We propose a method to jointly measure the intensity of (second-order) risk aversion, prudence (third-order or downside risk aversion), and temperance (fourth-order or outer risk aversion) in laboratory experiments. Our approach is non-parametric and model-independent, and elicits higher-order risk premia with a multiple price list technique.
The typical behavioral pattern we observe is risk aversion, prudence, and temperance. These traits correlate within subjects. Maybe surprisingly, downside risk compensations are significantly larger than those for second-order and outer risk. We illustrate that commonly used utility functions cannot predict this preference pattern, but cumulative prospect theory can. Since higher-order risk preferences have been almost exclusively studied within expected utility, this result motivates their study in other theories of decision making under risk.