Dynamically consistent alpha-maxmin expected utility

The alpha-maxmin model is a prominent example of preferences under Knightian uncertainty as it allows to distinguish ambiguity and ambiguity attitude. These preferences are dynamically inconsistent for nontrivial versions of alpha. In this paper, we derive a recursive, dynamically consistent version of the alpha-maxmin model. In the continuous-time limit, the resulting dynamic utility function can be represented as a convex mixture between worst and best case, but now at the local, infinitesimal level. We study the properties of the utility function and provide an Arrow-Pratt approximation of the static and dynamic certainty equivalent. We then derive a consumption-based capital asset pricing formula and study the implications for derivative valuation under indifference pricing.