Abstract:
Researchers have been forecasting ratings change for decades, most of them have estimated single-period classification model, which is static model with multiple-period ratings change data by neglecting the fact that firms change through time. Therefore, a new model, hazard rate model, is proposed (Shumway 2001). Theoretically, hazard rate model is more appropriate than static model for forecasting since hazard rate model explicitly considers the time-varying of both ratings and firms’ characteristics. The purpose of this study is to compare the accurate performances on predicting issuer ratings change between using static and hazard rate models. Empirical results indicate that hazard rate model outperforms static models in out-of-sample forecasts. Moreover, there were distress failures of great companies even though they had investment rates leading some to conclude that these collapses occurred because of accounting fraud and corruption. However, alternative explanation of these failures is that the rating agencies are now using softer standards in assigning ratings. This study examines standard of rating agency by employing ordered logit model and results suggest that rating standards are lenient. Therefore, the failure of great firms in the past can be partly explained by the softer standa.rd of rating agency.
