Two Echelon Perishable Inventory Management


Speaker


Abstract

We study a two-echelon perishable inventory system that supplies a two-period lifetime product to satisfy stochastic, price-sensitive demand. Inventory is issued using first-in, first-out discipline and unsatisfied demand is lost.  Both price and replenishment quantities at each echelon are endogenously determined by the model.  We formulate this two echelon problem as an inter-related set of Markov Decision Processes.  We characterize the optimal policies and show that the system can be solved iteratively and separately for each echelon, with the solutions converging to a steady, non-randomized system-wide solution. We derive several insights into fundamental properties of the model and provide exact analysis of system behavior through an extensive numerical study.  We also address cases of asymmetric information between echelons, assess the value of information, and compare supply chain performance to the coordinated, centralized case.