A Parametric Model of Ambiguity Hedging

The Ellsberg Paradox demonstrates that many people like to hedge ambiguity in a way that contradicts expected utility maximization. We postulate roughly that all violations of the expected utility theory are due to affinity for ambiguity hedging. The associated utility representation is a special case of maxmin expected utility (Gilboa and Schmeidler, 1989), where the set of priors Π has a well-known parametric structure called εcontamination. Both the weight ε ∈ [0, 1] and the contaminated prior p ∈ Π are uniquely derived from preference. The parameter ε is interpretable as a degree of affinity for hedging. A model of complete ignorance is obtained as a special case where the prior p is contaminated by the simplex of all probability measures.