Ellsberg Paradox and Second-order Preference Theories on Ambiguity: Some New Experimental Evidence

We study the two-color problem by Ellsberg (1961) with the modification that the decision maker draws twice with replacement and a different color wins in each draw. The 50-50 risky urn turns out to have the highest risk conceivable among all prospects including the ambiguous one, while all feasible color distributions are mean-preserving spreads to one another. We show that the well-known second-order sophisticated theories like MEU, CEU, and REU as well as Savage’s first-order theory of SEU share the same predictions in our design, for any first-order risk attitude. Yet, we observe that substantial numbers of subjects violate the theory predictions even in this simple design.