Reconsidering Risk Aversion

Risk aversion is typically inferred from real or hypothetical choices over risky lotteries, but such “untutored” choices may reflect mistakes rather than preferences. We develop a procedure to obtain a better measure of normatively relevant preferences: after eliciting untutored choices, we confront participants with their choices that are inconsistent with intertemporal-expected-utility axioms and allow them to reconsider their choices. We demonstrate this procedure via a survey about hypothetical retirement investment choices administered to 596 Cornell students. We find that, on average, reconsidered choices are more consistent with almost all axioms, with one exception related to a counterfactual reference point