The non-linear effect of technological diversification on regional productivity: implications for growth and Smart Specialisation Strategies

ABSTRACT

With a sample of 268 European Union (EU) regions and multilevel modelling, we demonstrate that different technological diversification measures measured as coherence and entropy-variety have different non-linear effects on regional productivity growth. These non-linear effects work in opposite directions to each other. Our analysis shows that higher regional productivity returns can be found in regions investing both around their existing technological capabilities as well as in more distant knowledge domains. Our findings have significant implications for understanding regional productivity growth processes and the implementation of Smart Specialisation Strategies.

KEYWORDS:

• regional productivity
• technological structure
• relatedness
• diversification
• Smart Specialisation

JEL:

• O30
• R11
• R58

ACKNOWLEDGEMENTS

The authors thank the following for precious comments on earlier drafts of this manuscript: Ron Boschma and Philip McCann, Alessandra Faggian, Emanuela Marrocu and the Spatial Dynamics Lab team in UCD Dublin. They are grateful to the editor and three anonymous referees for comments, criticisms and suggestions. The paper greatly benefited from comments and discussion by participants at the Regional Studies Association (RSA) Europe’s Socio-Spatial Dynamics Summer College in Cagliari, Sardinia, Italy, 4–7 September 2018. Precious feedback was also received at the 58th edition of European Regional Science Association (ERSA) 2018; the British and Irish Regional Science Section Annual Meeting 2019, at Association of American Geographers (AAG), Washington, DC, USA, 2019; and the Uddevalla symposium 2019, Sweden.

DISCLOSURE STATEMENT

No potential conflict of interest was reported by the authors.

Notes

1. We decided to employ the CPC because is an extension of the International Patent Classification (IPC) that also includes an additional section Y related to general tagging of new technological developments.

2. The technological classes that appear more frequently together in the regionalized patent documents.

3. Differently from other measure of revealed relatedness we compute the degree of cognitive proximity of patent classes at the regional level not simply based on patent documents. The other measures of technological relatedness are based on co-occurrence matrixes that count how many times two patent classes co-occur on a patent document. Coherence instead is based on a matrix that counts how many times two technological classes occur together in the regional economies. This is the reason why we think that coherence is the most suitable proxy of relatedness for this specific study.

 

Additional information

Funding

Silvia Rocchetta and Dieter F. Kogler acknowledge funding from the European Research Council under the European Union’s Horizon 2020 Research and Innovation programme [grant agreement number 715631, ERC TechEvo] as well as from the Science Foundation Ireland (SFI) under the SFI Science Policy Research Programme [grant agreement number 17/SPR/5324, SciTechSpace]. Raquel Ortega-Argiles acknowledges the fund received from the UK Economic Social Research Council (ESRC) award ‘From Productivity to Prosperity: Inclusive Growth for the West Midlands’ [grant number ES/S002278/1].