Non-parametric Elicitation of Risk Preferences

• Development Workshop

Date and Location

Monday, October 27, 2014, 4:10 PM - 5:30 PM
ARE Conference Room, 2102 Social Sciences and Humanities

Abstract

This paper introduces a new framework for analyzing risk preferences that does not rely on Von-Neumann and Morgenstern’s expected utility theorem. In particular, the proposed model does not require the (much debated) axiom of independence. In our setting, individuals choose between two lotteries L and M, depending on a cumulative probability gap: they compare the probability that the outcome W is lower than each possible value x for each lottery, and put a weight on each combination (x ; probability gap). Formally, the individual chooses between lotteries L and M depending on the sign of the cumulative weighted probability gap. It allows us to build a theory of decision under risk where risk aversion is not linked to the concavity of Bernoulli’s utility function. We show that in this framework, it is possible to estimate risk preferences non-parametrically and provide some simulation illustrations.