Document Type



The change of economic circumstances in recent years has increased the gaps of credibility among firms, and the necessity to measure the credibility adequately is an important ever-increasing factor. As the result, it is essential for banks to assess the rating evaluation of borrowers appropriately in order to measure the amount of credit risk. This paper reviews preceding studies regarding model building that enables us to explain or forecast the ratings of firms issued from creditassessment agencies. The paper then explains the features of the ordered logit model in detail that is nowadays used as a ratings evaluation model throughout the world. Finally, some defect points of these preceding models are pointed out and the improved ordered logit model on the noncompensatory rule that enables to compensate these defect points is proposed. In addition, the appropriateness of the proposed model is verified based on the actual rating data for the companies.