International policy emulation and university-industry technology transfer David C. Mowery Haas School of Business U.C. Berkeley
Overview Systems of innovation literature rarely considers interaction among different national systems. evolutionary : cross-border flows of capital, technology => some requirement for adaptation, reflecting competitive pressure. purposive : policymakers learn from success of one another s policies, etc. Purposive interaction => conscious emulation. OECD members evolving policies toward university-industry interaction are an example of purposive interaction, emulation.
International emulation in technology policy Examples: R&D collaboration (EU, US emulate Japanese collaboration in the 1980s; Japan now emulates SEMATECH in ASET, SELETE). Deregulation. Intellectual property rights. Venture capital. Characteristics: Learning is selective. Implementation of the imitative/emulative response further muddles the fidelity of the reflection. International emulation of Bayh-Dole shares these characteristics.
The Bayh-Dole Act of 1980 Act sought to encourage commercial development of federally funded inventions. Simplified procedures through which universities and gov t labs could patent federally funded inventions and license them. Bayh-Dole did not legalize anything previously prohibited. It replaced a complex web of Institutional Patent Agreements (IPAs) between individual federal agencies and universities. The Act provided a Congressional endorsement of university licensing of federally funded inventions and weakened federal agencies oversight of licensing agreements. Bayh-Dole has been cited (Economist, OECD) as a catalyst for university-industry technology transfer and US economic growth.
International emulation of Bayh-Dole Discussions or policy changes affecting technology transfer activities of national universities in Japan; Italy; Germany; Denmark; France; Canada, and other nations. Bayh-Dole widely cited as a model. Many policy initiatives focus on patenting of university inventions. Transfer ownership of patent rights from faculty to university (Denmark; Germany). Transfer ownership of patent rights from university to individual faculty (Italy). Some initiatives (Sweden, Japan) include authority or public financial support for creation of technology transfer offices. Others (France) liberalize leave-of-absence policies for gov t and university researchers to start new firms.
But several issues have not been addressed How important has the Bayh-Dole Act been in supporting university-industry collaboration and technology transfer in the United States? Would growth in these activities have occurred without the Bayh-Dole Act? Will emulation of the Bayh-Dole Act accelerate collaboration and technology transfer in other nations university systems? What side effects for academic research?
Roles of universities within national innovation systems Source of trained S&Es. Combination of research & training => an important mechanism for knowledge & technology transfer through flow of graduates to industry. Source of peer-reviewed knowledge placed in the global public domain. Magnet for S&E immigrants from diaspora & elsewhere. In some cases, universities support regional hightechnology agglomerations. Are they more important to catalyzing the formation of these clusters or to sustaining their growth? Stanford vs. Fairchild in Silicon Valley. Multiplicity of roles => variety of channels through which universities affect industrial innovation.
How do university and industrial innovation interact? The interaction is bidirectional: industrial research/innovation affects academic research, as well as the reverse. Schockley Semiconductor and Stanford. Major channels of interaction include training; publishing; faculty consulting; new-firm formation; patents & licenses. Channels of interaction affect one another. Importance of different channels varies among fields of research. Outside of pharmaceuticals, US industrial R&D managers in large firms indicate that patents are relatively unimportant channels of influence on industrial innovation.
Little work on relationship among channels of interaction Lack of knowledge about the nature of the interaction among channels => little understanding of how policy-related emphasis on one channel affects operation of others. How important are different channels to different university missions? Little work on role of licensing in spinoff formation, survival. Gender and patenting: Women at similar seniority, productivity levels in US universities appear to engage far less in patenting & licensing of inventions. We know little about relative importance of these, other channels of interaction in different national university, innovation systems. 9
Structural characteristics of US higher education created incentives for technology transfer before 1980 Large scale of national system. No centralized (e.g.,federal) control of administrative policies. Heterogeneous institutional structure (public; private; secular; religious; large; small) and quality. Dependence by many institutions on local sources of financial & political support. This dependence motivated research with local benefits, search for links with local industry. Inter-institutional competition for resources, prestige, faculty. Aghion et al. (2009), Katz & Goldin (2008): Institutional autonomy => higher performance among US universities.
U.S. university patenting predates passage of Bayh-Dole in 1980 Many U.S. universities avoided direct management role in patenting & licensing during the pre-1970 period. Significant change during the 1970s: Overall university patenting accelerates. Private universities expand patenting. Biomedical technologies grow as a share of university patenting and licensing (see Figure). Universities enter direct management of patenting and licensing. In late 1970s, HEW (parent of NIH) considered limitations on exclusive licensing agreements for NIH-funded inventions. Universities (led by Purdue, Stanford, MIT, Harvard, Columbia) responded by lobbying for flexibility (e.g., exclusive licensing agreements) in federal licensing policy. Bayh-Dole thus is an effect of growth in US university patenting during the 70s, as well as one of several causes of increased academic patenting during the 80s.
University patents by class, 1970-95 0.45 0.4 0.35 0.3 share (% ) 0.25 0.2 Chemicals Drugs/medical electrical/electronic mechanical 0.15 0.1 0.05 0 19 70 1971 19 72 19 73 1974 19 75 1976 19 77 1978 1979 19 80 1981 19 82 year 1983 19 84 1985 1986 1987 1988 19 89 1990 19 91 1992 19 93 1994
Post-Bayh-Dole trends University share of all US patents grew from 0.5% in 1980 to 5% by 2005. US universities account for as much as 11% of biotechnology patents by 2000. Entry into patenting by universities with limited pre-1980 experience. High-intensity academic patenters (more than 10 patents assigned during 1970-80) accounted for 87% of academic patents in 1975, 64% in 1992. Medium-intensity academic patenters (< 10 patents during 1970-80) accounted for 15% of academic patents in 1975, 30% in 1992. Entrant academic patenters (no patents during 1970-80) accounted for 6% of patents by 1992. Entrant universities receive lower-quality patents initially, close quality gap with experienced institutional patenters by early 1990s. Underscores complexity, difficulty of learning to patent in university OTTs. Industry-funded share of total university research in U.S. grows from roughly 3.9% in 1980 to 6.9% in 1995, 7.4% in 1999, drops to 5.1% in 2008. 2008 level is below that of 1957.
