New innovation policies in science based economy era: Empirical analysis by linked dataset of scientific papers, patents and firm level financial data
Use of genetic engineering in pharmaceutical R&D, quantum physics in semiconductor fabrication process and nanotechnology as a source of new material in chemical and metal products industries; all of these examples show that scientific findings become to be important inputs to industrial innovation. In such “science based economy” era, innovation policies requires more linkages with science and technology policies. Specifically, S&T policies focusing on universities and public research institutes (generating scientific findings) should be coordinate with industrial innovation policies for industry (appropriation of scientific findings into economic output). However, the actual process of converting scientific knowledge into new product and process has not been investigated enough. This research fills such gap between policy needs and analytical findings by providing statistical analysis on micro structure of industry use of scientific findings, as well as measuring the impact of academic research on industrial performance.
The understanding of the process and the impact of academic research on industrial performance is important particularly in Japan, where the reforms to science and technology innovation systems are undergoing. This policy action began with technology licensing organization (TLO) laws in 1998, followed by the Japanese version of the Bayh-Dole Act. In addition, both public research institutes (PRIs) and national universities were incorporated in 2001 and 2004, respectively, which facilitates competition among these institutes for innovation activities. Consequently, the number of joint patents with corporations and joint research projects in PRIs and universities is on the rise (Motohashi and Muramatsu, 2012). However, the impact of those policies on innovation performance at firms has not investigated enough. The direct interactions between research sector and industry can be measured by joint patenting, citation, the number of R&D collaborations etc, but it is more important to track to what extent scientific findings at universities and PRIs are involved in innovation outputs at firms such as new products and processes. Given the fact that most public funding of research and development flows to universities, public research institutions, or other parts of the science sector, it is important to measure the degree of contributions of academic activities to industrial innovation outputs, as the accountability of public spending on R&D is more and more required.
This research proposal directly addresses such policy questions as “to what extent scientific activities at universities and PRIs contribute to industrial innovation?” and “what are new innovation policies for maximizing the impact of scientific activities, without damaging academic integrity at research sector?”. Specifically, I will construct the linked datasets of bibliographic information on patents and academic papers at the level of individual researchers in Japan, the US, Europe, and China. Matches will be made between patent inventors and authors of papers, making it possible to identify inventions with possible industrial applications that also involved academic research (academic papers). By using this information it will be possible to clarify the importance of scientific findings in industrial innovation; what industries or technology fields are getting hotter, or the academic research areas that contribute to innovation within industry. The strengths of links between academia (as shown by academic papers) and technology (as shown by patents) will show how the impact of patents on corporate performance changes, and the econometric analysis will make possible an understanding of the economic impact of academic research at the national level. Decreases in the speed of global economic growth and increases in pension burdens for a burgeoning population of elderly have placed Japan and the countries of Europe in a severe fiscal bind. It is in these circumstances that we see arguments for reducing public funding of universities and public research institutions. The results of this study should be an important input to the policy debate in terms of the size of public R&D spending under severe budget constraints.
Many empirical studies related to industry-academia partnerships have been done in the past. However, these can be generally placed into two categories: those related to factors (including inhibiting factors) behind universities’ or corporations’ industry-academia partnership efforts; and analyses of the impact of industry-academia partnerships on universities and corporations. However, outcomes of these studies stop at providing analytical data and fragmented information in the various target fields studied. There are few studies that provide a systematic analysis of innovation processes beyond that of the university research phase to industrial performance (Ankrah and Al-Tabbaa, 2015). In addition, even in this case, the aforementioned study only examines the pharmaceutical industry, where tying academic research to new products is relatively easy. There are practically no inter-industry analyses. Additionally, there are many examples of research on industry-academia partnerships which use data on individual researchers, and which primarily focus on those cases in which university professors and students participated in corporate projects in examining the impact on university education and research activities (Agrawal and Henderson, 2002). However, there are few examples of research on industry-academia partnership projects as seen by corporations. This is because of the difficulty in obtaining information from individual researchers within corporations. Research on corporate researchers using patent data can be found from Professor Lee Fleming of UC Berkeley, though that study did not use a database of academic papers, and could therefore not analyze the impact of academic papers on academia. Accordingly, building a database that allows the systematic analysis of the innovation process from academic research to innovation output (patents and new products) would be a globally revolutionary accomplishment.
