University industry research relations and intellectual property: Some insights from the United States Bronwyn H. Hall UNU MERIT, University of Maastricht University of California at Berkeley NBER, IFS and NIESR, London
Outline for the presentation 1. Overview 2. Bayh Dole and tech transfer 3. Survey evidence on research partnerships between university and industry 1. motivation of the two partners 2. results from a small study of ATP partnerships with university members 4. Effects of patenting on subsequent research November 2014 UNU-MERIT 2
General setting Production of scientific knowledge, software, databases, share common features many actors contribute; production is cumulative some producers are also consumers Incentive problems due to indivisibility and inappropriability Two different modes of production have emerged November 2014 UNU-MERIT 3
Two worlds of invention Republic of science model Incentives from returns to priority Encourages early publication and dissemination Use of others output at low cost appropriate citation; reciprocity Private sector with strong IP Incentives primarily financial IPRs encourage publication, but only of codified knowledge; trade secrecy often used in addition Use of others output requires payment or reciprocal cross licensing (transactions costs) November 2014 UNU-MERIT 4
Gambardella Hall (2006) model Equilibrium researchers do both, but public domain share is unstable without coordination Coordination sometimes achieved via social pressure or social norms (academic research, open source licenses) Defection occurs when Number of participants grows large Size of the reward increases IP protection is available Numerous examples in early stage industry (collective invention) as well as in academia November 2014 UNU-MERIT 5
Implications for research software and databases More likely to be privatized because Greater market demand (higher profit) Scientific norms are weaker; reputation effects not as great Require maintenance and user help, and these are not rewarded by priority More non producing consumers Public and private production can coexist (as in our model) As market grows, privatization more likely individuals earn discrete profits from defecting, but have infinitesimal effect on value of public good November 2014 UNU-MERIT 6
University industry research relations in the United States Long history 100+ years, in agriculture and manufacturing Derives from the nature of US (public) universities and their close links to local communities See Mowery and Rosenberg, Technology and the Pursuit of Economic Growth. Increase in past 15 25 years Wide variety of organizational forms Still a relatively small fraction of university research funding in the U.S. (~5 to 7 percent, reaching a peak in 2000 and then falling slightly) November 2014 UNU-MERIT 7
Variety of partnership types Industry support of specific university researchers via grants and consulting Large laboratories funded by industry consortia involving 10s to 100s of firms, such as the Stanford Center for Integrated Systems Quasi permanent FFRDCs and UIRCs, partially funded by federal government Onetime projects that involve a university as a partner Ordinary research joint venture (RJV) with a specific goal Government cost shared RJV, such as those funded by ATP (now discontinued) comprehensive survey data that includes all types of funding does not exist studies usually based on one particular type, difficult to measure total trends November 2014 UNU-MERIT 8
Source of university research funding - U.S. 100% 90% 80% 70% 60% 50% 40% Industry Self and other Federal govt. 30% 20% 10% 0% 1953 1960 1970 1980 1990 2000 2004 2011 November 2014 UNU-MERIT 9
OECD is similar to US Sources of academic research funding Year 1981 2003 Government funding 82% 70% Industrial funding 3% 6% Self funded 13% 16% Other 4% 8% Source: Vincent-Lancrin (2006), as cited in Foray and Lissoni (2010) November 2014 UNU-MERIT 10
Some facts about US tech transfer Did Bayh Dole cause an upsurge in patenting by US universities? Probably not the main cause: Patenting per R&D dollar by universities was already growing (about 4% per annum) there was no increase in the rate of growth Slight increase in the rate of growth of university patenting as a share of all patenting However, it did increase the rate of growth of the share of universities with tech transfer offices November 2014 UNU-MERIT 11
November 2014 UNU-MERIT 12
0.05 US university patenting as a share of all US origin patents, by grant year 0.04 0.04 0.03 0.03 0.02 0.02 0.01 Bayh-Dole Act 0.01 0.00 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 2011 November 2014 UNU-MERIT 13
Some facts about US tech transfer Is tech transfer via TTOs an important source of income for universities in the US? Definitely not: licenses do not cover the costs of most tech transfer offices (>80% under water during past 20 years) Earners are fairly stable, and account for bulk of federal R&D funding Most patents earn little revenue TTOs are a small piece of the action UC system TTO profits = $16M/year in 1991 2003 One year industry contribution to UC research in 2003 was $235M (15 times as much) November 2014 UNU-MERIT 14
Some facts about US tech transfer Is tech transfer via (patent) license the most important way to transfer knowledge from university to industry? No: other methods such as publications, informal contacts, and conferences are much more important Source: Cohen, Nelson, and Walsh % of respondents rating info source as important Publications 41.2 Informal contact 35.6 Meetings/conferences 35.1 Consulting 31.8 Contract research 20.9 Recent hires 19.6 Cooperative research 17.9 Patents 17.5 Licenses 9.5 Personnel exchange 5.8 November 2014 UNU-MERIT 15
Some facts about US tech transfer Is university research essential to innovation in all sectors? Definitely not: primary areas where it was important in mid 1980s were food, agriculture, wood and paper, drugs, and some electronics products (15 out of 50 sectors surveyed) In mid 1990s, add nonferrous metals and specialized industrial machinery Over half of university patenting is in the biomedical sector Sources: Yale (1987) and Carnegie Mellon (2002) surveys of R&D in industry November 2014 UNU-MERIT 16
Conclusions? Bayh Dole and the associated growth of TTOs not the essential ingredient in promoting university scienceindustry linkage, nor the solution to the problem of funding university research. This conclusion does not imply that universityindustry linkages are not an important part of the innovation system. Next topic: Survey evidence on motivations and barriers to partnering November 2014 UNU-MERIT 17
