We consider the free-market and socially optimal outcomes in a general oligopoly model with many firms which first engage in R&D and then compete in either output or price. Strategic behavior by firms tends to reduce output, R&D, and welfare and so justifies higher subsidies except when R&D spillovers are low and firms' actions are strategic substitutes. It also reduces the benefits of R&D cooperation. Moreover, policies to encourage cooperation are likely to be redundant (since it is always privately profitable) and simulations suggest that the welfare cost of lax competition policy is high.

We examine the implications for strategic trade policy of different assumptions about precommitment in a two-period Cournot oligopoly game with learning by doing. The inability of firms and governments to precommit to future actions encourages strategic behaviour which justifies an optimal first-period export tax relative to the profit-shifting benchmark of an export subsidy. In the linear case the optimal subsidy is increasing in the rate of learning with government precommitment but decreasing in it without, in apparent contradiction to the infant-industry argument. Extensions to active foreign policy, distortionary taxation and Bertrand competition are also considered.