What determines the position of regions in European knowledge networks? A comparative perspective on R&D collaboration, co-patent and co-publication networks

The question of why certain regions are more efficient in creating new knowledge is one of the main research issues in Regional Science. In the recent past, inter-regional knowledge networks - defined as the inter-linkage of collaborative endeavours in knowledge creation across regional boundaries - have been increasingly considered as important element for knowledge generation processes. Such inter-regional knowledge networks may constitute promising channels to tap knowledge that is widely dispersed in geographical space. This paper investigates the position of regions in inter-regional knowledge networks at the European level, taking a comparative perspective on the R&D collaboration network as given by the European Framework Programmes (FPs), the co-patent network and the co-publication network. The aim is to identify how region-specific and spatial factors explain positioning in European knowledge networks. In order to take account of structural properties of each network type, we apply a Social Network Analysis (SNA) perspective, defining a region's network position in terms of its centrality according to inter-regional collaboration intensities. By this, we use eigenvector centrality, placing the region in a central hub position, and betweenness centrality, assigning central regions the ability to control knowledge diffusion in the network. In modelling network centrality we distinguish between regional characteristics that reflect a region's knowledge production capacities (capacity dimension), and the variety in a region's knowledge production activities (relational dimension). Our empirical model specification in form of a mixed effects panel version of the Spatial Durbin Model (SDM) allows accounting for the spatial autocorrelation in explaining network position by differentiating between direct, region-internal, and indirect, region-external impacts arising from our set of explanatory characteristics. The results provide first empirical evidence on regional factors that influence position in inter-regional knowledge networks from a comparative perspective on three different network types, given their distinct network structure in relational as well as spatial terms. The results reveal that internal capacity- and technology-related aspects but also spatial spillover impacts from surrounding regions prove to be particularly important for centrality in the co-patent network. We also find significant - region-internal and region-external - impacts of general economic conditions on a region's centrality in the FP network. However, we cannot observe substantial spillover effects of region-external factors on centrality in the co-publication network. Thus, the distinctive knowledge creation foci in each network seem to find expression in its regional determinants.