| 摘要Abstract | 信用評等是在資訊不對稱下，外部人評估公司之信用程度以及經營狀況等的 重要指標，且由於以往的研究通常將單一模型適用至各類公司，但每家公司的特質 有所不同，其信用評等模型之各變數的權重值應有所不同。本研究將公司治理因素 結合至信用評等模型，且採 Cubist 迴歸樹模型的方法來建構，其優點可以將影響信 用評等模型之關鍵因素萃取出來，而不同的門檻之下各有其所屬的信用評等模型， 各評等模型攸關之因素亦會有所差異，可以配適出適合各公司類型的信用評等模 型。且本研究融入公司治理因素強化信用評等模型，可以修正以往研究僅考慮財務 因素的不足之處。經實證與驗證結果發現，本研究所創建的信用評等模型是更完整 的模型，能更準確評估公司的信用評等狀況，以反映公司的經營現狀以及推估發生 信用違約的可能性，能有助於預測公司的未來發展，並提供財務報表使用者更有價 值的參考資訊。Credit ratings are an important index used by outsiders to predict the credit level and operational performance of a firm under the circumstance of information asymmetry. Most previous studies have applied a single model to fit all types of firms, often resulting in inaccurate estimations as different types of firms have different characteristics. The variable weights of a credit rating model should not be the same. This study incorporated the factor of corporate governance into the credit rating model, and used Cubist regression tree to construct an empirical model. The proposed model is capable of extracting the key factors that affect the credit rating model, thereby taking into consideration the thresholds and variables appropriate to firms. This model can overcome the insufficiencies of previous studies that only considered financial factors. Our empirical results show that the proposed model provides a more complete and accurate assessment of the credit rating of firms, reflecting current operational phenomena, and inferring the probability of default. This model can be used to predict the future development of firms and provide accurate information for users of financial statements. |