On the Possibility of Informationally Efficient Markets

In a dynamic asset pricing model informed traders receive a noisy signal of the value of a risky asset while uninformed traders learn to extract the information from the price. The relative popularity of the two strategies depends on past performance. An "intensity of choice" parameter is endogenous, reflecting the traders" confidence in selecting the better of the two strategies. The asymptotic properties of the model depend on the evolutionary process for modeling relative popularity. It also depends on how the treatment of the convergence of the model as the popularity of being informed declines towards zero. It is possible to create prices that are arbitrarily close to perfect efficientMarket Efficiency, Asset Pricing, Learning