Appropriability and the Timing of Innovation: Evidence from Mit Inventions

Abstract

At least since Arrow (1962), economists have believed that strong property rights are necessary for firms to invest in innovation. This belief was a key principle underlying the Bayh-Dole Act, which gave universities the right to own and license federally funded inventions, because the commercialization of university inventions requires private firm investment in development, given the early stage of these inventions at the time that they are licensed. However, surprisingly little research has examined this key principle. In this paper, we exploit a database of 805 attempts by private firms to commercialize inventions licensed exclusively from MIT between 1980 and 1996 to address this issue. The data allow us to examine the timing of subsequent commercialization or termination of the licenses to these inventions as a function of the length of patent protection, as well as other measures of appropriability. We model the firm's investment decision as an optimal stopping problem, and we characterize the hazard rates of first sale and termination over time. In both the theory and the empirical analysis, we find two opposing effects of time. The length of patent protection provides an incentive for the firm to invest that declines with time; while the probability of technical success increases in each period that the firm invests. Competing risks models to predict the resulting hazards of first sale and termination reveal that, for these data, the hazard of first sale has an inverted u-shape and the hazard of termination has a u-shape. We find that increased appropriability, as measured by Lerner's index of patent scope and effectiveness of patents in a line of business, decrease the hazard of termination and increase the hazard of first sale.

Keywords: Appropriability, Patents, Innovation, Discrete Hazard Models

JEL Classification: L00, D21, D23, O31, O32, O34