with Gaël Le Mens (Universitat Pompeu Fabra)

In this paper, we compare learning in two common settings in financial markets. One in which investors can observe the outcome of an investment alternative only if they invest in it, and another in which investors can always observe the outcome of an investment, even if they do not invest. We provide empirical evidence that investors’ beliefs are closer to the objective Bayesian beliefs given the observed information when investors are in a setting where they have access to financial information because of endogenous choice. We propose a mechanism that explains our findings. We show that the endogenous creation of the sample of information triggers different cognitive processes. These alternative processes cause better information processing and are of enough magnitude to help overcome the effect of sampling errors.