Modeling supply chain risks and risk mitigation strategies with Bayesian networks

The globalization of supply chains, increasing product variety, single sourcing, increasing interdependencies between companies, increasing customer expectations and the shortening of lead times lead to vulnerability of business to risks. Nowadays, supply chains are expanding rapidly and becoming more complex and risks are causing serious disruptions. Therefore, studies in supply chain risk management have attracted the attention of academics and practitioners. Companies need to identify and assess risks and determine appropriate risk mitigation strategies. This study focuses on the automotive sector, which is highly vulnerable to risks due to the high level of complexity of its supply chain and interdependence of chain members. In this study, firstly, a current and comprehensive literature review on supply chain risks and risk mitigation strategies was conducted. Subsequently, semi-structured interviews were conducted with 20 supply chain experts working as managers in 15 different automotive companies to identify risk drivers, risks, and appropriate risk mitigation strategies. After evaluating the interviews with the experts, a Bayesian network model was created to determine the risk probabilities and the probabilities of applying risk mitigation strategies. Scenario and sensitivity analysis on the model were conducted and the results of these analyses were examined. According to the model outputs, supply chain risk has the highest probability of occurrence. Operations risk is ranked the last. The most important reason for this finding is that internal risk factors in the automotive industry are largely under control. It has been observed that the collaboration strategy, one of the risk mitigation strategies, is an important strategy that is effective against all risks and has a high probability of implementation. As a result of this study, important findings of the automotive supply chain risks and risk mitigation strategies were obtained based on the practitioners' evaluations and their real-world examples. The findings will contribute to the knowledge of managers working in the automotive sector and researchers in the field of supply chain risk management. Also, A Bayesian network model, which can be updated as new data are available, is presented that can help decision-makers in the automotive sector to determine the possibilities of applying risk mitigation strategies according to risk probabilities.