Evaluating alternative Frequentist inferential approaches for optimal order quantities in the newsvendor model under Exponential demand

Abstract

Three estimation policies for the optimal order quantity of the classical newsvendor model when the demand is Exponentially distributed are evaluated in this paper. The evaluation is performed analytically for different combinations of sample sizes and values of the requested critical fractile. The statistical measures that have been chosen to perform the evaluation are (a) the actual critical fractile, namely, the actual probability for the estimated order quantities to meet the demand of the period, (b) the mean square error of the estimators for the optimal order quantity, and (c) the range of deviations of estimated order quantities from the optimal order quantity, provided that the probability of taking such a range is the same for the three estimation policies. For small and moderate sample sizes, no estimation policy predominates over the other two approaches, and the choice should be made on a subjective base according to the individual preferences of researchers or practitioners.