Cash Management and Its Risk Control under No-Stationary Uncertain Demand

Abstract

This paper considers the cash balance problem with non-stationary uncertain demand. Online algorithm with risk-reward is used by decision-makers who employ base stock policy. In the online model, the manager could not describe probability distribution but knew exactly the lower bound and upper bound of future demand. We design online risk strategy in terms of forecasting, and measured the reward of online risk strategy by comparing its performance with that of optimal deterministic online algorithm when forecasting is successful. We also proposed a set of online risk strategy whose risk was restricted to a tolerance level by designing the upper bound and range of forecasting while forecasting isn't successful. The online strategy gains reward if successful, the risk is acceptable if unsuccessful. The decision-makers can design an online strategy to improve performance in the light of his risk tolerance level.