A note comparing support vector machines and ordered choice models' predictions of international banks' ratings

摘要

We find that Support Vector Machines virtually always predict international bank ratings better than ordered choice models.\ud\udRatings of sovereign risk, corporate bonds and financial institutions conducted by rating agencies (RAs) may be seen as instruments that provide investors with prima facie information about the financial position of the subject in question and on the price of credit risk. Pinto (2006) argues that RAs opinions facilitate capital allocation through supplied information about the financial position of the companies in question. Indeed, the RAs' exclusive position may be justified because they reduce asymmetric information between investors and companies. Ratings are ordinal measures that should not only reflect the current financial position of sovereign nations, firms, banks, etc. but also provide information about their future financial positions. There has been extensive research in predicting bond ratings using multi-variate discriminant analysis, ordered choice models, non-parametric techniques and combined methods’ forecasts to predict bond ratings - see, Altman and Saunders, (1998), Kamstra et al (2001) and Kim (2005). Thus, we employ financial variables, in addition to country risk (which we model using country specific dummy variables), as determinants of bank ratings in our modelling. The main challenge in modelling ratings is to increase the probability of correct classifications. This motivates our comparison of Support Vector Machines (SVMs) with ordered choice models for predicting individual bank ratings as produced by Fitch Ratings (FR).