News Impact on the Probability of Default: Evidence from Russian Retail Companies

This study is aimed at investigating the issue of the news sentiment application to bankruptcy prediction model and the impact of the news sentiment on the probability of the default on the example of Russian retail. The study analyzes 190 companies: 95 Russian retail companies that went bankrupt in 2015-2019 and 95 non-default analogues. The predictor set consists of 34 financial variables and sentiment variables aggregated using «bag-of-words» from the news corpus of 4877 articles. The novelty of this study is the application of the news sentiment to the bankruptcy prediction problem. The methodological base of this work is the Logistic Regression used as a benchmark model, and machine learning models: Random Forest, Support Vector machine and Multilayer perceptron. The results show that adding sentiment variables improves the accuracy of the probability of the default model and provides some word characteristics of companies which are financially stable and not financially stable.