How should managers take into account the propagation of supply chain disruptions and risks (i.e. the ripple effect) when they design their inventory policies? For over 60 years, various extensions and applications to the popular newsvendor model have been suggested, where cost/profit are often the focal objective. We propose a new version of the traditional single-period newsvendor model - the "Rippled Newsvendor" - with supply chain severity (i.e. risk propagation) as the primary objective while taking into account network structure. Our model considers exogenous and endogenous risk(s) of disruption while exploring the tension between under-supply and "wear-and-tear" (i.e system breakdown). To model the intricacies of this trade-off whilst minimizing the potential spread of risk, we leverage a Bayesian Network whereby the conditional probability distributions are functions of the inventory ordering decisions. We use a simulation study to understand the nature of our objective function as well as to gain insight into the potential optimal ordering policies of this new model. Furthermore, the simulation seeks to understand how the various factors in our system impact total risk severity, and if they do so in different ways. Our simulations indicate that local exogenous risk is of greater importance than non-local exogenous risk. Furthermore, we show that the type of risk, as well as the structural characteristics of the supply chain and inventory system, impact risk severity differently.
Keywords: Bayesian networks; Inventory management; Newsvendor; Ripple effect; Supply chain risk propagation; Supply chain risk structure.
© 2020 Elsevier B.V. All rights reserved.