The Geography of Unconventional Innovation

Abstract

This paper studies the role of population density as a driver of innovation. Using a newly assembled dataset of georeferenced patents in the U.S., we show that overall innovation activity is not concentrated in high-density areas as commonly believed. However, when we restrict attention to unconventional innovations - innovations based on unusual combinations of existing knowledge - we show that these are indeed more prevalent in high-density areas. To interpret this relation, we propose the view that informal interactions in densely populated areas help knowledge flows between distant fields, but are less relevant for flows between close fields. We then provide evidence supportive of this view. We build a model of innovation in a spatial economy that endogenously generates the pattern observed in the data: specialized clusters emerge in low-density areas, whereas high-density cities diversify and produce unconventional ideas.