Where to Invest Resilience in Supply Chain Network under Stochastic Disruptions?

Much of the previous research on supply network resilience examined what capabilities firms should develop and how they develop the capabilities as resilience investment. However, what has received less attention is where to invest resilience in the network. Scholars have debated over what nodes affect the resilience of a supply network the most (e.g., critical nodes vs. nodes on critical paths). Yet they have not investigated where (which nodes) to invest resilience given limited resources. Motivated by real problems from Cainiao Network, this study reconciles the debate over node vs. path and proposes the optimal strategy of resilience investment that maximizes expected max-flow through the network. We consider two routing mechanisms with respect to whether materials have to flow through pre-specified paths and three levels of disruption probability (rare, general, and frequent). The analysis reveals that under both routing mechanisms, investing in the nodes of the highest capacities is optimal under rare disruptions, whereas investing in the short paths with large flow is best under frequent disruptions. However, the optimal investment problem under general disruptions is NP-hard. Therefore, we provide node-greedy and path-greedy algorithms to solve the problem. Using the distribution network and operational data provided by Cainiao, we show that the two greedy algorithms perform better than their theoretical performance guarantees. We further identify the driving forces of the optimal resilience investment