Methods for Supplier Default Risk Assessment in Supply Chains: A Review

Abstract

Supplier default risk represents a critical concern in supply chain management. This paper systematically reviews three quantitative approaches for supplier default risk assessment: weighted scoring, logistic regression, and Bayesian methods. The weighted scoring method constructs indicator systems with subjective or objective weights to rank suppliers by composite scores. Logistic regression models the log-odds of default as a linear function of supplier characteristics to estimate default probability. Bayesian methods, which are closely related to logistic regression, combine prior distributions with observed data to generate posterior probabilities, enabling dynamic updating and small-sample inference. Based on the comparative analysis, this paper proposes a multi-dimensional framework to guide method selection in procurement decisions. Research gaps and future directions are also identified, including dynamic updating mechanisms, integration of unstructured data, and the convergence of Bayesian methods with machine learning. Future research should explore real-time risk assessment and hybrid approaches that combine methods with machine learning to support more scientific supplier evaluation and enhanced supply chain resilience.