US research univ. patents % of all domestic-assignee US patents, 1963-99 0.04 0.035 0.03 0.025 share 0.02 0.015 0.01 0.005 0 1963 1965 1967 1969 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 year
Post-Bayh-Dole trends (2) Did the Act catalyze university patenting? Other factors are as important as Bayh-Dole. Chakrabarty & Diehr decisions, US IPR policy evolution; NIH funding of biomedical research; universities search for revenues after DoD research support declines in early 1970s. Much of the post-1980 licensing growth would have occurred without Bayh-Dole. In spite of rhetorical emphasis on university spinoffs, new firms consistently account for less than 20% of US university licensees. How large is the pot of licensing gold? UC systemwide gross revenues averaged $99M/year for FY 2001 2006; net licensing revenues averaged $28M/year. Industry-funded research exceeded $200M during FY 2006. Gross licensing revenues at UC, Stanford, Columbia, elsewhere are dominated by small # of patents, majority of which cover biomedical inventions.
University management of IP For many US universities, net licensing revenues are modest or negative. Other (non-revenue) motives for university patenting: Faculty pressure. Economic development/technology transfer. Funding of research by industrial collaborators/licensees. Research freedom, especially in the absence of experimental use infringement defense. Different goals => different policies, performance measures. University administrators often are not clear about priorities, unrealistic about potential licensing revenues. Fixed costs, operating expenses of TTOs are high. Does every university need an independent TTO? Considerable scope for multi-institutional collaboration. Need for face time with faculty, licensees => decentralization.
Measures of university-industry linkages in US, other OECD nations since Bayh-Dole US cited by OECD for strong links between university & industrial research, especially since 1980. But indicators of university-industry links (% of university research funded by industry), role of universities in national R&D show US ranking behind other nations. Crespi et al. (2006): majority of faculty patents in UK, France, Italy, Germany, Netherlands, Spain are not assigned to universities. European emulation of Bayh-Dole may be based on faulty premises concerning faculty patenting, role of patenting in tech transfer. Factors influencing links, impacts are more complex, difficult to capture via indicators. 18
Does university patenting affect research content and exploitation? Does greater emphasis on one channel of universityindustry interaction have a chilling effect on others? Little evidence that faculty patenters publish less. Ding & Choi (2008) find differences in predictors of faculty participation in SABs vs. firm foundation. Cumulative pubs predict SAB involvement, contemporary pubs more strongly predict firm-foundation involvement. Is increased academic patenting (for many reasons beyond Bayh-Dole) impeding science? Evidence is inconclusive. Cohen et al. (2006): Academics are not constrained by patents. Murray & Stern (2009): Patented discoveries experience relative decline in citations. Sampat et al. (2003): Slowdown in citations to university patents since 1980; slowdown is greatest for industry citations to university patents (Fabrizio, 2007).
Gov t., industry criticism of universities Emphasis on patenting => frictions with some nonbiomedical firms. Hewlett-Packard cites less restrictive IPR regime in non-u.s. universities as a factor in expanding foreign research collaboration. National Institutes of Health efforts to rationalize Materials Transfer Agreements (MTAs) governing transfers of research tools, materials among academic scientists also have encountered problems. universities take inconsistent positions on fair terms of access to research tools depending on whether they are importing tools or exporting them. (Working Group on Research Tools, 1998). Late-2010 National Academy of Sciences report: Patenting and licensing practices should not be predicated on the goal of raising significant revenue for the institution. The likelihood of success is small, the probability of disappointed expectations high, and the risk of distorting and narrowing dissemination efforts is great.
Some universities move to less patentcentric policies Gradual shift to recognize that: Licensing revenues may not be either large or worth the expense and conflict with industry. Much academic research does not resemble the biomedical field. Patents are less valuable, lucrative sources of licensing revenues. Both Stanford, UCB now emulate MIT in combining management of IP with industrial liaison activities supporting industry-sponsored research. New policies have reduced universities emphasis on ownership of tangible IP from IT industry collaborations. December 2005 Open Collaboration Principles agreement between 7 research universities and 4 IT firms.
Conclusion Bayh-Dole extended, rather than creating a new era in US university patenting, licensing, industry collaboration. Long history of such collaboration reflects structural characteristics of US higher education system. Growth of patenting in the 1980s was rooted in developments during the 1970s. In important respects, Bayh-Dole was a response to increased university patenting, rather than an exogenous causal factor. Other developments in US IPR policy, growth in biomedical research funding and scientific advances, contributed to the growth of patenting during the 1980s. The endogeneity of Bayh-Dole may limit the effects of its emulation by other governments.
Conclusion (2) More than a Bayh-Dole policy is needed to stimulate closer interaction between universities and industry. Structure of public research funding. Structure of university system. Importance of institutions external to the university (labor mobility; venture capital). Multiplicity of channels through which universities and industry interact => the importance of patents for university-industry technology & knowledge transfer varies across and within industries. Neither necessary nor sufficient in some research fields. Necessary but not sufficient in others. What role for different types of firms (established, small, startup)?