We develop a typology of regional innovation networks based on the extent to which the inventors in a given location span structural holes within the internal collaboration network or connect local inventors with inventors outside their region. We identify four types of network configurations characterized by different levels of internal brokerage and external boundary-spanning (fortresses, multi-level brokers, playing fields and absorbers). Using data on co-inventing networks within and between United States Metropolitan Statistical Areas between 2000 and 2014, we show that configurations balancing internal brokerage (closure) and low (high) external boundary-spanning lead to higher innovation performance than isomorphic configurations. In a nutshell, our findings convey a more nuanced, multi-level understanding of the trade-offs between different brokerage roles in regional innovation dynamics.
Read full abstract