Exploring the Benefits of Information Sharing in a Distribution System

In a distribution system, the distribution center (DC) observes order streams from all retailers it serves while each individual retailer may only have the local demand information available. This paper investigates how the DC can exploit this informational advantage over the retailers to improve its own performance as well as the performance of the entire supply chain. Specifically, we study a stylized model with one DC and two retailers. The demand processes of the retailers are subject to potential market interactions and thus are modeled by a vector ARMA process. Given that each individual retailer only observes its own demand process, we show that the DC can recover the joint external demand process from the retailers’ order histories. If the DC shares with the retailers the information derived from the joint demand process, the retailers can always improve their forecasting accuracy and reduce inventory cost. However, the DC’s forecasting accuracy and cost performance may deteriorate because additional information alters the retailers’ order processes and may decrease their predictability. In the special case of zero lead times, we establish a necessary and sufficient condition for the DC itself to benefit from information sharing. In the case of general lead times, we quantify the magnitude of the benefit/loss by a numerical study. We also examine the information security and privacy issues raised in our settings that may prohibit the DC from sharing information and materializing the associated benefits. The existence of privacy preserving information-sharing mechanisms is established, and a protocol is developed for the DC to prove that the improved demand forecasts are indeed credible without releasing one retailer’s demand information to the other. (Key Terms: Information Sharing; Forecasting; Distribution System; Privacy Preserving)