Resources | techdiv

Data

The index and figures presented in this platform have been elaborated from OECD REGPAT database. OECD-REGPAT derives from PATSTAT, and it contains all patent applications to European Patent Office (EPO) and their technological classes (International Patent Classification codes - IPCs) from 1977 to 2011. OECD-REGPAT assigns each patent application to the corresponding NUTS2 region based on the address of the inventor. For a complete description of the method for regional assignation method click here. For a complete description of the database content click here.

Methodology

The Team

Pierre-Alexandre Balland

Ron Boschma

Joan Crespo

David Rigby

Professor at Utrecht University

Researcher at Utrecht University

Professor at University of California Los Angeles

Data manipulation and measures computation has been done by using R free software and its EconGeo package.

Number of patents

Patents are counted per region and technology for each period of time (last period available, 2005-2009).

Multiple technological classes: patents are often assigned to more than one technological class (multiple IPCs). In these cases, we counted one for each technological classes involved. In these cases, the patent count is one for each of the technological classes cited in the patent.
Multiple locations: patents may be assigned to more than one location (multiple NUTS2) when two inventors in the patent live in different regions. In these cases, the patent count for the region is proportional to the share of inventors participating in the patent and located in the region.

For computation issues, regions with less than 50 patents per period are excluded. Similarly, technological classes with less than 30 patents per period are also excluded. These small categories generate important inconsistencies in the computation of complexity.

Relatedness

Relatedness computation relies on co-occurrence of technological classes. Two technological classes are related if they often occur together in the same patent document.

References: Boschma et al., (2015) ; Hidalgo et al., 2007

Complexity

Complexity computation relies on the structure of the region-technology network: region A is linked to technology 1, if region A is specialized on technology 1. Complexity is iteratively computed combining two types of information i) which regions produce specific technologies, and ii) how common specific technologies are across cities.

References: Balland and Rigby (2016) ; Hidalgo and Hausmann (2009)

Software

Acknowledgements

We would like to thanks the European Commission and Regional Studies Association for their financial support.

We also would like to thank Chris Espósito, Wolf-Hendrik Uhlbach and Gianluca Tarasconi for their assistance in data collection and data cleaning.

This platform has been strongly inspired by the Observatory of Economic Complexity.