Network-driven transfers, localized spillovers, and the economics of innovation

The dissertation research concerns the detection and empirical analysis of flows of knowledge capital in patent citations data, and the examination of determinants and mechanisms associated with knowledge spillovers. I define and construct network-driven transfers data which acknowledge separate components of knowledge flow. The term is chosen to suggest the important productivity-enhancing aspect to inventor relationships, and the analyses are a window to the social context of innovative activity. I design the methodology and datasets to be used widely, to complement the typically undifferentiated patent citations increasingly popular in studies of innovation. As a tool for observing knowledge flows, network-driven transfers are associated with odds of four to one to observe actual transfers, an improvement by a factor of 3.0 to 7.5 over full citations data. Citations to newer patents are also more likely to be associated with inventors, and thus actual flows. The final chapter describes a number of operational and methodological contributions to the search for geographically localized knowledge spillovers. The unique focus is on the role of individual inventors, and to examine, at this fundamental level of innovative activity, the impact of geographic distance, time, and political boundaries on transfers of knowledge. The analyses exploit network-driven transfers---citations for which transmission mechanisms emphasize repeat contact with prior inventors. The analyses reveal systematic patterns to the attenuation of localization across the aforementioned dimensions. The main finding is that network-driven transfers are correlated with increasing proximity effects over time. This suggests that inventor proximity becomes more important for productivity, despite communications advances which, according to conventional thinking, enhance collaboration within inventor networks.