Green technologies and Smart Specialisation Strategies: a European patent-based analysis of the intertwining of technological relatedness and key enabling technologies

ABSTRACT

This paper investigates the green diversification of regional technologies and its consistency with the Smart Specialisation logic of related diversification. It also analyses the role that key enabling technologies (KETs), as a key pillar of Smart Specialisation Strategies (RIS3), have in green branching. Working on a patent-based panel (1981–2013) of 240 European regions, it is found that the relatedness to pre-existing knowledge makes a new green-tech specialization more probable. This holds true for the relatedness to both green and non-green pre-existing knowledge, but to a greater extent for the latter. Thus, the ‘hybridization’ of non-environmental technologies seems to require lower cognitive proximity than ‘pure’ green branching. Regional KETs also facilitate the transition towards sustainable technologies and negatively moderate the green impact of the relatedness to pre-existing technologies, both green and non-green. The results confirm that Smart Specialisation policies and the support to KETs could also help regions move towards environmental sustainability.

KEYWORDS:

• Smart Specialisation Strategies
• environmental sustainability
• regional branching
• key enabling technologies

JEL:

• O31
• O33
• R11
• R58

DISCLOSURE STATEMENT

No potential conflict of interest was reported by the authors.

Notes

1. As shown in Tables A1 and A2 in Appendix A in the supplemental data online, the identification of green technologies and KETs relies on both International Patent Classification (IPC) and Cooperative Patent Classification (CPC) codes. We have accordingly used the union of the two classifications to obtain regions’ patenting activity in the green and KETs domain. In the same respect, it should also be noted that the OECD’s Env Tech classification is based on technological classes at different levels of disaggregation. While we have used the full-length specification of technological classes to tag regional patents, the data have been eventually homogenized by using the four-digit disaggregation level. In this process, being green has been considered a dominant feature, such that if, for example, a four-digit class consists of some nine-digits classes that are non-green and one nine-digit class that is green, then the class is deemed to be green. We acknowledge that this might imply an overestimation of the green patents.

2. The RTA index is built up by working out a standard Balassa measurement of the patent applications (PAT_ist) registered within regional boundaries. As a rule, (the absence of) a green-tech specialization is indicated with RTA_GREEN_ist (between 0 and 1) being > 1. However, given the relevance of the transition from the absence and the presence of a specialization, a threshold > 1 has also been used as a robustness check (see Appendix A2 and Table A3 in the supplemental data online). In the same respect, moving averages have been used in constructing this and other patent-based indicators (e.g., KETs) to attenuate the problems of their erratic trends over time (see Appendix A2 online).

3. Moderation in the inclusion of controls is imposed by the need of attenuating the effects on our estimates of incidental parameter problems (see below) and by the non- or very limited availability of technology-region and/or region-specific data for the long period of time we have considered (1981–2013).

4. Regional per capita income (PCI) has alternatively been used in a robustness check (see Table A4 in Appendix A in the supplemental data online).

5. This is an alternative to the Box–Cox transformations, defined by: $\mathrm{i}\mathrm{n}\mathrm{v}\mathrm{e}\mathrm{r}\mathrm{s}\mathrm{e}y=\mathrm{l}\mathrm{o}\mathrm{g}[y_i+(y_i^2+1{)}^{1/2}]$. Except for very small values of y, the inverse sine can be interpreted as a standard logarithmic variable. However, unlike a logarithmic variable, the inverse hyperbolic sine is defined at zero (Burbidge, Magee, & Robb, 1988; Johnson, 1949; MacKinnon & McGee 1990).

6. Results with other values of k in the domain [1–5] are robust and available from the authors upon request.

7. While apparently modest, the size of these marginal effects is consistent with that obtained for relatedness/density variables by other recent studies (e.g., Broekel & Mewes, 2017; Cortinovis, Xiao, Boschma, & van Oort, 2017).

8. The present paper provides an empirical analysis that does not allow one to appreciate the very peculiar dynamics involving connections among specific green technologies and KETs, and the role of local institutions and research infrastructures. This is beyond the scope of this study and will be the focus of our future research agenda.