Proceedings of International Conference on Management, Economics and Finance
Decentralized Supply Chain Performance When Demand Information Is Not Shared
Youssef Tliche, Atour Taghipour and Béatrice Canel-Depitre
In decentralized logistic structures, information management policies generally do not allow information sharing among actors of the same supply chain, or in the best cases, allow the sharing of some non-key information. In order to enhance overall decentralized supply chain performance, we investigate a recent strategy called “Downstream Demand Inference” which allows an upstream actor to infer the demand at his formal downstream actor without the need of formal information sharing. Downstream Demand Inference was surveyed through several forecasting methods, and was successfully experimented with only Simple Moving Average method. Thus, in this paper, we investigate Downstream Demand Inference through other method, namely, the Weighted Moving Average forecasting method which affects non-equal weighting to past observations. First, we establish the Mean Squared Error and Average Inventory Level expressions for the upstream actor. Second, we formalize the upstream actor’s Forecast Optimization Problem and propose the application of the Newton’s method to solve it. Finally, based on simulated time-series demands, the results show that our approach allows further optimized solutions, in terms of forecast errors and inventory levels, which can improve the competitiveness of companies in the market.
Keywords: weighted moving average; supply chain management; newton method; downstream demand inference; autoregressive moving average processes.