| Summary: | 碩士 === 逢甲大學 === 金融碩士在職專班 === 104 === This study builds a Financial Distress Prediction Model to investigate the causal relationship of financial ratio and corporate governance as variables to the financial distress of an enterprise. The Financial Model, Corporate Governance Model, and Integrated Model are built respectively in order to compare the prediction ability of the models. The studied object is listed and OTC domestic companies, and the studied period is 2005-2015; this study chose 74 companies with financial distress and 148 normal companies based on the paired sample method, and used Logit Regression Model as the Prediction Model. The modeled period is 2005-2012, and the verified period is 2013-2015.
The empirical results showed that the variables having a significant positive correlation with the occurrence of financial distress include financial liabilities ratio due within one year, debt ratio, number of times CPA was switched in the last three years, and turnover of CFO within 3 years. Variables having a significant negative correlation with the occurrence of financial distress include earning before tax margin, operating margin, cash flow per share, earnings per share, equity ratio, total asset turnover, directors and supervisors holding rate, control rights, pyramid structure, cross-shareholding structure, and independent directors and supervisors seats.
The classification in this study is created according to the probability threshold value of 0.5 and the probability threshold value of Martin(1977), when the probability threshold value of Martin(1977) is adopted, all models can be used to effectively reduce Type I error, and improve the classification accuracy of distress of the company, enabling banks to make less credit loans to high-risk customers and avoid bad debt and losses, but on the other hand, the classification error rate of Type II error will be raised, so the overall classification accuracy cannot be improved.
The closer the data is to the year before the financial distress occurred, the overall classification accuracy of modeled samples in the integrated model is more likely to reach 98.85%, and the overall classification accuracy of verified samples in the integrated model is more likely to reach 91.67%, indicating that the serious financial deterioration and ineffective corporate governance play an important role on corporate financial distress, the integrated model can more effectively detect indicators of financial and corporate governance weakening, and achieve the early prediction.
Analyzing the samples of the year before the financial distress occurred which are classified as having a Type I error in the integrated model, the study found that they were already classified as financial distress in the Prediction Model 2 and 3 years before the financial distress occurred, indicating that before a financial distress occurs, it is actually traceable; using the integrated model in any year can achieve the purpose of early prediction. When the sample is classified as having a Type II error in the integrated model the year before the financial distress occurred, it may be an indicator that the companys financial ratios and corporate governance are weakening. Predicted Default Rate of finance and corporate governance implies some risk information that can be provided to decision-makers and investors to determine the risks. Logit Regression Model calculates the predicted default rates, enabling a better understanding of the meaning of credit risk degree of each sample, as well as the acquirement of credit risk information of the counterparty, to achieve the purpose of risk management. Based on risk preference, banks can assess the most beneficial and appropriate model to manage and monitor enterprises credit risks.
|
|---|
