In the recent past, regional, national and supranational Science, Innovation and Technology (STI) policies have emphasized supporting interactions and networks between organisations of the innovation system. The policy instrument of the EU in this context are the European Framework Programmes (FPs) that support pre-competitive R&D projects, creating a pan-European network of actors performing joint R&D. In this study, we focus on the embeddedness of European regions in this network. By embeddedness we refer to the notion of centrality in the sense of the Social Network Analysis (SNA) literature. In network theory, vertices that have a more prominent and central network position will more likely benefit from network advantages than actors that have a more distant, peripheral position in the network. A higher network embeddedness of a region, i.e. of organisations located in that region, may increase information and knowledge access in the network, and, thus, create a competitive advantage when it comes to the formation of new collaborations and alliances. The objective of the study is to explain why some regions are able to obtain a better network embeddedness in the European network of R&D cooperation than other regions. For this reason we aim to identify determinants that influence a region's embededdness, involving region-internal factors, such as regional characteristics on their innovation capability, their economic structure and technological specialisation, as well as region-external factors considering the influence of these variables in the neighbourhood of a specific region, referred to as spatial spillovers. To address this question we employ spatial panel modelling techniques, explicitly taking into account the time dimension in our data and the influence of spillovers by specifying a panel spatial durbin error model (SDEM). The dependent variable is the regions' centrality in the FP network for the years 1998-2006, using a sample of 241 NUTS-regions of the EU-25 member states. We aggregate individual FP cooperations to the regional level leading to a network where the nodes are represented by regions and the edges by cross-region collaboration intensities. Using these matrices we calculate a region's centrality relying on two different centrality concepts, namely betweeness- and eigenvector centrality. The independent variables involve regional characteristics related to a region's knowledge production capacity and a region's general economic structure. The results will significantly enrich our understanding of the relationship between a regions network embededdness and its internal and external characteristics.
