Inventory Control When Selling to Human Newsvendors

Abstract

We develop inventory control strategies for a supplier who is selling a product to a group of human newsvendors. Recent research in newsvendor experiments using human subjects revealed significant behavioral tendencies in the decision making processes of human newsvendors. If the supplier would incorporate these behavioral tendencies while managing her inventories, her costs would be significantly reduced. Using data from our own experiments, we estimate the possible reductions in supplier costs and determine the factors that significantly impact it. We observe a significant improvement in the supplier’s inventory decision if she estimates the demand distribution with a model that captures the anchoring tendencies instead of assuming aggregate randomness. As the model selection relies much on the data, it is an important task to determine whether the retailers are mean-anchoring, demand-chasing, both, or neither. In addition, we observe that the information about retailers’ behaviors is more beneficial for the supplier when the end-customer demands are more variable. It is certain that more precise information on the data will lead to a better inventory decision for suppliers. The information includes the retailers’ behavioral tendencies, individual-specific order behaviors, and the variability of end-customer demands.