A Study of Credit Rating Model of Financial Institutions in Taiwan

[[abstract]]本文以中華信評2005至2008年受評之台灣上市(櫃)金融機構為研究樣本，並將金融機構中的銀行業提出做個別分析，使用財務、公司治理與市場面變數納入多元區別分析(Multiple Discriminant Analysis, MDA)模型和Ordered Logistic Regression模型來進行信用評等的預測。研究結果顯示，金融機構如果規模較大且有好的資本適足率、獲利能力、流動性資產比例高以及違約風險低，較容易獲得較佳的評等等級；銀行業則是規模較大與違約風險較低的銀行較容易獲得較佳的評等等級。整體而言， MDA模型不論是何種群體或加入市場變數與否都比Ordered Logistic Regression模型之預測準確度高，因此本研究建議將財務、公司治理和市場面三種因素同時放入MDA模型中進行估計，較為符合金融機構信用評等評估之模式。[[abstract]]The research investigated the public financial institutions which were rated by credit rating agency – the Taiwan Ratings Corporation (TRC) during 2005 to 2008. Multiple Discriminant Analysis (MDA) and Ordered Logistic Regression Model were employed. Additionally, financial variables, corporate governance variables, and market variables were also used in predicting credit rating of these financial institutions. Results indicated that while a large-scale financial institution with better capital adequacy ratio, higher profitability, higher ratio of liquid assets and lower default risk, it subsequently received a higher credit rating. Similarly, a public large-scale bank with lower default risk received a higher credit rating. In general, the MDA model received better accuracy rate than the Ordered Logistic Regression Model regardless what kind of samples or market variables applied. Therefore, the author suggests that it seemed to be more appropriate to add in financial, corporate governance, and market variables in the MDA model, in order to obtain credit rating prediction for financial institutions