We study the prevalence of the higher order risk attitudes of prudence and temperance in an experiment with a large demographically representative sample of participants. Under expected utility, prudence and temperance are defined by a convex first, and concave second, derivative of the utility function, and have direct implications for saving behaviour and portfolio choice. In the experiment, participants make pairwise choices that distinguish prudent from imprudent, and temperate from intemperate, behaviour. We correlate individuals' risk aversion, prudence, and temperance levels to their demographic profiles and their financial decisions outside the experiment. We observe that the majority of individuals' decisions are consistent with risk aversion, prudence, and temperance. Prudence is positively correlated with saving, as predicted by precautionary saving theory. Temperance is negatively correlated with the riskiness of portfolio choices.

We administer a theory-driven, lab-in-the-field experiment to study the disposition effect among financial professionals. Our novel design identifies, at the individual participant level, key behavioral drivers of the disposition effect: reference-dependent risk attitudes (“tastes”), second-order uncertainty attitudes (including “ambiguity”), and subjective likelihood assessments (“beliefs”). Among the 237 professionals in our sample, 34% exhibited the disposition effect, which seems to be primarily driven by non-Bayesian beliefs. Our experimental results suggest that, when faced with new information about their asset’s performance, financial professionals failed to update their beliefs sufficiently leading them to sell the asset that gained (lost) value more (less) readily.

Proper scoring rules provide convenient and highly efficient tools for incentive-compatible elicitations of subjective beliefs. As traditionally used, however, they are valid only under expected value maximization. This paper shows how they can be generalized to modern (“non-expected utility”) theories of risk and ambiguity, yielding mutual benefits: users of scoring rules can benefit from the empirical realism of non-expected utility, and analysts of ambiguity attitudes can benefit from efficient measurements using proper scoring rules. An experiment demonstrates the feasibility of our generalization.