演化式多重組合羅吉斯迴歸模型－應用於信用評等

新巴塞爾協定建議銀行採用內部評等法自建授信系統以減少人為錯誤帶來的作業損失，並且能夠快速正確處理授信放款。目前被廣泛應用於信用風險預測模型的是羅吉斯迴歸，此模型雖然可以分成多類，但其S曲線通常以等距或經驗法則切割門檻值做信用分等，當違約機率產生變動時會造成評等等級變動的不對稱現象。因此，本研究提出演化式多重組合羅吉斯迴歸模型(EMCRS)，每一個信用評等設計一個羅吉斯迴歸模型進行違約預測以決定評等的等級，希望能提供分類的準確性；模型的門檻值與預測變數是藉由遺傳演算法以非線性方式做最佳化；以此研究模型建立一套演化式多重組合羅吉斯迴歸信用評等系統。預測變數是擷取自財務變數、基本資料變數、會計師變數、總體經濟變數及公司治理變數季資料等五個構面進行預測。另外，其目標函數是以新巴塞爾協定建議的驗證方法對本研究模型進行違約預測效力、評等穩定性以及等級同質性三方向的模型驗證。實驗結果發現，EMCRS信用風險違約評等的預測效力以及等級同質性方面相較於其他四種評等模型（TCRI、區別分析、決策樹、羅吉斯）均有不錯的表現；多季的違約時間點實證中，反應出最近期的財報與公司治理等相關資訊的揭露對模型具有較佳的預測效力；代表穩定性指標的移轉矩陣率會隨著使用者設定的評等級數增加而呈下降的常態現象；具有關鍵因子含財務變數：每股淨值、常續性EPS、每股稅前淨利、總資產週轉率（次），其中常續性EPS被選擇的次數為最多，其次是公司治理變數的董監酬勞佔稅前淨利%，足見EMCRS評等模型受財務變數類與公司治理變數較具影響力。本文的發現對於建構信用評等模型有重要的貢獻與影響。

關鍵字

羅吉斯迴歸模型；遺傳演算法；新巴塞爾資本協定；信用評等；違約機率

並列摘要

Some serious financial issues, such as the Asian Financial Crisis and Subprime Mortgage Crisis, have occurred in the last two decades. It is not self-evident that the real economy may suffer from the credit crunches as a result of the financial crises and bank inadequate management. Due to one of banks' major sources of profits being loan growth, especially in enterprise loans, it is important to manage and evaluate corporate financial risk effectively. Published in June 2004, Basel II is a well-known international initiative that requires banks to have a more risk sensitive framework. It establishes regulatory expectations for credit risk through the Internal Ratings Based (IRB) approach, which allows banks to assess key risk drivers as the primary capital calculation. In statics, the logistic regression is only suitable for probabilistic binary classification, but it cannot provide multiple classifications. Although cumulative logic regression (CLR) introduces a multi-class algorithm, it is hard to decide the thresholds in CLR. This paper proposes an evolutionary MCLR credit rating system (EMCRS) that uses an evolutionary approach to optimize multiple combinatorial logistic regression models. We implement GA to estimate non-stationary time-series data with dynamic non-linear searching capabilities. Finally, the EMCRS is verified by (1) capably predicting the default rate (e.g. KS, ROC, CAP), (2) rating stability (e.g. TM), and (3) grade homogeneity (e.g. CIER). The experimental results demonstrate that EMCRS has better competence to predict the enterprise default rate than TEJ. It is reasonable that rating stability will decrease if the number of ratings increases. Profitability, earnings per share, and management factors are critical for evaluating the performance of EMCRS.

並列關鍵字

Multiple Combinatorial Logistic Regression Model ； Genetic Algorithm ； Basel Ⅱ、Credit Rating、Default Probability

參考文獻

Altman, E. I.(1968).Financial Ratios, Discriminant Analysis and the Prediction of Corporate Bankruptcy.Journal of Finance.23(4),589-609.

Google Scholar

Back, B.,Laitinen, T.,Sere, K.(1996).Neural Networks and Genetic Algorithms for Bankruptcy Predictions.Expert Systems with Applications.11(4),407-413.

Google Scholar

Baesens, B.,Van Gestel, T.,Viaene, S.,Stepanova, M.,Suykens, J.,Vanthienen, J.(2003).Benchmarking State-of-the-Art Classification Algorithms for Credit Scoring.The Journal of the Operational Research Society.54(6),627-635.

Google Scholar

Balcaen, S.,Ooghe, H.(2006).35 Years of Studies on Business Failure: An Overview of the Classic Statistical Methodologies and Their Related Problems.The British Accounting Review.38(1),63-93.