Comparative Analysis of Discriminatory Power of Credit Ratings and Credit Spreads in Problems of Default Prediction

In this paper, a comparative analysis of the discriminatory power of credit ratings and credit spreads in the prediction of defaults was conducted. In the theoretical part, the credit spread was considered as an alternative approach to assessing the creditworthiness of the borrower (issuer) and predicting defaults. Data bases were created for all rated bonds, whose issuers had a credit rating for the period from January 1, 2010 to December 31, 2015. At the same time, an assessment was made of the degree of influence of the credit event (changes in the issuer's rating by the international agency) on the prediction of a default, as well as an assessment of the degree of influence of credit spreads on the prediction of a default. I also compared the correctness of the classification of borrowers on the ratings between the credit ratings of agencies and the estimates of credit risk inherent in the prices of financial instruments for the Russian market. To compare the discriminatory power of credit ratings and the discriminative ability of bond spreads, a logit model was used and ROC curves were constructed. A representative sample included 62 banks and 40 companies from the real sector of the Russian market. The ratings of Russian and international rating agencies and G-spreads of corporate bonds were considered. Further statistical test to compare the area under the ROC curves led to the conclusion that the G-spread has a high discriminatory power and predicts a probability of default better than the ratings assigned to RA Expert. At the same time, the ratings assigned by such rating agencies as NRA, Standard & Poor's and Moody's Interfax Rating Agency and the corresponding G-spreads do not differ in their predictive power.