Multicriterion concept of risk under incomplete information

Abstract We present three behavioral hypotheses reflecting decision makers' simultaneous propensity towards high and possibly speculative gains on the one hand and the dislike of extreme expected losses on the other. A multidimensional measure of risk, leading to the Prospect Ranking Vector (PRV), takes into account the possibilities of both the extreme gains and the extreme losses associated with a given venture. The “distribution-free” approach utilizes some well-known probability inequalities in order to derive useful approximations of prospect-ranking vectors under the conditions of partial information. Because the decision makers are neither exclusively risk-averse nor risk-prone but rather exhibit both propensities at situationally dependent and changing levels of mutual conflict, the concepts presented in this paper could form a basis of a less restrictive general theory of risk.