Patents, misallocation and spillovers

This article studies the relationship between patent protection, knowledge diffusion and aggregate productivity. Motivated by the strengthening of U.S. patent enforcement associated with the creation of the Court of Appeals for the Federal Circuit in 1982, we combine patent and firm-level data for 1976–2006 to construct firm-specific exposure to patents granted to technologically related firms. Greater patent exposure is associated with a smaller measured spillover pool. This association remains negative and statistically significant after including industry-year fixed effects and technology-class-year controls, using alternative measures of technological proximity and patent timing, weighting patents by citations and excluding patent-thicket industries. Dynamic specifications show that future patent exposure does not predict current spillovers. We use these micro estimates to discipline a general equilibrium model with heterogeneous firms, endogenous R&D and knowledge spillovers. The estimated decline in spillovers maps into a reduction in the spillover elasticity from 0.45 to approximately 0.10, lowering aggregate TFP by 8.3%; eliminating spillovers lowers TFP by 10.8%. Weaker diffusion also reduces participation in R&D and concentrates innovation among larger firms. The results imply that patent regimes that hinder knowledge diffusion may entail substantial aggregate productivity costs by constraining follow-on innovation and narrowing participation in R&D.