Collaborative Research: Experiments on Procedures and Prediction in Economic Decision Making

California Institute Of Technology, Pasadena CA

Investigators

Abstract

This project contributes to research in behavioral economics. We know that standard economic models of individual behavior often do not accurately describe how people behave in specific situations. Another way to say this is that these models are not descriptive. However, these models are built on assumptions that people may think include valuable principles that should guide behavior. If this is the case, then these standard canonical models would instruct people on how they should behave. Such a model would be prescriptive, because it prescribes the behavior that people should choose to follow the principle. This project examines whether or not standard economic models are prescriptive in laboratory decision making experiments. The results of this research will further our understanding of whether and how economic models of individual behavior can guide policy interventions. This project will develop two novel experimental methods to assess the normative (prescriptive) nature of expected utility, discounted expected utility, and subjective expected utility. The first methodology will treat axioms as procedures that make a choice on behalf of the individual. Indi- viduals will see axioms represented in a visually simple manner and will report whether they want their choices to be consistent with the axiom. Then, individuals will make choices from decision problems related to the axiom, and will be given the opportunity to reconcile any discrepancies between their preference over axioms and choices. This methodology will test whether individuals treat the axioms as normative in decision-making, even in situations where they originally violated the axiom. We will employ this methodology in the domains of temporal choice and choice under risk and ambiguity. The second methodology will directly elicit whether individuals prefer recommendations that are predicted by a model. This will reveal whether the model itself is viewed as normative by evaluating predictions, without needing to explain the model to experimental participants. We will employ this methodology to study the normative appeal of expected utility theory. Together, this research agenda will show whether individuals want their choices to be consistent with canonical economic models. If they do, then the frequently observed violations of these models might be better thought of as mistakes rather than reflecting underlying preferences. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.