Collaborative Bullwhip Effect-Oriented Bi-Objective Optimization for Inference-Based Weighted Moving Average Forecasting in Decentralized Supply Chain

Abstract

Downstream demand inference (DDI) emerged in the supply chain theory, allowing an upstream actor to infer the demand occurring at his formal downstream actor without need of information sharing. Literature showed that simultaneously minimizing the average inventory level and the bullwhip effect isn't possible. In this paper, the authors show that demand inference is not only possible between direct supply chain links, but also at any downstream level. The authors propose a bi-objective approach to reduce both performance indicators by adopting the genetic algorithm. Simulation results show that bullwhip effect can be reduced highly if specific configurations are selected from the Pareto frontier. Numerical results show that demand's time-series structure, lead-times, holding and shortage costs, don't affect the behaviour of the bullwhip effect indicator. Moreover, the sensitivity analysis show that the optimization approach is robust when faced to varied initializations. Finally, the authors conclude the paper with managerial implications in multi-level supply chains.