Comparison of Financial Distress Prediction Models Based on Likelihood Ratio Test, ROC Curve, and Classification Table

• Main Authors: Deng, Bou-yuan, 鄧博遠
• Other Authors: Chang, Ching-fu
• Format: Others
• Language: zh-TW
• Published: 2005
• Online Access: http://ndltd.ncl.edu.tw/handle/87448975817389877082

碩士 === 國立政治大學 === 會計研究所 === 93 === The 1999 Basel II Accord suggests banks measure the impossibility of reimbursement of debtors to calculate capital minimums by internal ratings-based approach. To reduce the credit risk, it is important that banks construct accurate financial distress prediction systems to determine the probability of financial distress of debtors. This study employs logistic regression and discrete-time hazard analysis to construct nested models to which the financial, non-financial, and corporate governance corporate variables are added step by step. I therefore make comparison of the performance of three models under logistic regression and discrete-time hazard analysis, respectively. Meanwhile, the comparison of the performance of logistic regression and discrete-time hazard analyses under each of three models is also made. The empirical results show that the in-sample predictive ability of financial distress is enhanced by gradually incorporating different kinds of variables in both analyses. Although the out-of-the-sample predictive ability of financial distress is not improved by gradually incorporating different kinds of variables in one analysis, the model performance is quite well overall. The entire discriminability of discrete-time hazard analysis is better than logistic regression under each model.