PhD Seminar: Estimation of R&D Network Formation based on Knowledge Creation Model

Estimation of R&D Network Formation based on Knowledge Creation Model

This paper develops a stochastic model of R&D network formation in which firms choose R&D effort and collaboration partners, taking their technology portfolios as given. Based on the knowledge creation model, this paper captures technological overlap and complementarity as sources of spillovers. Using patent data, I construct the R&D network in the IT sector in the US, and estimate which of the common and unique technologies plays the larger role in collaboration. The estimates indicate that the common technologies contribute more to the formation of collaboration links than the unique technologies. Ignoring this technological structure overestimates the spillover parameter. Counterfactual simulations also show that uniform subsidies to collaboration costs do not improve welfare.