Technological diversification and Smart Specialisation: the role of cooperation

ABSTRACT

Smart Specialisation is closely associated with the concept of diversification. For a better understanding of Smart Specialisation, this paper examines one little-explored explanatory factor of technological diversification: cooperation within and between regions. Using Organisation for Economic Co-operation and Development (OECD) REGPAT data on co-applications for patents, the empirical analysis investigates the role of cooperation between organizations on technological diversification in 226 European regions over 10 five-year periods, 2000–13. Cooperation within and between regions emerges as an important determinant of regional diversification, but both forms of cooperation should evolve hand in hand – singly, each form may prove ineffective at boosting regional diversification.

KEYWORDS:

• Smart Specialisation
• diversification
• patents
• European regions
• cooperation

JEL:

• O31
• O38
• R11

ACKNOWLEDGEMENTS

The author thanks Fulvio Castellacci for guidance, help and advice with this paper. His comments significantly improved the quality of the manuscript. The author also thanks Thomas Brenner, Sophie Urmetzer, Ron Boschma, Henrik Schwabe, Philippe Laredo, Magnus Gulbrandsen, three anonymous referees, the journal Editor, David Bailey, and the Associate Editor, Michael Fritsch, for detailed and helpful comments on earlier versions. Thanks also to the organizers of and participants at the 2017 European Meeting on Applied Evolutionary Economics (EMAEE) in Strasbourg, the 2017 Heilbronner Symposium in Heilbronn, and the Special Session on ‘Territorial Innovation Models, Smart Specialisation and Public Policies’ at the 24th Congress of the Portuguese Association for Regional Development in Covilhã, 2017. The responsibility for any remaining errors and shortcomings is the author’s alone.

DISCLOSURE STATEMENT

No potential conflict of interest was reported by the author.

ORCID

Artur Santoalha http://orcid.org/0000-0001-6942-6948

Notes

1. February 2016 version.

2. Patent applications are accounted for by the fractional counting method and regionalized by applicants’ address (Maraut, Dernis, Webb, Spiezia, & Guellec, 2008).

3. Following Xiao et al. (2018), this paper uses overlapping periods to maximize the number of available observations. However, this is not problematic (see the coefficient of the lagged dependent variable in Table 3).

4. 2010 classification.

5. The EU-28, except Slovenia.

6. As a robustness check, the empirical strategy proposed in the fifth section was also implemented considering a higher level of disaggregation (633 IPC subclasses), with results similar to those presented in the sixth section.

7. In most cases, the gain/loss of technologies by the regions is not the result of slight variations in the RCA. Such movements generally result from the introduction of completely new technologies/full abandonment of older ones (see Tables A1 and A2 in Appendix A in the supplemental data online).

8. For the computation of these indexes, patent applications are counted as whole-counting (Maraut et al., 2008).

9. System GMM failed to reject the nonexistence of an autocorrelation of order 2.

10. Improvement in the ratio necessarily implies an improvement in regional diversification.