In the last decades, technologies became more complex which increased the degree of uncertainty in R&D. To overcome the uncertainty, firms frequently engage in R&D collaborations, e.g., Research Joint Ventures (RJVs), and licensing agreements. While RJVs are well explored in the literature, very little is known about the use of licensing agreements. Building on a novel database which includes licensing agreements in the semiconductor industry between 1989 and 1999, we estimate a bivariate probit model with self-selection. Our estimation results highlight the fact that firms' relatedness in technology and product markets determine their choices to engage in ex ante and ex post licensing agreements. Our study provides evidence that the decision to engage in ex ante and ex post licensing is formulated with regard to uncertainty in R&D. While ex ante licensing agreements serve as a useful mechanism to insure against uncertainty in R&D, ex post licensing agreements rather serve as an instrument to resolve uncertainty in R&D. Hence, licensing agreements are appropriate instruments to lessen R&D uncertainty.
innovation licensing research joint ventures R&D cooperations R&D uncertainty semiconductors