The Geography of Breakthrough Innovation in the United States over the 20th Century

Over the 20 century, the geography of breakthrough innovation in the United States – defined as the spatial distribution of the production of patents that are both novel and impactful – underwent three broad changes. At the start of the 20 century, breakthrough innovation was concentrated in populous and knowledge-diverse metropolitan areas. By the 1930s, breakthroughs were created less frequently across the entire country and so their invention had a less distinct geography. The substantial creation of breakthroughs resumed in the 1960s and was once their invention was concentrated in large and knowledge metropolitan areas. However, during the latter part of the century the invention of breakthroughs also frequently involved long-distance collaborations between inventors. In this paper, I document these historical changes to the geography of breakthrough innovation and propose a model to explain why they occurred. The model suggests that the geography of breakthroughs is established by four factors: (1) the prevailing knowledge intensity of breakthrough inventions, (2) the distancebased frictions incurred by technologies used for collaboration, (3) the distance-based frictions incurred by the technologies used for knowledge-sourcing, and (4) the disruptiveness of the regime of technological change. I generate support for the model, and conclude the paper by discussing lessons that the 20 century’s geography of breakthrough innovation provide for anticipating possible futures for the geography of innovation in the 21 century, including in the years beyond COVID-19. Acknowledgements: I thank David Rigby, Michael Storper, Allen Scott, Wolf Uhlbach, Anthony Frigon, Carolin Ioramashvili, Ryan Hynes, participants at the SEG seminar series, and participants at the NARSC annual meeting for their thoughtful feedback. In addition, I thank Frank van der Wouden for providing pre-1975 inventor location data.