The theoretical literature on managing R&D pipeline is largely based on real option theory making decisions about undertaking, continuing, or terminating projects. The theory typically assumes that each project or causally related set of projects is independent. Yet, casual observation suggests that firms expend much effort on "managing" and "balancing" its R&D pipeline, where managing appears to be related to the choice of R&D selection thresholds, project risk and whether to buy or sell

more »

... ts to fill the pipeline. Not only do these policies appear to differ across firms, they also appear to vary over time for the same firm. Changes in management policies suggest that the choice of R&D selection thresholds is a time-varying strategic decision and there may be some type of vertical interdependency among R&D projects in different stages. In this paper we develop a model using dynamic programming techniques that explains why firms vary in their R&D project management policies. The novelty and value of our model derives from the central insight that some firms invest in downstream cospecialized activities that would incur substantial adjustment costs if R&D efforts are unsuccessful while other firms have no such investment. If transaction costs in technology markets are positive, which implies that accessing the market for projects is costly, these investments lead to state-contingent project selection rules that create a dynamic and vertical interdependency among R&D activities and product mix. We describe how choices of R&D selection thresholds, preferences over project risk, and use of technology markets for the buying and selling of projects differ by the state of the firm's pipeline, the magnitude of transaction costs in the adjustment market, and the magnitude of technology costs. These results yield interesting managerial implications and public policy.