A Collaborative Decentralized Distribution System with Demand Forecast Updates

In this paper, we study inventory pooling coalitions within a decentralized distribution system consisting of a manufacturer, a warehouse (or an integration center), and n retailers. At the time their orders are placed, the retailers know their demand distribution but do not know the exact value of the demand. After certain production and transportation lead time elapses, the orders arrive at the warehouse. During this time, the retailers can update their demand forecasts. We first focus on cooperation among the retailers - the retailers coordinate their initial orders and can reallocate their orders in the warehouse after they receive more information about their demand and update their demand forecasts. We study two types of cooperation: forecast sharing and joint forecasting. We show that the cooperative games associated with both situation have non-empty cores. However, by using an example we illustrate how forecast sharing collaboration might lead to bad performance, and asymmetric forecasting capabilities of the retailers might harm the cooperation. On the other hand, joint forecasting always results in higher total expected profit. Finally, we analyze the impact that cooperation and non-cooperation of the retailers has on the manufacturer's profit. We focus on coordination of the entire supply chain through a three- parameter buyback contract. We show that our three-parameter contract can coordinate the system if the retailers have symmetric margins. Moreover, under such a contract the manufacturer prefers retailers' cooperation since he can get a share of improved performance