Using multi-agent systems to explore information sharing in arborescent supply chain networks

The impact of information sharing in supply chains has been traditionally restricted to classical serially-linked echelon structures, an assumption which does not often correspond to reality. In this paper, the potential benefits of information sharing in a divergent supply chain are analyzed. To do so, a Java-based multi-agent simulation is developed to compare the performance of a serial supply chain with that of a divergent supply chain, both under non information sharing and information sharing conditions. Results show that the traditional (i.e. without information exchange) arborescent supply chain structure performs worse than the serial supply chain when the market demand suffers violent and unexpected changes. On the contrary, a significant reduction of the discrepancy between both SCNs can be noted when implementing information sharing strategies.