Application of Survival Analysis to Predicting the Risk of Company’s Bankruptcy in the Russian Food Production Sector

At the current stage of development of the Russian economy, characterized by instability and constant economic deterioration, the ability to predict company’s bankruptcy plays an extremely important role in the identification and elimination of financial distress. This work is aimed at applying methods of survival analysis, such as Cox proportional hazard model with time dependent covariates and Kaplan-Meier estimator to predict the bankruptcy risk of firms from Russian food production sector. To achieve the aim of this work it was needed to perform some tasks: analyze the literature, aimed at exploring the question of bankruptcy prediction; review the theoretical foundations of survival analysis; collect and process data, derived from financial statements of firms; build the survival curves to determine the probability of bankruptcy of existing companies at some fixed point of their existence; build a Cox regression model to determine the factors that affect the risk of firms becoming bankrupt, and a logistic regression model which is commonly used in predicting financial insolvency; compare the previously constructed models to select the one with best specification. During the study, it was found that the Сox model with time dependent covariates shows a better quality than the parametric logistic regression model having a predictive power of 76%. This model assumes that the risk of the company’s bankruptcy depends on five time varying covariates, represented by different financial ratios, and on eight fixed variables responsible for the qualitative characteristics of companies. By constructing the survival curves using the Kaplan-Meier estimator, it was revealed that the most dangerous period in company’s lifetime occurs in the first five years of its existence.