Many writers subscribe to Joseph Schumpeter's view that, while perfectly competitive firms allocate resources efficiently in a static sense, they perform poorly when it comes to innovation. From this point of view, the optimal form of market structure is unlikely to be perfect competition, but some other type of dynamic competition which includes significant elements of monopoly. Recently, considerable effort has been focused on modelling Schumpeter's notion of competition. Perhaps best exemplified by the 1980 work of Partha Dasgupta and Joseph Stiglitz (hereafter D-S),2 this approach views free entry to the RD see Nelson and Winter (1977). 3Readers will recognize a similarity of this approach with the notion of contestable markets discussed most recently by William Baumol (1982) and by Baumnol, John Panzar, and Robert Willig (1982). For an interesting comparison of the Schumpeterian with the Marxian notion of competition, see John Elliott (1980). 4This was noted, for example, by Robert Wilson (1975). 5In such industries, the long-term gains from dynamically efficient innovation become of paramount importance; consequently, the optimal market structure would consist of a small number of firms. The driving force behind such a result is what Scherer (1972) has called the Lebensraum effect. Firms performing R&D must at least break even. They derive their profits from

This paper examines the use of compulsory lincensing as a policy to combat the monopoly problem associated with the patent system. It introduces the notion of an optimal patent--one where the patent life and the licensing royalty rate are both determined optimally. Under certain simplifying assumptions it is shown that the optimal patent will have an indefinite life, for both process and product innovations. Some preliminary calculations suggest that the use of compulsory licensing may lead to substantial welfare improvements, even if the patent life is left unchanged at 17 years.

This paper examines the use of compulsory lincensing as a policy to combat the monopoly problem associated with the patent system. It introduces the notion of an optimal patent--one where the patent life and the licensing royalty rate are both determined optimally. Under certain simplifying assumptions it is shown that the optimal patent will have an indefinite life, for both process and product innovations. Some preliminary calculations suggest that the use of compulsory licensing may lead to substantial welfare improvements, even if the patent life is left unchanged at 17 years.