Welfare economics with intransitive revealed preferences: A theory of the endowment effect

Economists use the standard rational model to predict behavior after a policy change and to determine the policy's welfare implications. Recent experimental observations are casting doubt on the predictive accuracy of the standard model, but the more realistic behavioral alternatives often provide a poor basis for making normative evaluations. This paper suggests that we can still predict behavior and measure welfare within the same model. We show that optimizing agents with standard preferences will in some cases behave as if they are subject to an endowment effect. Even so, we may still be able to uncover information about their preferences.