Knowledge spillover from university research before the national innovation system reform in Japan: localisation, mechanisms, and intermediaries

Abstract

This study examines whether and how university knowledge affected industry R&D in the period when university–industry collaborations encountered institutional barriers. Panel estimation of the regional knowledge production function (1983–1996) revealed that university research had localised impacts on industrial innovations, measured by patents filed by the Japanese firms, with a five-year lag. University–industry joint research did not act as a conduit of university spillover, which suggests that informal channels, such as voluntary transfer of academic inventions in return to donation, worked in the pre-reform period. Intermediaries as a part of regional innovation policy, represented as local public technology centres, exerted a positive impact on industrial innovations in the region. Their immediate impacts suggest the contribution of technology diffusion, such as technical consultation, that help small local firms solve current problems while their far-reaching impacts suggest the importance of knowledge creation, such as own research.

Keywords:

• innovation intermediaries
• knowledge spillover
• national innovation systems
• regional innovation systems
• university-industry collaborations

JEL Classification:

• O25
• O32
• O33
• O38

Notes

1. As well as advantages of the promotion of university technology transfer, previous studies pointed out potential concerns about the extensive involvement of university scientists in the realm of proprietary technology, such as the tragedy of anti-commons (Heller and Eisenberg 1998). Subsequent research revealed working solutions employed in the real world to avoid negative impacts on open science (Walsh, Arora, and Cohen 2003; Sampat mimeo).

2. Other major changes in innovation policy are as follows. The Industrial Technology Enhancement Act of 2000 established procedures under which university scientists can consult for, establish, and manage companies. The Intellectual Property (IP) Basic Law of 2002 had universities establish university IP Headquarters.

3. One exception is Motohashi and Muramatsu (2012) that compared citations of joint patents between university scientists and corporate researchers to those of firms’ solo patents without collaborations before and after 2000. They found a significant difference in forward non self-citations, suggesting far-reaching impacts of university-industry joint patents because of the basic nature of university knowledge. Such tendencies were weakened after 2000 when focusing on collaborations between small firms and universities.

4. Varga (1997) identifies several alternative approaches to examine localized university spillover. One is statistical analysis of the probability of the locational coincidences of firms that cite academic patents and universities that originated the knowledge (Jaffe, Trajtenberg, and Henderson 1993; Almeida and Kogut 1995; Gittelman 2007). The other is questionnaire surveys asking R&D managers the importance of the location of universities in achieving industrial innovations based on academic research (Mansfield 1991, 1995).

5. The NUTS classification (nomenclature of territorial units for statistics) is a hierarchical system that divides up the economic territory of the EU to NUTS 1 (major socio-economic regions), NUTS 2 (basic regions for the application of regional policies), and NUTS 3 (small regions for specific diagnoses).

6. Another important perspective on distance is the cognitive distance. Knowledge transfer would be more efficient when the technological profiles between the provider and user of knowledge are more similar as they are predicted to share greater knowledge base. Using the French data, Autant-Bernard (2001) measured the impact of scientific proximity (i.e., the number of publications of the nearest scientific neighbor of the region) on innovations as well as that of geographical proximity, and found that university research with scientific proximity did not increase patents generated by local firms while the geographical proximity to universities had a positive effect on them. Due to data constraints, this study exclusively addressed the issue of geographical proximity.

7. The significance of geographical proximity may vary not only by the phase of research but also by sector because localized knowledge flow is observed in biotechnology (Zucker et al. 1998b), but not in the semiconductor industry (Almeida and Kogut 1995). Again, this suggests the needs for controlling for the sectoral variations when analysing localized impacts of university research on industrial innovations.

8. Izushi argues that channels of technology transfer from local public technology centres to small firms can be determined through the degree of information gaps between them (Izushi 2003, 2005).

9. See Howells (2006) for numerous and diverse functions of innovation intermediaries.

10. Ronde and Hussler (2005) incorporated R&D expenditure by local authorities as a proxy variable of regional innovation policy and found a negative impact on patents generated by firms in a department. They interpreted the results as reflecting the regional innovation policy in France to support less innovative regions.

11. For more detailed information on the historical background and functions of local public technology centres, see Fukugawa (2008, 2009).

12. Agglomerations make technology transfer by local public technology centres more efficient because they enable policymakers to target and enable additional technology transfer activities to exert scale economies (Storey 1994). This study controls for the regional variations in the degree of industrial specialization by using location quotients.

13. An average prefecture is approximately 8000 sq km, which is even smaller than an average state in the US (approximately 196,500 sq. km) and larger than an average department in metropolitan France (approximately 5700 sq km).

14. The Institute of Intellectual Property (IIP) is an external body of the Japan Patent Office. The IIP Patent Database collects all information on patents applied for the Japan Patent Office.

15. The proportion of labor costs in R&D expenditure is approximately 45% according to Survey of Research and Development by the Ministry of Internal Affairs and Communications.

16. Most of the innovative attempts are doomed to fail, thus are not patented. Furthermore, not all the inventions are patented since patent propensity greatly varies according to sectors and firm size. Regardless of these potential shortcomings, this proxy variable is selected because of the comprehensiveness of the IIP Patent Database in representing innovative activities by the Japanese firms.

17. Suppose X is the specific manufacturing sector's employment in a prefecture and Y is the manufacturing employment in the prefecture. If X′ and Y′ are that manufacturing sector's employment in Japan and manufacturing employment in Japan, respectively, then the location quotient is (X/Y)/(X′/Y′).

18. It should be noted that studies cited here used R&D flow whereas R&D stock was used in this study, which makes it impossible to make direct comparison. My intention here is not international comparison but illustration of the diminishing impact of industrial R&D and the presence of university spillover in the pre-reform period in Japan.

19. Fukugawa (2005) shows that the research quality of faculty, represented as forward citations to publications, is significantly correlated with firm size of industrial partners, and local public technology centres are important intermediaries for smaller firms to initiate collaborations with university scientists. This implies that local public technology centres may have linked smaller firms with less absorptive capacity with universities with lower research quality.

20. Other results of Equation (4) are statistically significant at the 10% level.

21. The share of high-impact scientific papers in natural sciences and engineering was 0.8% (16th in the world) for South Korea and 0.5% (21st) for Taiwan in 2000 while they ranked 13th (1.5%) and 17th (1.0%), respectively, in 2010 (NISTEP 2013).

22. Amid serious concerns over decreasing competitive advantage in the manufacturing industry, the US government benchmarked local public technology centres in its industrial modernization program (OTA 1990).

23. Fukugawa (2008) evaluates how technology transfer has been performed by local public technology centres in accordance with regional innovation systems in which they are located.

Additional information

Funding

This research was supported by Grants-in-Aid for Scientific Research from the Japan Society for the Promotion of Science [15K03411], Seimeikai, Zengin Foundation for Studies on Economics and Finance, and Kikawada Foundation.