We model today's privatized crop research industry as a small number of firms, developing and selling differentiated products to heterogeneous producers. Crop variety research is modeled as a search process, which allows us to differentiate between applied and basic research and recognize research as a stochastic process. We use the framework to develop a number of propositions regarding private research incentives, the spillovers of knowledge, and the impact of public policy. The results suggest an underinvestment in research even when property rights have been established.