Development of MDA and Logit models to predict bankruptcy probability for Russian coal enterprises

Subject. The study deals with forecasting the probability of bankruptcy for coal enterprises, represented by two variants of mathematical models: the MDA model and the logit model. Objectives. The aim is to develop a methodology for predicting the probability of bankruptcy based on data from Russian coal industry enterprises for 2014–2022. Methods. The study employs the coefficient method, correlation analysis, k-means method, cluster and factor analysis, MDA and logit models. Results. The development of models included several stages: formation of a basic system of financial coefficients; creation of a sample including two classes of coal enterprises (bankrupt/liquidation stage and operating), calculation of financial coefficients based on financial statements of organizations; correlation analysis to exclude multicollinear indicators; factor analysis to reduce the number of variables; model development and interpretation. As a result, we devised two bankruptcy assessment models (the MDA model and the logit model), with high quality indicators, based on the modern initial base of financial coefficients of Russian coal industry enterprises. Conclusions. The offered models are able to predict the bankruptcy of coal enterprises in the short term (two years) with forecast accuracy of up to 90% or more.