As to research methods for this project, below I describe (1) building a database and (2) empirical analyses that use that database in more detail.
In building a database, I shall focus my efforts on researcher-level connections in databases of science and technology papers (scientific research) and patents (interim outputs of the innovation process). The target data is global, and includes the US, Europe, and China, which lead out on global work in science and technology. Specifically, academic paper data will come from the Scopus database of Elsevier, B.V. This database provides identifying information on researchers that show up as authors of papers. Information on inventors in patent data is not identified in the Scopus database, which therefore requires disambiguation. Learning from prior research, a clustering analysis will be done on such information as inventor names, addresses, institutions, patent technology classifications, citations, etc., and then machine-learning based identification will be done using teacher data (ASE: Approximate Structuring Equivalence, Li, et al., 2014). It is possible to use PATSTAT, a global patent database, as a sources for this data, though the accuracy of identification can be improved by using Chinese characters for Japanese and Chinese names and addresses, which will entail improvements to the ASE-based algorithm. In addition to the Japanese (IIP patent database: Goto and Motohashi, 2007) and China (SIPODB: Intellectual Property Office data), identification of inventors in patent applications from Japan, the US, Europe, and China will also be done. As to identification of inventor data from Japanese patents, a preliminary work has been conducted under the NISTEP project (Ikeuchi et. al, 2017). The database used for this analysis is completed with author IDs in academic papers and individual inventor information obtained from patent data.
Moreover, I shall conduct a questionnaire survey of incentives for researchers to engage in external partnerships (industry-academia partnerships in this case), and conduct research related to deciding factors in external partnerships for both universities and corporations. Using a database with bibliographic information such as papers and patents makes it possible to describe phenomena such as technology spillovers from networks of corporations and universities that may be co-authoring papers, as well as citation information. However, these types of data are insufficient to analyze the causal relationships behind the other phenomena (deciding factors in industry-academia interactions). Building a behavioral model at the individual researcher level using a policy simulation requires adding information by modifying the questionnaire with networking incentives. First, approximately ten thousand researchers will be extracted from each university and corporation, out of researcher information generated from core data on authors and theses (there are several million inventors in from four countries and several tens and thousands of orders in Japan’s patent information). The percentage of researchers participating in industry-academia partnership projects in universities is known to be extremely low (4-5% according to a European analysis; Lissoni, et al., 2009). Half of the sample of researchers with observable partnerships with corporations for patents, as well as half of the sample of researchers with no partnerships will be selected, with a survey done of these researchers as to the reason for participating (or not participating) in the partnership; individual attributes such as the nature of research, restrictions on research costs, attitudes toward partnering with corporations, etc.; organizational attributes such as assessment criteria of their university or faculty, teaching load, etc.); and system attributes (industry-academia partnership policies of national and local governments, etc.). Corporate researchers will also be separated into samples of those that have participated in partnerships with universities and those that haven’t, with each sample being similarly surveyed. Though there are few examples of analyses of incentives for industry-academia partnerships with corporate researchers, many studies have analyzed intellectual property knowhow and licensing via the technology market (Arora and Gambardella, 2010, etc.). Using the outcomes of these studies as reference points, I shall use the questionnaire to identify individual characteristics (freedom to use research projects, relationships with university researchers, including information exchanges, etc.) in addition to corporate characteristics (supplementary technology, product market competitiveness, importance of scientific findings, etc.) and system characteristics (enforcement of intellectual property rights, etc.).
