Comparing Apples and Oranges: Information Sharing in Supply Chains for Fresh Fruits

Abstract

Dual and multi-sourcing strategies have previously been shown to decrease the risk exposure that retailers have to supply chain disruptions. However, when modeling the inventories of suppliers as uncertain, it is often assumed that retailers have complete knowledge of their suppliers' yield distributions prior to making ordering decisions. We study a reverse information sharing framework whereby a series of retailers initially have no knowledge of their suppliers' yields. By modeling the previously observed inventory realizations of suppliers as a time series, the retailers make predictions about future supply and base current ordering decisions off of these. Using real-world supply and demand data, we simulate ordering decisions for a series of sixteen hypothetical retailers that implement dual and multi-sourcing strategies under such an information sharing model. When retailers have inaccurate perceptions of supply, they order more or less than what they optimally should, leading to extra costs as well as supply chain waste and stockouts. Our simulations show that as retailers gain more information about their suppliers' yields over time, their inventory predictions become increasingly accurate, and their ordering decisions begin to approximate to what they would be if they had complete foresight. However, empirically, we find that the rate of this convergence is slower the more suppliers there are. Our results demonstrate that when studying dual and multi-sourcing, it is critical for information gaps between agents to be considered, as they play a large role in dictating the retailers' ordering decisions.