On the Onassis problem

Abstract

A decision problem is by convention characterized by its outcome matrix and by a subsequent utility evaluation. In trying to set up an outcome matrix based on wealth values it may occur that due to ambiguities inherent in the decision problem it is not clear which standard of value (or numéraire) should be used in order to measure wealth. A typical example of this kind is Stützel's so called Onassis Paradox. We show that problems of this kind can be solved within the conventional framework of decision theory. The analysis proceeds in two steps. First, state-dependent utility functions are derived; second, a model for evaluating these utility functions is presented.