The Decision to Patent, Cumulative Innovation,and Optimal Policy

Optimal patent breadth is an issue that is still being vigorously debated at both the theoretical and empirical levels. This paper analyzes optimal patent policy in the context of cumulative innovation in a model that endogenizes the patenting decisions of early innovators. In the theoretical literature on cumulative innovation, it is generally assumed that all innovations are patented. However, studies such as Cohen et al. (2000) and Levin et al. (1987) report that firms frequently rely on secrecy to protect their discoveries. Cumulative innovation implies that innovators may have significant incentives to keep their innovations secret to get a head start in subsequent R&D races. This paper shows that if innovators cannot rely on secrecy to protect their innovations, it is optimal to have relatively narrow patent protection. This happens if the government has a weak trade secret policy or if innovators cannot monitor the flow of their technological information. This is because when innovators cannot rely on secrecy to protect their innovations, they have increased incentives to patent them and it is not necessary for the government to give them extra incentives to patent. In the case when innovators always prefer secrecy over patenting, it becomes optimal to have a flexible antitrust policy rather than a flexible patent policy. Since non-disclosure reduces the investment incentives in the second R&D race, allowing collusive licensing agreements between competing innovators becomes optimal in order to stimulate investment in the second R&D race.