Search

Increasing awareness of the intrinsically complex nature of supply networks has brought the field of supply chain management into the domain of network science. However, due to the difficulties of acquiring large-scale and consistent empirical data sets, a more complete picture of a real-world supply network has remained remarkably elusive. In this paper, we present novel data that characterize the Toyota supply network, and identify key structural features using measures from social network analysis and the more recent field of network science. We show that the network structure for the Toyota supply network departs widely from the simplified models on which much previous work is based. Our analysis reveals the heterogeneous composition of the network and identifies key firms. Further analysis reveals the existence of constituent sub-networks, and we show that their structures reflect various factors, such as product categorization, geographical closeness and business alignment. Mapping the topology, geography, and distribution of productive capabilities for this supply network provides a critical first step for developing a more empirically-grounded theory of distributed production.

We assemble a large‐scale empirical dataset that allows us to examine the local and global topology of relationships between firms in the Toyota supply network. On this basis we propose novel measures that allow us to characterise the resilience of the entire supply network. Our findings show that simple linear supply chain models are inadequate, because they neglect important lateral dependencies between suppliers. Hence, we argue that it is necessary to describe and model the supply chain as a complex network. We observe that the degree distribution for this network scales exponentially, so that disruptions at randomly chosen suppliers have little impact but vulnerability to disruptions at highly connected suppliers is significant. These potential vulnerabilities are mitigated by the network’s ‘small‐world’ structure, where the average path to any given supplier via other suppliers is very low, and the number of relations between suppliers that produce the same product types is high, especially within the Kyoho‐kai supplier association. Membership of the tightly‐knit Kyoho‐kai supplier association is positively correlated with an increase in the number of connections that a supplier has, leading to a segmentation that favours highly connected hubs. The network also exhibits a high degree of product type redundancy, where multiple suppliers offer similar product types. When we examine product diversity along the chain we find that upper tiers show higher diversification given their vulnerability to changing customer demands. Our analysis of a unique, large‐scale empirical dataset aims to move the field of supply chain management beyond stylised facts, and to demonstrate how methods from interdisciplinary work on complex networks can contribute novel insights.

We use a map of the interconnections between Toyota’s extended supply base to review the concept of ‘length’ in supply chains, and show that the chain is significantly shorter than expected. To explore why, we draw on ideas from network science and develop an alternative theoretical frame to the conventional ‘transaction cost economics’ approach. Using particular inspiration from the domain of food webs we discuss whether additional explanations such as supplier selection bias, product complexity, size and dynamical constraints, and a common lean culture can help explain the dimensions of the Toyota chain.

Working paper