Robust Decision-making Under Ambiguity

Most of management research, following on the paradigm of expected utility theory, has developed complex models of optimal managerial action in the presence of uncertainty. Still, the assumption that managers assign probabilities to outcomes, and consequently optimize their actions has come under criticism from a number of empirical studies. Researchers, working on probability assessments, have found an interesting empirical regularity: while managers have difficulty coming up with one single number representing a probability, they very often can judge (remarkably) accurately a range for such probabilities. This study examines managerial decision-making under ambiguity by examing how having a family of probability distributions (instead of one unique distribution) influences the nature of “optimal” decisions. In a remakably simple result, we show an equivalence result that demonstrates that the “optimal” decision retains its core structure and that there is a unique distribution under which an identical decision may be expected to be made.