The Location of Industry R&D and the Location of University R&D - How Are They Related?

1. The Location of Industry R&D and the Location of University R&D- How Are They Related? Charlie Karlsson Martin Andersson

2. Introduction (1) • Strong tendencies of a globalisation of R&D, but also • Strong spatial clustering of R&D and related innovative activities • Clustering is assumed to bring about external knowledge economies, typically in the form of knowledge flows, which tend to be spatially bounded

3. Two major performers of R&D: industry and universities

4. Aggregated industrial and university R&D in Sweden: descriptive statistics Industry and universities perform over 90 % of the total R&D in Sweden. Private non-profit organizations and public authorities perform a very limited volume of R&D.

5. R&D and innovation • Generally significant positive effects of industry R&D on patent output • Mixed results concerning the effects of university R&D on patent output • Significant positive effects in the US • No significant effects at the overall level in Sweden in Germany • Significant positive effects in a limited number of industries in Sweden

6. Research problem • Despite a vaste literature on how university and industry R&D affect innovation output, the literature on how the location of university R&D and the location of industry R&D are related to each other is rather limited and consists mainly of case studies. • There is a general lack of knowledge about how the location of industry R&D and the location of university are related, i.e. about their possible mutual interaction.

7. University R&D as an attractor for industry R&D • Much theoretical and empirical literature suggests that the location of industrial R&D is attracted to locations offering good opportunities to take advantages of knowledge flows from universities (and public research institutes), • i.e. there are strong incentives to locate in locations with high accessibility to research universities doing research in fields relevant to industry. • Already as far back as in the 1960’s a number of case studies confirmed the important roles played by Stanford University and MIT for commercial innovation and technological entrepreneurship.

8. Industrial R&D as an attractor for university R&D? • Few studies (Jaffe, 1989; Anselin, Varga & Acs, 1997) • Arguments: • Industry doing R&D in a region may use part of their R&D investments to finance university R&D (commissioned R&D/joint R&D projects, etc.) • Universities in regions with industrial R&D might find it easier to attract R&D funds from national and international sources due to R&D co-operation with industry. • Political decision-makers may decide to start or expand university R&D at locations where industry already is doing R&D to support industrial R&D.

9. Purpose • to analyse how the location of industry R&D and the location of university R&D are related using a simultaneous equation approach, i.e. • does industrial R&D tend to expand in locations with high accessibility to university R&D and vice versa?

10. Accessibility and network formation (1) • Underlying assumption: the extent of links between industry R&D units and university research departments is a function of the spatial proximity between them and the contact intensity of the pertinent activities. • Activities involving knowledge exchange and knowledge generation – such as joint R&D projects between industry and universities – are highly contact intensive.

11. Accessibility and network formation (2) • The outcomes of R&D projects are often uncertain and the transmission of complex knowledge normally requires face-to-face communication. • The implication is that physical accessibility is an important condition for network formation.

12. standard measure of accessibility with exponential distance decay Accessibility and random choice • Assuming that quality and prices are the same at universities across space and that monetary travel costs are proportional to distance, the probability that an industrial R&D unit k located in location i choose to cooperate with university researchers in location j can be expressed as (see e.g. Train, 1993): • Letting , the denominator becomes: An industrial R&D unit with high accessibility to university R&D is likely to have more frequent contacts and durable links with university researchers. A similar approach can be applied to university researchers which wish to establish contacts with industrial R&D units.

13. Accessibility measures Location i’s accessibility to industrial R&D: Location i’s accessibility to university R&D: Locations = municipalities Distinction between: (i) intra-municipal accessibility, (ii) intra-regional accessibility and (iii) extra-regional accessibility, (Johansson, Klaesson & Olsson, 2002) Regions defined as functional regions in terms of commuting patterns

14. Empirical application • University and industrial R&D measured in man-years • Time period: 1995-2001 • Accessibility calculations based upon travel time distances by car between districts within municipalities and between municipalities.

15. The spatial concentration of population and industrial and university R&D across Swedish municipalities in 2001

16. The top 5 municipalities in terms of R&D man-years of university and industrial R&D in 2001 • Despite that the cumulative percentage is larger for the university R&D, industrial R&D show a larger concentration to specific municipalities. • Stockholm and Göteborg alone hosts more than 40 % of Sweden’s total industrial R&D.

17. Equations to be estimated

18. Results (2SLS estimation procedure) **) One municipality (Solna) excluded from the sample

19. Conclusions • Preliminary findings in favor of that industrial R&D tend to expand in locations with high accessibility to university R&D and vice versa. • The results suggest that intra-municipal accessibility is the most important type of accessibility. • The negative parameter for intra-municipal accessibility to university R&D in the equations for the location of university R&D indicates a deconcentration of university R&D • Low R-square for the location of university R&D indicate that also other (political?) factors influence the location of university R&D

20. Future work • Important to make a distinction between • science areas • industrial sectors • Interesting to try to get hold of data illustrating co-operation between industry and research universities, such as • data on funding • joint R&D projects