Topics in the design of an auction for licenses
I consider a situation wherein a research lab has developed a process innovation and wants to sell licenses to members of an industry. The literature has mostly considered the sale of exclusive licenses. I demonstrate, using both theoretical and experimental methods, that when firms sell differentiated products in the final product market, there is a strong case for selling non-exclusive licenses. I also analyze the effect of product market characteristics on the optimal number of licenses. In the first essay, I consider the auction of licenses to a cost-reducing technology when each firm can bid for only one license. I derive the value of winning a license and how this value varies with the number of licenses auctioned. I then show that, given any fixed number of licenses, several well-known auctions are revenue-equivalent. The auction revenue, however, changes with the number of licenses and the lab can maximize its revenue by optimally choosing the number of licenses. I show that it may be optimal to auction multiple licenses when the level of product differentiation among the firms in the industry is high. Moreover, in the presence of negative externalities, the optimal reserve price can be zero. In the second essay, I employ laboratory methods to test the predictions developed in the first essay. The experimental revenues are consistent with the predicted revenues. However, there are some differences in the distribution of the experimental revenues and the predicted revenues and I propose a possible explanation rooted in a simple bidding heuristic for the difference. In the third essay, I develop a framework to analyze the sale of multiple licenses to use a cost-reducing technology when each firm is allowed to bid for multiple licenses. The payoff of each firm depends on the number and abilities of the licensees and the seller maximizes her revenue by optimally choosing the licensees. The optimal mechanism is determined both when each firm's ability to reduce its cost is publicly observable and when it is not. I also show the role of several product market factors in the determination of the optimal number of licensees.