SRA-E Lisbon 2017

The effective forecasting of default is crucial for financial institutions and other counterparties to make lending decisions and partnership agreements. In accordance with the recommendations of the Basel Committee on banking supervision risk analysis of corporate borrowers should incorporates external ratings of international rating agencies and internal ratings-based approach. Consistent with the Japanese philosophy of kaizen this approach has to regularly improve in order to correspond to changing market conditions. In our study we used a two-step Heckman selection model to solve the problem of sample selection bias. The first step is to develop a logistic regression to evaluate the probability of default. The second step is to develop OLS or GMM model (Heckman's lambda is used as one of the predictors) to evaluate the scale of the problem in the case of default. It is defined as "holes” in the capital of companies (negative net worth) - the book value of assets minus the book value of liabilities. According to our previous estimates corporate governance factors have a positive impact on the quality of the predicted probability of default models in Russian construction industry. As the dependent variable tested several modifications of default. Additionally we checked overfitting problem and after that have been tested two kinds of regularization (Ridge and Lasso) of logistic regression – result was sustainable. The following hypotheses were confirmed: other things being equal, the probability of default of the company is lower, if CEO is co-owner; other things being equal, the probability of default of the company is higher if the company is a subsidiary. Moreover the companies with small and more cohesive board of directors (co-owners) better overcome financial distress – possibly due to the greater flexibility and the ability to quickly make key decisions. Moreover we tested the hypothesis about the significance of company’s expectations about the volume of production and demand during a year to evaluate probability of default. In continuation of this topic we conducted a comparative analysis by manufacturing industry of Russia, results of which will be presented at the conference.