Default Prediction for Companies Operating in Information Technology Sphere Using a Proportional Hazards Model

This study examines the issue of predicting bankruptcy for IT companies in Russia. We use a proportional hazards model (Cox regression) as a bankruptcy model , which allows assessing the company’s risks at each discrete point in time. The sample of IT companies includes 97 companies with bankruptcy arbitration cases between January 1, 2021 and February 29, 2024 and 194 healthy companies. In this study we consider several specifications of the model: only with financial predictors, only with non-financial predictors, and with their combination. The model specification with a combined set of variables has the greatest forecast accuracy. Its overall accuracy is 78%, which is 5.1% higher than the accuracy of the specification based only on financial variables. The quantitative impact of each explanatory variable on the bankruptcy risk of IT companies in Russia has been studied. We predict survival probabilities for a period of 30 months for IT companies using the best specification of the model.