- 學位論文
破產預測方法之實證研究
Empirical Investigation of Alternative Bankruptcy Prediction Approaches
若您是本文的作者，可授權文章由華藝線上圖書館中協助推廣。
摘要
本篇論文將焦點放在破產預測方法的比較上，所包含的方法包括類神經網路方法、Z-Score、KMV、和單獨使用ROA變數來預測破產等方法。在類神經網路方面，我們另用了兩個不同的變數-Moody的變數和Z-Score的變數來建構模型。 為了比較這些方法，我們使用contingency tables去驗證模型的正確性，並使用power curve去驗證模型的預測能力。我們得到一致性的結果，模型的表現於好到壞分別是：用Moody變數所建構的類神經網路模型、用Z-Score變數所建構的類神經網路模型、KMV、Z-Score、和ROA。
關鍵字
破產預測並列摘要
In this paper, we pay our attention to compare the performance of approaches which contains neural network approach, Z-Score, KMV model, and ROA variable. We structure Back-Propagation Network which uses two difference variable, Moody’s variables and Z-Score variables. To examining the validation of models, contingency tables are used to test the model calibration and power curve is used to test model power. We get the identical conclusion that the performance form best to worst are BPN using Moody’s variable, BPN using Z-Score variables, KMV model, Z-Score approach, and ROA variable. Comparing with the result of Jorge R. Sobehart, Roger M. Stein (2000), the predict power ranking is almost the same.
並列關鍵字
bankruptcy參考文獻
1. Yesilyaprak, Ara (2004), “Bond Ratings with Artificial Neural Networks and Econometric Models”, American Business Review
2. Edward I. Altman (1968), “Financial Ratios, Disccrimimamt Analysis and the Prediction of Corporate Bankruptcy ”, Journal of Finance
6. Leland Hayne E. (2002), “ Predictions of Expected Default Frequencies in Structural Models of Debt”, Working paper
7. Merxe Tudela, Garry Young (2003), “A Merton-Model Approach to Assessing the Default Risk of UK Public Companies”, Bank of England working papers with number 194
9. Peter J. Crosbie, Jeffery R. Bohn, ”Modeling Default Risk”, KMV