Structural characteristics of complex supply chain networks

Abstract

There is a scarcity of analyses of topological properties of real world supply chain networks (SCNs), mainly due to lack of data. Most theoretical research has used the popular Barabasi-Albert (BA) growth model, in which new nodes (firms) entering the network attach preferentially to the existing nodes with more links (contractual relationships), to generate network topologies supposedly representative of real world SCNs. This paper looks at real world SCN topologies using data presented in Willems (2008) and establishes a general set of topological characteristics for the manufacturing sector. Based on observations from twenty six SCNs, it is found that, although manufacturing SCNs exhibit power law node degree distributions, they tend to be hub and spoke like, with hubs that are significantly larger than those predicted by the BA model. Also, the SCNs indicated the presence of closely knit, vertically integrated communities of firms, while highly connected firms were found to avoid connecting with each other. Since the above structural properties are not features of the networks generated by the BA model, it is concluded that a new mechanism is required to model the growth of SCNs.