Innovative activities have public good characteristics in the sense that the cost of producing the innovation is high compared to the cost of producing subsequent units. Moreover, knowledge of how to produce subsequent units is widely known once the innovation has occurred and is, therefore, non-rivalrous. The main question of this paper is whether mechanisms can be found which exploit market information to provide appropriate incentives for innovation. The ability of the mechanism designer to exploit such information depends crucially on the ability of the innovator to manipulate market signals. We show that if the innovator cannot manipulate market signals, then the efficient levels of innovation can be implemented without deadweight losses–for example, by using appropriately designed prizes. If the innovator can use bribes, buybacks, or other ways of manipulating market signals, patents are necessary.

This paper presents a model of cumulative innovation where firms are heterogeneous in their research ability. We study the optimal reward policy when the quality of the ideas and their subsequent development effort are private information. The optimal assignment of property rights must counterbalance the incentives of current and future innovators. The resulting mechanism resembles a menu of patents that have infinite duration and fixed scope, where the latter increases in the value of the idea. Finally, we provide a way to implement this patent menu by using a simple buyout scheme: The innovator commits at the outset to a price ceiling at which he will sell his rights to a future inventor. By paying a larger fee initially, a higher price ceiling is obtained. Any subsequent innovator must pay this price and purchase its own buyout fee contract.