This paper uses elementary nonparametric regression theory to integrate economic and sociological thinking about expectations and choice. The main result is a Social Learning Proposition giving conditions implying that a person who observes the outcomes realized by a set of 'role models' can use these data to estimate nonparametrically the expected utility of an action. The Social Learning Proposition is first developed for static discrete choice and then extended to finite-horizon dynamic discrete choice problems. The key to the extension is a Representation Lemma showing that any dynamic problem has an observationally indistinguishable static representation.
ASJC Scopus subject areas
- Economics and Econometrics