Risk preference elicitation without the confounding effect of probability weighting

In this paper we show that the wildly popular Holt and Laury (2002) risk preference elicitation method confounds estimates of the curvature of the utility function, the traditional notion of risk preference, with an estimate of the extent to which an individual weights probabilities non-linearly. We show that a slight modification to their approach can remove the confound while preserving the simplicity of the method which has made it so popular. Data from a laboratory experiment shows that our new method yields significantly different levels of implied risk aversion than the Holt and Laury task even after econometrically controlling for probability weighting in the latter. Implied risk aversion from the traditional Holt and Laury task is relatively insensitive to payout amount, but our new method reveals increasing relative risk aversion and risk neutrality at low payout amounts.