Estimating uncertainty aversion using the source method in stylized tasks with varying degrees of uncertainty

Abstract

The source method provides a tractable way to analyze uncertainty aversion using source functions. In this analytical framework, uncertainty aversion is defined using the index of pessimism and the index of likelihood insensitivity. The focus of this paper is to use the source method to examine how uncertainty aversion differs across events that have the same underlying objective probabilities but are presented under varying degrees of uncertainty. In a controlled laboratory setting, subjects are presented with three lottery tasks that are ranked in order of increasing uncertainty. Given the choices observed in each task, a “source function” is estimated jointly with risk attitudes assuming Rank-Dependent Utility Theory under different probability weighting specifications for the source function. Results from the Prelec (Prelec (1998)) specification suggest that, as the degree of uncertainty increases, subjects display increasing pessimism; in contrast, the Tversky-Kahneman (Tversky-Kahneman (1992)) and the Power specifications detect no such difference. Thus, conclusions regarding uncertainty aversion are contingent upon which probability weighting specification is assumed for the source function, and caution should be taken when selecting specifications for the source function.