Designing Auctions in R&D: Optimal Licensing of an Innovation ∗

We study an R&D game in which a research unit undertakes a (non-observable) research effort and, if an innovation is obtained, auctions licenses of the invention to a pool of pro-ducers. We assume that each producer has a private valuation for the license and suffers a negative externality when a competitor becomes a licensee. We compare the optimal rule for the allocation of licenses and the level of research effort implemented by the innovator in two scenarios: decentralized (or free) licensing by the innovator vs. optimal regulation. We show that, as long as the cost of public intervention is sufficiently low, free licensing induces two different types of inefficiencies: an excessively high price for licenses that results in a suboptimal dissemination of knowledge, and an excessively low research effort. This result indicates that optimal public intervention should combine the following measures: (i) an an-titrust agency which limits the royalties that innovators can ask for a license, and (ii) a direct subsidy to research activity in order to compensate innovators for their loss in the revenue raised with licensing