Marshallian externalities in innovation

A quality ladder model is used to test for Marshallian externalities in innovation. The model predicts that, in the absence of spillovers, the geographical distribution of research should be the same as that of production. This hypothesis is strongly rejected: innovation in two-digit industries exhibits strong spatial clustering independently of the distribution of employment. We find also, in support of Romer and Lucas, that there are strong spillovers from aggregate innovative activity in a region to the research intensity of individual industries. The location of a sector's R&D activity is determined more by the location of other sectors' innovation than by the location of its own production.