Sammendrag
As interorganisational collaboration is increasingly seen as a crucial nexus of innovation by scholars and policymakers alike, it is important to investigate whether and in what way research collaboration networks affect innovation outcome. Answering the research question “How does a firm’s position in a research network affect its innovation outcome?” this thesis employs a novel combination of data sources to test the innovation effects of three variables related to firm network position: Node strength, reach, and ego-network redundancy. Using a network constructed from a database of publicly funded research collaboration projects from the Research Council of Norway (RCN), the effect of network variables are tested on variables from firm-level data from the 2014 Norwegian Community Innovation Survey that indicate process and product innovation, as well as new-to-firm and new-to-market innovation. Using the unique RCN dataset, the thesis implements a few novel methodological approaches, such as delineating sectoral network by tags denoting knowledge areas, and weighting the network ties according to project budget. The findings indicate that while a firm’s research network position has significant effects on product innovation, it does not significantly affect process innovation. Furthermore, while both node strength and reach are positively associated with innovation outcome, the effect of the former is heavily dependent upon the firm’s R&D-intensity. These results are explained in light of both the network and modes of innovation literature, which both emphasise the different nature of the knowledge being created and exchanged in different innovation processes. Meanwhile, the effects of ego-network redundancy on innovation outcome are inconsistent, as a significant positive effect is found for innovation in general, but no effect can be found for new-to-market innovation. The ambiguous results are used as the basis for an exploratory reflection on how the effects of knowledge creation and diffusion may be inhibited by imitative behaviour in dense ego-networks.