| Summary: | 碩士 === 國立中山大學 === 財務管理學系研究所 === 99 === A financial failure prediction model should be dynamic by adding latest information in an effort to improve the current predictive power, and this model also can be applied to different industries and periods. That is, it has prominent goodness of fit and stable parameter. In this study, I testify that if the modified independent variables, industry relative ratios, can improve the prediction rate by using logistic regression. My research is based on public information.
This study constructs two kinds of model：Model I is constructed with original financial ratios and Model II with relative industry ratios. Both models incorporate additional variables related to corporate governance. My empirical results suggest that relative industry ratios enhance the predictive power of financial failure prediction within three partially overlapping periods.
Further study focus on Model II, I isolated firms which are confronted with financial difficulties and they can’t be discriminated from other normal firms by using the prediction model. My result demonstrates that the main difference between the former and the latter is debt/equity ratio. Those firms which can’t be detected afford less liability. In addition, my studies also compare these undetected firms with their control group and find they still can be distinguished from their control group by using logit model. The accuracy rate of prediction can reach 92.42%.
Last study we use event study to research the links between the default possibilities of firms and their stock prices. My results demonstrate that the default possibilities may cause abnormal returns.
|
|---|
