Here’s the abstract:
This paper investigates how physical, organisational, institutional, cognitive, social, and ethnic proximities between inventors shape their collaboration decisions. Using a new panel of UK inventors and a novel identification strategy, this paper systematically explores the net effects of all these ‘proximities’ on co-patenting. The regression analysis allows us to identify the full effects of each proximity, both on choice of collaborator and on the underlying decision to collaborate. The results show that physical proximity is an important influence on collaboration, but is mediated by organisational and ethnic factors. Over time, physical proximity increases in salience. For multiple inventors, geographic proximity is, however, much less important than organisational, social, and ethnic links. For inventors as a whole, proximities are fundamentally complementary, while for multiple inventors they are substitutes.
In other words, we find that physical proximity is critical to break the ice in a research collaboration; once the relationship has been established, however, other forms of proximity become more important. Crucially, for multiple inventors we find that co-location basically disappears as a driver of collaboration.
Obvious, you might think. But I’d argue that the multiple inventor group finding is pretty counter-intuitive. Our results also imply that the gains from incubators and research labs are strong for young researchers, but may fall off quite quickly. And our numbers chime with the ‘nursery cities’ hypothesis – basically, big cities are better for small young firms than older, larger ones.
UPDATE: an improved version of the paper is forthcoming in Research Policy. Read it here.