A network science approach to analysing manufacturing sector supply chain networks: Insights on topology

Abstract

Due to the increasingly complex nature of the modern supply chain networks (SCNs), a recent research trend has focussed on modelling SCNs as complex adaptive systems. Despite the substantial number of studies devoted to such hypothetical modelling efforts, studies analysing the topological properties of real world SCNs have been relatively rare, mainly due to the scarcity of data. This paper aims to analyse the topological properties of twenty-six SCNs from the manufacturing sector. Moreover, this study aims to establish a general set of topological characteristics that can be observed in real world SCNs from the manufacturing sector, so that future theoretical work modelling the growth of SCNs in this sector can mimic these observations. It is found that the manufacturing sector SCNs tend to be scale free with degree exponents below two, tending towards hub and spoke configuration, as opposed to most other scale-free networks which have degree exponents above two. This observation becomes significant, since the importance of the degree exponent threshold of two in shaping the growth process of networks is well understood in network science. Other observed topological characteristics of the SCNs include disassortative mixing (in terms of node degree as well as node characteristics) and high modularity. In some networks, we find that node centrality is strongly correlated with the value added by each node to the supply chain. Since the growth mechanism that is most widely used to model the evolution of SCNs, the Barabasi - Albert model, does not generate scale-free topologies with degree exponent below two, it is concluded that a novel mechanism to model the growth of SCNs is required to be developed.