Why partner? Industry Access to new and complementary research also found to be important by Cohen et al survey (1997) Development of new products Maintaining a relationship with the university Obtaining new patents Solving technical problems Less important Improving products, recruiting students Source: survey of ~400 R&D managers by Lee (2000) November 2014 UNU-MERIT 18
Why partner? university researchers Funds for research assistance, lab equipment, and one s own research agenda. Insights into own research; field test theory and empirical research. Less important Practical knowledge useful for teaching Student internships and job placement Patentable inventions and business opportunities Source: survey of ~400 R&D managers by Lee (2000) November 2014 UNU-MERIT 19
Evaluating the benefits Henderson and Cockburn (1996) in pharmaceutical industry, access to university research enhances sales, R&D productivity, and patenting Zucker, Darby, and Armstrong (2001) collaborating (publishing) with star university scientists important for firm performance in biotechnology Adams, Chiang, and Starkey (2001) Ind. Univ. Cooperative Research Centers, especially those funded by NSF, promote tech transfer and increase patenting rates at industrial laboratories. Rosenberg and Nelson (1994) university research enhances and stimulates R&D in industry, rather than substituting for it. Pavitt (1998) augments capacity of business to solve complex problems. Hall, Link, and Scott (2001) studied university industry RJVs partially funded by ATP (NIST) Universities are included in research projects that involve new science and where there is more difficulty in assimilating knowledge. Performance commercialization may take longer, but less likely to fail November 2014 UNU-MERIT 20
Barriers that inhibit universities from partnering with industry? Hall, Link, and Scott (2001) based on interviews (n=9, so caution is urged): Most significant barrier inhibiting partnerships with industry intellectual property concerns, specifically patenting rights. Universities want to be able to patent whatever research results from their partnering relationship, but found industry extremely difficult to deal with on this issue Publication rights were, for the most part, an non issue from the perspective of the university. Small companies Subcontract with universities rather than include them as a research partner Universities reported a higher false start rate with small companies primarily because they seemed less familiar with the university bureaucracy. (less tolerant of?) November 2014 UNU-MERIT 21
Barriers that inhibit industry from partnering with universities? Intellectual property reasons frequently cited. Representative comments: IP is often a stumbling block for collaborations because many universities want to publish results prior to IP protection, and sometimes will not grant exclusivity of results. Universities have an over inflated view of their intellectual property value, and university licensing officers have an over inflated view of the value they bring to the project. November 2014 UNU-MERIT 22
Evidence on IP impacts When university scientists patent upstream results, is there an impact on scientific research? Walsh et al. 2000 industry adapting vast majority of respondents say that there are no cases where valuable research projects were stopped due to IP problems. Firms take licenses, invent around patents, infringe, develop public databases, challenge patents in court November 2014 UNU-MERIT 23
Evidence on IP impacts Murray & Stern 2007 life science patent paper pairs citation to research in the paper declines 9 17% after patent grant Walsh et al 2007 survey 507 academic biomedical researchers, access to knowledge inputs largely unaffected by patents. access to other researchers cell lines, reagents, or unpublished information more problematic due to scientific competition, the cost of providing materials, a history of commercial activity on the part of the prospective supplier, and whether the material in question is itself a drug. Williams (2011) compare citation to genes sequenced by Celera and Human Genome Project (matched samples, and within gene panel evidence) IP protection reduced subsequent scientific research and product development outcomes by about 30% November 2014 UNU-MERIT 24
Evidence on IP impacts Thompson, Mowery, Ziedonis 2011 look at ~500 UC patent paper and MTA paper pairs find negative impact only in life sciences, which disappears when matched on pre grant cites small negative impact for MTAs with the private sector, and licenses on research tools Nagaoka & Okada (2014) introduction of pre grant publication in US increased inventor non self citation Nagaoka evidence on the introduction of a grace period in Japan Earlier paper publication increased citation based knowledge flows November 2014 UNU-MERIT 25
Conclusions Little impact broadly but Maybe in life sciences Definitely in MTAs As predicted by Gambardella & Hall (2006), In areas where rewards are large When IP becomes available or stronger (e.g., subject matter broadening) On data and materials inputs that are costly to provide and where buyers may not be producers (specialized expertise) November 2014 UNU-MERIT 26
Selected references (BHH) Papers available on my website: University Industry Research Partnerships in the United States, in J. P. Contzen, D. Gibson, and M. V. Heitor (eds.), Rethinking Science Systems and Innovation Policies, Purdue University Press, Winter 2004. "Universities as Research Partners," with A. N. Link and J. T. Scott, Review of Economics and Statistics 85 (2003): 485 491. "Barriers Inhibiting Industry from Partnering with Universities: Evidence from the Advanced Technology Program," with A. N. Link and J. T. Scott, Journal of Technology Transfer 26: 87 98 (2000). Proprietary vs. Public Domain Licensing of Software and Research Products, with Alfonso Gambardella, Research Policy, forthcoming 2006. November 2014 UNU-MERIT 27
Selected references (others) Two special journal issues: Journal of Technology Transfer (2000, 2001): R. Nelson, M. Thursby, J. Thursby, A. Link, J. Scott, M. Feldman, I. Feller, L. Zucker, M. Darby, J. Adams, W. Powell, S. Maital, P. Stephan, R. Morgan, N. Rosenberg, D. Blumenthal, etc. Research Policy (2006), special issue entitled Property and the Pursuit of Knowledge: IPR issues affecting scientific research, edited by P. A. David and B. H. Hall Mowery, Nelson, Sampat, and Ziedonis, 2004, Ivory Tower and Industrial Innovation, Stanford University Press. Foray and Lissoni, 2010, University Research and Public Private Interaction, in Hall and Rosenberg (eds.), Handbook of the Economics of Innovation, Elsevier. November 2014 UNU-MERIT 28