A Study on the Prediction Models ofCorporate Financial Distress

• Main Authors: Horng-Cherng Ku, 顧紘誠
• Other Authors: none
• Format: Others
• Language: zh-TW
• Online Access: http://ndltd.ncl.edu.tw/handle/98887933065748885035

碩士 === 逢甲大學 === 經營管理碩士在職專班 === 92 === Abstract The prediction of financial distress used to rely on the financial statements of each company, which were also used by banks to evaluate each company’s past performance and estimate the riskiness of potential borrowers. However, the financial reports may easily be managed or manipulated. Thus, how to design a more reliable financial distress precaution model has been an important issue emphasized by financial institutions. This research uses the market-value financial ratios and the characteristics variables of the companies to establish the prediction model of financial distress, and also to investigate whether the market-based financial ratios are better predictors of financial distress, compared with the traditional book-value financial ratios. The major financial ratios, firms’ characteristics variables, and the combination of these variables are used in this study. The logistic regression and the neural-network methods are employed to analyze the various models of the research. The empirical evidence of our study shows that when the logistic regression method is used, the firms’ characteristics variables provide a better prediction of financial distress. The average correct prediction percentage is 76.47% during a three-year period. On the other hand, when the neural-network method is used, the combined model gives a better prediction. The average correct forecasting percentage is 76.21% during a three-year period. Both methods tend to provide a better prediction as the financial distress approaches. 【Keywords】 Financial Distress, Financial Ratios, Logistic Regression, Neural Network