Does Geography Matter for Science-Based Firms? Epistemic Communities and the Geography of Research and Patenting in Biotechnology

The localization of knowledge flows has been attributed to agglomerations of knowledge-intensive firms in a region, but I argue that knowledge localization is not consistent with the wider social and professional networks of the firms' employees. I predict that small, knowledge-intensive firms participate in both distant and clustered collaborations, and that these are differentially related to project value. Specifically, I hypothesize a "nested network" structure of collaboration, in which the geography of collaboration reflects different underlying processes by which new technologies and scientific knowledge are generated by firms. Ties in the home region are predicted to provide a platform for technological innovation and exploitation of firm knowledge ("D") while ties outside the region are more likely to be associated with participation in wider scientific communities and acquisition of new knowledge ("R"). The dataset consists of over 5000 publications in the scientific literature by 119 biotechnology firms. The papers involve collaboration between firm scientists and scientists working in universities and other research organizations. The data show that even though the firms are small and highly agglomerated in a few regions, their research collaboration networks are far-flung, reflecting the global distribution of scientific knowledge. The data further reveal a bimodal pattern in the geography of collaboration, with about 20 per cent of papers involving co-located authors (within 50 miles of eachother) and a larger group involving coauthors over 1000 miles apart. To test whether this bimodal pattern in the geography of collaboration reflects the operation of "nested networks" described above, I model two forward citation streams to the papers: technological innovation is measured as patent-to-paper citations, and research impact is measured as paper-to-paper citations. The models confirm that collaborations in which coauthors are located within 50 miles of one another are more likely to be cited in subsequent patents by the firm relative to papers in which coauthors are located at greater distances; however, prestige in the scientific community is not associated with subsequent patenting. In contrast, consistent with literature in the sociology of science, the scientific prestige of coauthors has a significant impact on citations in the literature, but geographic clustering of coauthors is negatively associated with high-impact articles. The contrast between the models of the two citation streams points to different processes of scientific research and technology development that is reflected in the geography of collaboration. The paper adds to the literature establishing that proximity is associated with technological innovation, but also shows that small firms are simultaneously linked to wider professional communities outside their home region, and that these distant ties create opportunities for new knowledge acquisition