Everyday econometricians : selection neglect and overoptimism when learning from others

There are many important decision problems where learning through experimentation is costly or impossible. In these situations, individuals may try to learn from observing the outcomes of others who have made similar decisions. Often, however, information about others comprises a selected dataset, as outcomes are observed conditional on specific choices having been made. In this paper, we design an investment game which allows us to study the influence of selection when learning from others. Using the theoretical study of selection neglect in Jehiel (2018) as a guide, we test (i) for the presence of selection neglect in this investment context, and (ii) some comparative static predictions of the model. We find strong evidence for selection neglect- even though subjects are fully informed about the data generating process. As theoretically predicted, the degree of bias due to selection neglect increases when other decision makers become more informed, or become more rational. It decreases when signals are correlated.