| Abstract: | Innovators may have incentives to incur the expense of the search for new knowledge even if others can use the results without contributing to the costs of the search for them. Why? Assume firms choose their actions independently. Assume new knowledge is a free public good: a firm can make the product at lower cost even if it spends nothing on research, if another does the research. The most plausible noncooperative equilibrium in this situation is stochastic: firms will decide what to do on the basis of a random device. The theory determines the probability that the firm spends nothing, or something, on research. In consequence, no firm can be certain that a rival will do the research; no firm can be certain of becoming a free rider. In consequence, there will be no relation on average between returns on research-outlays and size of outlays; but the yield on research always will be positive. Innovation can lead to differences among firms, but these differences should not persist, as firms can gain by eliminating them. In a non-cooperative equilibrium, where firms have different cost conditions, the allocation of output among them is inefficient. Such inefficiency can be eliminated if information about methods of production is sold, say by way of licensing. But royalty receipts are a very small fraction of total revenues, even in research and development intensive industries such as chemicals and scientific instruments. Merger is another avenue of cooperation. The theory which assumes research results are private predicts a positive relation between merger activity and technical change. Empirical evidence for the 1879–1930 inverval supports this prediction. |
