Finding Cross-Border Collaborative Centres in Biopharma Patent Networks: A Clustering Comparison Approach Based on Adjusted Mutual Information

The recent speedy development of COVID-19 mRNA vaccines has underlined the importance of cross-border patent collaboration. This paper uses the latest edition of the REGPAT database from the OECD and constructs the co-applicant patent networks for the fields of biotechnology and pharmaceuticals. We identify the cross-border collaborative regional centres in these patent networks at NUTS3 level using a clustering comparison approach based on adjusted mutual information (AMI). In particular, we measure and compare the AMI scores of the clustering before and after arbitrarily removing cross-border links of a focal node against the default clustering defined by national borders. The region with the largest difference in AMI scores is identified as the most cross-border collaborative centre, hence the name of our measure, AMI gain. We find that our measure both correlates with and has advantages over the traditional measure betweenness centrality and a simple measure of foreign share.