On the consistency of the individual behavior when facing higher-order risk attitudes

We propose and analyze three procedures to elicit participants' behavior regarding their higher -order risk attitudes. Our procedures relax the rigidity of having to choose among only two alternatives (as usual in the literature), and allow for choosing a third or even a fourth alternative. This way, the remaining choices for the two original alternatives carry lower levels of noise, i.e., lower randomness in behavior from the subjects without a strong preference in the binary setting, that are forced to choose among only two alternatives. Although the proposed procedures only filter out a small proportion of the choices, the effect on the consistency of the individual behavior across moments is highly significant.