Risk averse demand selection with all-or-nothing orders

Consider a firm that operates in consecutive single selling seasons, delivering its products across several markets with unique revenue and uncertain demands in each market. Using a profit maximization approach based on a newsvendor-type model, the firm may still incur several losses across consecutive periods in the short run. Risk analysis with demand selection has been modeled where customer/market demands follow a normal distribution. Often a firm faces a set of potential unconfirmed orders, where each order will either come in at a predefined level or it will not come in at all. In this paper, we consider these all-or-nothing (AON) demands and provide insights into their effect on expected profit and the frequency of extremely costly procurement policies. Instead of solely identifying the market/demand set and procurement quantity that maximizes expected profit, we use a conditional value-at-risk approach that allows a decision maker to control the number of profitable but risky demands to consider in the overall procurement policy. This approach is compared against an expected profit objective, and several managerial insights are provided.