For (2) economic analysis, I will connect the aforementioned questionnaire survey data to papers, patents, and other bibliographic information to see the impact on corporate innovation performance of (1) determining factors from industry-academia partnerships, as seen by both universities and corporations; and (2) industry-academia partnerships and scientific research outcomes within universities. As to the former, I shall build an econometric model of incentives for researchers to partner with corporations (or universities) that includes individual characteristics of researchers, organizational characteristics, and system characteristics. Unlike existing research outcomes, this study will be able to make international comparisons, such as whether low researcher mobility, a characteristic of Japan’s innovation system, inhibits partnerships between industry and academia. Some theorize that having star scientists or personnel that act as bridges is important when forming industry-academia partnerships (Zucker and Darby, 2001), and it will be possible to assess the validity of this notion by adding various types of centrality indices researchers get from social network analyses as explanatory variables in our econometric model. Research on innovation performance of industry-academia partnerships, (2) above, will be analyzed using the econometric model as to the indirect impact of foundational research within universities, and not merely industry-academia partnership activities, on corporate innovation (patents, new product development, etc.). Many studies have observed a positive correlation between industry-academia partnerships and corporate innovation, though university researchers working in industry-academia partnerships with research that is understaffed, can exert a negative impact on corporate innovation. In this study, this sort of indirect impact and other influences of industry-academia partnerships can be modeled for use in analysis.
Agrawal, A. and R. Henderson (2002), Putting patens in context: exploring knowledge transfer from MIT, Management Science, 48, 44-60
Ankrah, S. and O. Al-Tabbaa (2015), University-industry collaboration: A systematic review, Scandinavian Journal of Management, 31, 387-408
Arora, A. and A. Gambardella (2010), Ideas for rent: an overview of markets for technology, Industrial and Corporate Change, 19(3) 775-803
Goto, A. and K. Motohashi (2007), Construction of a Japanese Patent Database and a first look at Japanese patenting activities, Research Policy, 36(9), 1431-1442
Kenta Ikeuchi, Kazuyuki Motohashi, Ryuichi Tamura, Naotoshi Tsukada (2017) Measuring Science Intensity of Industry using Linked Dataset of Science, Technology and Industry , RIETI Discussion Paper, 17-E-056, 2017/03
Li, GC., R Lai, A D’Amour, DM Doolin, Y Sun, VI Torvik, ZY Amy, L Fleming (2014), Disambiguation and co-authorship networks of the US patent inventor database (1975–2010), Research Policy 43 (6), 941-955
Lissoni, F., Llerena et. al (2009), Academic Patenting in Europe: Evidence from France, Italy and Sweden from KEINS Database, in Learning to Compete in European Universities, Cheltenham, Edward Elgar, 187-218
Motohashi, K. and S. Muramatsu (2012),Examining the university industry collaboration policy in Japan: Patent analysis, Technology in Society, 34(2), 149-162
Zucker, LG and MR Darby (2001), Capturing technological opportunity via Japan's star scientists: Evidence from Japanese firms' biotech patents and products, Journal of Technology Transfer, 26(1), 37-58
Kazuyuki Motohashi is a professor at the Department of Technology Management for Innovation, Graduate School of Engineering. He is also served as a Faculty Fellow of RIETI (Research Institute of Economy, Trade and Industry) and Visiting Scholar of NISTEP (National Institute of Science and Technology Policy). Until this year, he had taken various positions at the Ministry of Economy, Trade and Industry of the Japanese Government, economist at OECD and associate professor at Hitotsubashi University.
His research interest covers a broad range of issues in economic and statistical analysis of innovation, including economic impacts of information technology, international comparison of productivity, national innovation system focusing on science and industry linkages and SME innovation and entrepreneurship policy. He has published several papers and books on above issues, including Global Business Strategy: Multinationals Venturing into Emerging Markets (2015, Springer) and Productivity in Asia: Economic Growth and Competitiveness (2007, Edward Elgar). He was awarded Master of Engineering from University of Tokyo, MBA from Cornell University and Ph.D. in business and commerce from Keio University.
Prof. Motohashi is a recipient of the 2017 Michelin Research Fellowship at the Center for French-Japanese Advanced Studies, EHESS in Paris. He will conduct his research program titled "New innovation policies in science based economy era" in Paris from April 1st, 2017 to December 31st, 2017.
Personal Webpage: URL: http://www.mo.t.u-tokyo.ac.jp/