b2647998bc953781490049fe2ac28bf0-Supplemental-Conference.pdf

The decision maker is risk-neutral, therefore we can chooseu(v) = v; note that utility functions canbemodified bylinear transformations without changing thecertain equivalent. The required result holds if the certain equivalent of the situation which includes the augmented reward, denotedCE, is alsoCE . The result also holds forVoPH after recognizing that the optimal policy does not change here, because the uncertaintyX is not downstreamofadecision. This result holds for all utility functions. It follows from recognizing that the augmented rewardr()that yields0for all states ofX lies within the set of balanced rewards over which the optimization is conducted.