Investors are being encouraged after the global crisis to reduce their dependence on the largest credit rating agencies for risk assessments of companies and securities. Comparing risk assessments from different sources rapidly becomes non-trivial when more than three credit rating agencies are involved. We propose a method for comparing rating scales, and hence constructing correspondence diagrams and tables, thereby treating the rating scales used by different agencies as objects of study. Scales are compared by looking at sets of ratings assigned to similar entities (hereinafter banks) with the assumption that the risk being measured by each credit rating agency is the same for a given rated entity at a given point in time. Studying international bank ratings for a five-year period shows that there are subtle differences for the largest credit rating agencies. A mechanism for constructing mappings between scales could lead to more competition with new credit rating agencies.
The aim of this paper is to construct a reliable banks’ rating model for the main international agencies based on public information for the potential practical use. The Bankscope database for the period from 1996 to 2011 was used in the research. The ordered probit models show that inclusion of macroeconomic variables as well as the regional dummies improve their explanatory power. Moreover, the significance of the time dummies allowed us to conclude that rating agencies do change their grade when an economy operates on the different business cycle stages. Furthermore, the conclusions of a conservative nature of Standard & Poor’s ratings and overvalued Moody’s grades compared to the rating agency Fitch were performed. The models were checked for the in-sample and out-of-sample fit including distributional comparisons across agencies. The obtained model was classified as practically useful, as it gave 31 % of precise results and up to 70 % forecasts with an error within one rating grade. Moreover, 62 % of rating classes of banks were predicted without an error and more than 95 % of rating classes’ forecasts had an error within one rating class.
We compared the ability of various empirical methods to reproduce public credit ratings (PCRs) of industrial companies (ICs) from BRICS countries using publicly available information. This task is important for researchers and practitioners because many of BRICS’ ICs lack PCRs from reputable rating agencies such as Moody’s, Standard and Poor’s, and Fitch. This paper aimed at filling the gap in the existing research as insufficient efforts were focused on prediction of PCRs of ICs from the entire BRICS IC community. The modeled variables are credit ratings (CRs) of 208 BRICS’ ICs assigned by Moody’s at the year-end from 2006 to 2016. The sample included 1217 observations. Financial explanatory variables included companies’ revenue, operating profitability, interest coverage ratio, debt/book capitalization, and cash flow debt coverage. Non-financial explanatory variables included dummies for home region, industry, affiliation with the state, and a set of macroeconomic data of IC’s home countries. The set of statistical methods included linear discriminant analysis (LDA), ordered logit regression (OLR), support vector machine (SVM), artificial neural network (ANN), and random forest (RF). The resulting models were checked for in-sample and out-of-sample predictive fit. Our findings revealed that among considered methods of artificial intelligence models (AI), SVM, ANN, and RF outperformed LDA and OLR by predictive power. On testing sample, AI gave on average 55% of precise results and up to 99% with an error within one rating grade; RF demonstrated the best outcome (58% and 100%). Conversely, LDA and OLR on average gave only 37% of precise results and up to 70% with an error within one grade. LDA and OLR also gave higher share of Type I errors (overestimation of ratings) than that of AI. Therefore, AI should have higher practical application than DA and OLR for predicting the ratings of BRICS ICs
We compare several models for estimating the default probabilities of Russian banks using national statistics from 1998 to 2011, and find that a binary logit regression with a quasi-panel data structure works best. We conclude that there is a quadratic U-shaped relationship between a bank's capital adequacy ratio and its probability of default. In addition, macroeconomic, institutional, and time factors significantly improve model accuracy. These results are useful for the national financial regulatory authorities, as well as for risk-managers in commercial banks.