg and Nelson (1994), Henderson, Jaffe and Trajtenberg (1998), Jensen and Thursby (1998), and Branscomb, Kodama, and Florida (1999).

12. See Mansfield et al. (1977), Griliches (1992), Jones and Williams (1998), Borrus and Stowsky (1998).

13. See Acs and Audretsch (1988), Alic et al. (1992), Scherer (1999), Bidhé (2000), and Kortum and Lerner (2000).

14. Located in Palo Alto, Sand Hill Road is the Wall Street of the venture capital industry in Silicon Valley.

15. The term “angel investor “comes from the theater—see also Part I, section 3C, of this work. [please note: all this material also appears verbatim later in the text.]

17. Cited in Gompers and Lerner (2000, p. 2). These authors quite accurately point out an apparent contradiction in the quote from Dr. Good, which appears in its edited form to suggest that venture capitalists are reluctant to provide risk capital. Of course, this is not the case. As Gompers and Lerner describe, the venture capital mode of finance is precisely that which is specialized in providing finance in contexts where uncertainty is high and information asymmetries severe. At the same time, however, as Morgenthaler (2000) and other venture capitalists report, the risk/reward ratio for seed-stage technology-based ventures is not as attractive to venture capital firms as it is for ventures at a slightly later stage. We develop this argument further below.

18. The hypothesis of such a capital gap in seed-stage funding for new ventures is discussed by Sohl (1999), and consistently corroborated by practitioners (see, for instance, comments by participants at a Senate Small Business Committee Forum, <www.senate.gov/~sbc/hearings/internet.html>).

20. See, for example, accounts from participants at a 2001 Senate Small Business Committee Forum, <www.senate.gov/~sbc/hearings/internet.html>. The phenomenon is not restricted to the United States. The U.K. Department of Trade and Industry published a report in 1999 titled “Addressing the SME [small and medium-size enterprises] Equity Gap.”

21. See Mansfield et al. (1977), Griliches (1992), and Jones and Williams (1998)

22. Hellman (1998) describes the manner in which control rights in venture capital contracts mitigate the sorts of risks described by Arrow (1962).

23. To emphasize this point, Zeckhauser offers the following illustration: “A thought experiment might task what would happen if information remained a public good, but were susceptible to contract. Fortunately, there are public goods that offer relatively easy contracting, such as songs or novels, which offer an interesting contrast with information. Such goods appear to be well-supplied to the market, with easy entry by skilled low-cost songwriters and novelists.” Zeckhauser identifies five distinguishing characteristics of technical information that complicate contracting:
• Technical information is difficult to count and value.
• To value technical information, it may be necessary to “give away the secret.”
• To prove its value, technical information is often bundled into complete products (for instance, a computer chip or pharmaceutical product).
• Sellers’ superior knowledge about technical information makes buyers wary of overpaying.
• Inefficient contracts are often designed to secure rents from technical information (1996, 12746).

24. . Leading contributors to this literature include Harvey Brooks, Wesley Cohen, Michael Darby, Paul David, Maryann Feldman, Christopher Freeman, David Mowery, Richard Nelson, Keith Pavitt, Nathan Rosenberg, Donald Stokes, and Lynne Zucker.

26. Illustratively, venture capital funds disbursed to firms reached a peak of over $100 billion in the year 2000, before dropping off to $37 billion in 2001. In 1998 (our reference year) total venture capital disbursements were $17 billion.