Problem definition: We collaborate with an industrial partner whose supply chain uses multiple tiers, locations, and shipping speeds to efficiently serve customers. In practice, our partner also faces the possibility of upstream disruptions, which limit inventory availability. We model these key features of our partner’s network as a multiechelon distribution system (central warehouse and retailers) with expediting and disruptions. Methodology/results: We prove a novel stochastic program lower bound on optimal cost in this model and use this program to develop a heuristic base-stock policy. Our analysis demonstrates that there is a pronounced benefit from centralized inventory (i.e., holding inventory at the central warehouse) in distribution systems with expediting and disruptions as it can be used to both clear backlogs through expediting and hedge against future disruptions. Further, in the disrupted mode, we provide a simple criterion to determine when decentralization (i.e., holding inventory at the retailers) is preferred over complete centralization. Then, we validate our policies using data from our partner’s nationwide distribution network in the United States. Managerial implications: We provide novel inventory policies for managing a distribution system with expediting and disruptions that are understandable and implementable in practice. Our analysis provides the insight that facilitating the right level of central warehouse inventory is a critical hedge for improving performance in these systems. Finally, our industrial partner’s data suggest that our policies can provide significant cost savings in practice. Supplemental Material: The online appendix is available at https://doi.org/10.1287/msom.2023.0249 .
Zhao et al. (Thu,) studied this question.