Website CV Paper
Fields
Theory
Job market paper
Risk Aversion Reversals
Standard stochastic choice models used to estimate risk preferences from binary choices can lead to risk-aversion reversals, where a more risk-averse individual chooses a riskier lottery more frequently than a less risk-averse individual. We focus on weak utility, a flexible class of stochastic choice models consisting of a utility specification and a noise specification that is allowed to depend on the menu. We find that whether or not reversals arise depends critically on the utility specification, and that standard specifications, such as CARA and CRRA, lead to reversals however the noise is modeled for each individual. Moreover, under standard noise specifications, reversals arise for empirically relevant coefficients and lotteries, raising concerns about resulting estimates and out-of-sample predictions. We characterize which utility specifications can avoid reversals with a simple curvature condition satisfied by equicautious HARA, sum-ex, and sum-power specifications. Finally, we develop two notions of reversals for multinomial choice and show that the same curvature condition characterizes when reversals can be avoided.