A toolbox for smart specialization: relatedness, complexity and KETs based on technology flow matrices

ABSTRACT

This paper proposes a toolbox for smart specialisation strategy (S3) that can provide region-specific suggestions about the prioritisation of technology domains based on technology flow matrices estimated with patent citation data. Upstream–downstream relations across technologies are found using citations as proxies, and the key enabling technologies and knowledge complexity are defined accordingly, integrating three principles of S3 that were formerly analyzed separately in the literature. The average propagation lengths model in production theory, which captures long-term and indirect linkages, is then utilised to improve the measure of relatedness across technologies in the current literature. The S3 domains of Lombardy, Italy, are re-examined, and suggestions are made regarding previously ‘undiscovered gems’.

KEYWORDS:

• Smart specialisation strategy
• innovation network
• relatedness
• complexity

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Three major patent classification systems: International Patent Classification (IPC), United States Patent Classification (USPC) and Cooperative Patent Classification (CPC).

2 Classifications of products or economic activities: Harmonized Commodity Description and Coding Systems (HS), International Standard Industrial Classification (ISIC) and Standard International Trade Classification (SITC).

3 A patent may be assigned multiple IPC codes by the EPO examiner.

4 This number includes non-patent literature and cited patents outside of EPO.

5 Unlike in the calculation of k_mn, non-patent literature and forward citations outside of the EPO are not observable, so only EPO-citing patents are included in the calculation. This means that the absorption coefficient matrix is estimated with less precision.

6 There are three newly added classes that are extremely inactive: B33, G12 and G16. Not many applications occur in the chosen cohort of patents from 1998 to 2007.

7 A similar application of RTA to patents can be found in Montresor and Quatraro (2017) and Vlčková, Kaspříková, and Vlčková (2018). In applications to regions with very few patents, users of the RTA and other indicators developed in this paper should be cautious. These indicators are not good indicators of local strength for the regions without a foundation for any kind of smart specialisation.

Additional information

Funding

This paper was partly supported by the National Natural Science Foundation of China (Grant number 72103099).

Notes on contributors

Haixiao Wu

Haixiao Wu received the Ph.D. degree from the George Washington University, USA. His research interests includegeography of innovation, regional and urban economics. He is currently an Assistant Professor with Nanjing Audit University.