Demand Tests for Expected Utility in Incomplete Markets

Abstract

Asset demand tests for Expected Utility have almost universally been implemented in contingent claim settings where markets are complete. However when markets are incomplete, these tests cannot be applied since contingent claim prices cannot be uniquely recovered from given asset prices and the asset demand problem cannot be reduced to an equivalent contingent claim problem. Our key innovation is to show that by allowing not just asset prices but also probabilities to vary, it is possible to construct contingent claim prices that support contingent claim demand resulting from the maximization of Expected Utility. Complete market functional form and local demand tests can then be extended to incomplete markets.