On the ''Bayesability'' of chance-constrained programming problems (Q1073719)

It was hoped that the aversions to risks of goal-attainment failures modeled by chance constraints could be equivalently embodied in some von Neumann-Morgenstern utility function, the expectation of which could be maximized in the standard Bayesian fashion to yield solution sets at all chance-constraint levels identical to those of the original chance- constrained program (CCP). But I shall show that a finite-act E-model CCP cannot have any such Bayesian equivalent if any goal modeled by a chance constraint can fail to be attained by a strategy which optimizes the chance-unconstrained objective function.