本研究之目的在於針對KMV模型對於違約點之設定，並檢定此違約點是否適合於台灣上市、櫃公司的違約預測。而最適違約點的預測在文獻上非常少，因此，本研究提出以基因演算法（Genetic Algorithms；簡稱GA.）為基礎的KMV模型求解最適違約點。在實證上，我們比較了GA-KMV及KMV模型，結果顯示，前者無論在樣本內或樣本外、全產業、電子業或非電子業，其績效均優於KMV模型。我們進一步使用ROC曲線比較，結果還是GA-KMV優於KMV模型，這結果說明了GA-KMV有較高的適合度，此違約點適合應用台灣地區的上市、櫃公司，有助於我們預測違約機率和銀行授信風險管理的建構。 The purpose of this article is to investigate the optimal default point of Moody’s KMV model. We will test whether the default point is suitable for Taiwan’s list and OTC Companies or not and propose a new method based on genetic algorithms to be able to solve the optimal default point of KMV model. In empirical study, we have compared the GA-KMV with the KMV model. Results demonstrated that the percentage of correct either in-sample or out-sample, full industries, electronic industries or non-electronic industries, the GA-KMV model seems to be better than the KMV model. In order to find these results, we further to use the ROC curve to test these two models. Similar results show that the GA-KMV model can outperforms the KMV model which means that the GA-KMV model is a best-fit. In consequence, we obtained the optimal default point of Taiwan’s list and OTC companies. This will help us predict the optimal default point and improve the performance of risk management of banks.
