Not just financial support? Another role of public subsidy in university–industry research collaborations

Abstract

Management of university–industry research collaboration (hereafter UIC) is the key to its success. In this respect, government can play an essential role in UIC. A public subsidy for research and development (hereafter R&D) is not only an important financial support for UIC but may also serve as a useful means of promoting trust among UIC members, resulting in higher innovation performance. However, few empirical studies have investigated the role a public R&D subsidy plays in promoting trust in UIC. To this end, by using original survey data, this study examines empirically whether a public R&D subsidy for UIC contributes to trust formation and, thus, to higher innovation performance based on trust. Our findings suggest that a public R&D subsidy promotes trust formation, which then increases the innovation performance of UIC participants, partially mediating the more direct effects of R&D subsidy on innovation performance.

Keywords:

• public subsidy
• R&D
• research collaboration
• university
• trust

JEL Classifications:

• H25
• L24
• O32
• O38

Acknowledgements

Early versions of this paper were presented at the 9th IECER (Interdisciplinary European Conference on Entrepreneurship Research) in Munich, Germany, in February 2011; the 38th Annual Conference of EARIE (European Association for Research in Industrial Economics) in Stockholm, Sweden, in September 2011; the 10th JEPA (Japan Economic Policy Association) International Conference in Nishinomiya, Japan, in November 2011; the seminars at ZEW (Zentrum für Europäische Wirtschaftsforschung) in Mannheim, Germany (in February 2011), KU Leuven, Belgium, Utrecht University, the Netherlands, and the University of Bergamo, Italy (in March 2011); and the Annual Spring Meeting of the JEA (Japanese Economic Association) in Sapporo, Japan, in June 2012. The authors are grateful for the valuable comments by seminar participants, including Michael Dowling, Georg Licht, Bettina Peters, Reinhilde Veugelers, René Belderbos, Dirk Czarnitzki, Erik Stam, Sadao Nagaoka, Jong-Rong Chen, and Jun Suzuki, as well as two anonymous referees and the managing editor of this journal. The usual disclaimer applies.

Funding

This study is an outcome of the research project entitled ‘The Determinants, Organization, and Outcomes of University–Industry Interaction in Innovation Systems’, which was funded by the Volkswagen Foundation in Germany.

Notes

1. Of late, public support for R&D consortia has been increasingly provided as part of intellectual cluster policies. See Nishimura and Okamuro (2011) for Japan, Eickelpasch and Fritsch (2005) and Engel et al. (2013) for Germany, and Fontagné et al. (2013) and Martin, Mayer, and Mayneris (2011) for France. Okamuro and Nishimura (2011) provide a comparative study of recent national cluster policies in Japan, Germany, and France.

2. Several studies investigate the innovation performance of publicly funded cooperative R&D projects in Japan (Branstetter and Sakakibara 2002), the USA (Irwin and Klenow 1996), and the EU (Bayona-Sáez and García-Marco 2010; Matt, Robin, and Wolff 2012). However, these projects are based on R&D cooperation between private firms and thus include no or few UIC. Moreover, these studies do not investigate the mechanism of increased innovation output.

3. We conducted a Wu–Hausman test with regard to the endogeneity of a public subsidy. The result shows that the null hypothesis that the variable d_subsidy is exogenous is rejected at the 5% significance level.

4. Among 1455 responding firms that did not engage in or complete a UIC, 7.8% (114 firms) were still conducting a UIC project when the survey was carried out, and thus excluded from the sample frame. The largest reason for non-participation was the lack of internal resources for UIC (46.4%, 680 firms). There were 139 firms (9.5%) that were interested in UIC but could not find appropriate partners, and 10.3% (151 firms) were not interested in UIC mainly because in-house R&D is enough for them. Our econometric estimation of the determinants of UIC participation using a probit model shows that R&D intensity and the CEO's higher education positively affect UIC participation, while firm age and firm size have no impact after controlling for R&D intensity, CEO's education, and technology field. These results are available upon request.

5. The firms (projects) that did not obtain a public subsidy comprise both firms (projects) that did not apply for a public subsidy and those whose applications were rejected. From our questionnaire, we cannot distinguish between these groups. However, this would not cause severe empirical problems since we control for the endogeneity of obtaining a public subsidy in the empirical estimation.

6. No further information was available about a public subsidy including its rules, targets, or processes.

7. Of the projects in our sample, 40% include two or more universities. Therefore, by focusing on the most important university partner, we cannot exclude the possibility of upward bias in these projects, although we control for the number of project members and the (subjective evaluation of) research capability of the university partners in the empirical analysis.

8. This is not a minor increase because the mean value is −0.12 and the maximum is less than 3.

9. On weak instrument biases, see, for example, Nelson and Startz (1988) and Staiger and Stock (1997).

10. Regarding the mediating effect of trust, Santoro and Saparito (2003) report that trust mediates the relationship between communication behavior and UIC outcomes.

11. The estimation results remain unchanged even if we exclude prefecture_ratio and industry_ratio from the instrumental variables.

12. Podsakoff et al. (2003) provide a detailed and critical review of the potential sources and mechanisms of common method biases and some techniques for controlling them.

13. As mentioned before, several respondents preferred anonymity and did not provide their affiliation and name, but we could identify them using the numbers printed on the response envelope.

14. If we find that the mean values of the relevant variables for anonymous respondents are constantly lower than those for another group, we regard these differences as ‘consistent’.

15. Information on this program was obtained from the website of METI. http://www.meti.go.jp/ (accessed November 6, 2010).