Probabilistic modeling of multi- echelon retail supply chains in crisis conditions

This paper examines a large centralised multi-echelon supply distribution system of the O’STIN retail chain during a period of exogenous shock. Empirical studies of multi-echelon supply chains based on real data from large logistics systems remain relatively scarce. Of particular interest is the analysis of changes in the distribution policy under exogenous constraints at key logistics nodes, illustrated by a temporary reduction in the central warehouse throughput caused by major scheduled maintenance works. The aim is to quantify changes in the supply distribution policy across retail stores under resource constraints. A distribution model across retail outlets is formalised, and the sensitivity of distribution to store sales-volume rank is estimated. The empirical analysis draws on data from a retail logistics system comprising a central warehouse, a regional warehouse, and over seventy stores. A Bayesian hierarchical multinomial logistic regression model is employed, with parameters estimated via Markov chain Monte Carlo. The results demonstrate increased dependence of supply distribution on store rank during the resource constraint period. The response across system levels proved heterogeneous: the central warehouse exhibits growing concentration of supplies, whereas distribution from the regional warehouse becomes more uniform. These findings suggest that store prioritisation mechanisms sustain commercial efficiency even under constrained resources. The practical contribution lies in the applicability of the proposed approach to analysing distribution mechanism robustness in large retail networks. Limitations include the analysis of a single logistics system over a restricted observation window. Future research may focus on expanding the sample and modelling alternative supply chain scenarios.