Are complex patents more likely to be collaborated and how does the geography of this collaboration look like? In this paper I explore 140 years of US patent data and find that complexity is positively associated with collaboration. Moreover, if a patent is complex, it is more likely to be collaborated with inventors locally. Geography matters!
Research suggests that increasing collaboration in knowledge production is explained by rising complexity of knowledge. Yet, there is little long-run, systematic, empirical evidence on the relationship between complexity and collaboration. A new database is introduced that identifies all (co-)inventors on more than 3 million US patents between 1836 and 1975. Empirical analysis reveals (i) collaboration on US patents began to increase in the 1940s; (ii) there is a robust positive relationship between complexity and collaboration; and (iii) increasing complexity is associated with local rather than nonlocal collaboration.