Information Aggregation, Strategic Behaviour and Efficiency in Cournot Markets

When is the modeller introducing more error when analysing a Cournot market with private cost information - when ignoring market power or when ignoring the impact of incomplete information? Is the welfare loss at the market outcome driven by private information or by market power? The answer, both to the positive and to the normative questions, is that in large enough markets abstracting from market power provides a much better approximation than abstracting from private information. More precisely, in a replica market with n firms facing increasing marginal costs of production subject to independent shocks, while the effect of market power decays quickly with n (it is of the order of 1/n for prices and 1/n2 for (per capita) expected total surplus (ETS)), the effect of private information decays more slowly with n (it is of the order of 1/ for prices and 1/n for ETS). Increasing n is more effective in reducing the welfare loss due to market power than the one due to private information. Simulations show that the result holds for moderately sized markets (small n) whenever uncertainty is significant. In this case information policy is more relevant than classical competition policy.