Club convergence in innovation activity — Evidence from the US

This paper uses a non-linear dynamic factor model to investigate the innovation convergence dynamics across US states from 1992 to 2020. Our results indicate that multiple innovation clubs exist within the US, and the heterogeneity between those clusters is increasing over time. Moreover, we observe a geographical pattern in the club membership, with several states from the Western US (especially on the Pacific Coast) and the Northeastern US (particularly New England) forming part of the top cluster. While education and industrial structure are important determinants of innovation clubs, corruption emerges as the most decisive factor in determining whether a US state is on a high or low-innovation trajectory.