Empirical Rationality in the Stock Market

This paper approximation errors are introduced in a Luca (1978)-type model to reflect model uncertainty. The purpose is twofold. First, the rational investor is allowed to take model uncertainty into account when asset prices are determined. Second, the statistical degeneracy, common to most structural models, is broken and maximum likehood inference made possible. The model is estimated using U.S. stock data. The equilibrium price is seriously affected by the existence of approximation errors and the descriptive and normative properties are greatly improved. This suggest that investors do not and should not ignore approximation errors.