Clustering of R&D collaboration in Cournot oligopoly

This paper complements the Cournot collaboration game outlined in Goyal and Joshi (2003, sect. 4), with the hypothesis that pairwise R&D alliance is constrained by knowledge distance. Potential asymmetry of distance between two knowledge sets is formalized through a quasi-metric in knowldge space. If the knowledge constraints to collaboration are weak enough, the paper replicates the result by Goyal and Joshi (2003, sect. 4), that a firm is either isolated, or is connected to every other non-isolated firm in the industry. If absoprtion of ideas from one’s potential partner requires sufficiently high knowledge proximity, the stable R&D networks in Cournot oligopoly are shown to display the clustering property, that is characteristic of real-world industry networks, and of social networks more generally.