Sistema de classificação de risco de crédito: uma aplicação a companhias abertas no Brasil

This paper examines whether default events of public companies in Brazil are predicted by a credit rating system based on accounting ratios. The credit rating system developed uses cluster analysis to classify companies in eight ratings, seven for solvent and one for default companies. The variable used to assign ratings to companies is the probability of default estimated by credit risk model developed by Brito and Assaf Neto (2008). The credit rating system assigns annual ratings to non-financial companies listed on Brazilian stock market from 1994 to 2006. Based on these ratings, several risk migration matrices are generated for the analysis period. The migration matrices show the risk increase of default companies prior to the year of occurrence of default event. Most of these companies are classified in the lowest rating or migrate to lower ratings in the years preceding the default. In addition, mortality rates of firms grow in the credit rating system. These results demonstrate that default events of companies can be predicted by a credit rating system based only on accounting ratios.