Determinants of Collaboration in European R&D Networks : Empirical Evidence from a Binary Choice Model Perspective

R&D collaboration networks stimulated by the European Union in its Framework Programs (FP) on Research and Technological Development exhibit - like many other large social networks - characteristic features of complex networks. Existing analyzes of these networks, however, are mostly built on strongly simplifying assumptions for network construction, like fully connected project graphs or star-graphs. Still, apart from qualitative evidence, little is known about the determinants that affect partner choice and network dynamics. It is the aim of this paper to provide empirical evidence for a more differentiated picture of partner selection - and thus network formation - in the European FPs. We adopt an econometric perspective to identify determinants of link formation, including various actor characteristics, relational and network effects as well as geographical effects. We employ a binary choice model estimated by means of logistic regressions, with a dependent variable that represents the establishment of a formal cooperation between two organizations in FP5. We use data on EU FP projects from the sysres EUPRO database and from a representative survey of FP 5 participants. The study produces statistically significant evidence that R&D collaboration choices of organizations participating in European FPs are affected by geography, FP experience and relational factors including network characteristics. Thematic proximity matters more than geographic proximity, while most influential for collaboration appears to be prior acquaintance of the actors. Also, network effects significantly determine the collaboration choice, but to a slightly smaller extent than geographical effects