论文部分内容阅读
本文基于KMV模型对我国上市公司的信用风险进行度量与评估。本文介绍了KMV模型的推导过程并对Merton’s KMV模型的违约点进行了修正,然后选取2006—2015年沪深两市ST处理与非ST处理各25家小型公司为样本进行实证分析。首先,采用GARCH模型对公司的股权收益波动率进行估测并运用最小二乘法估算公司资产价值及其波动率。然后,运用机器学习Random Searching对公司的特殊处理情况进行拟合优化,确定最优违约点,并计算违约距离与预期违约概率。最后,运用独立样本T检验对比Merton’s模型与Moody’s模型计算的违约距离。本文发现,这两种模型对判断公司是否会发生违约的预测均有效;但Merton’s KMV模型计算的违约距离总体过大,即预期违约概率总体过小。
This article measures and evaluates the credit risk of listed companies in our country based on KMV model. This paper introduces the derivation process of KMV model and modifies the default point of Merton’s KMV model. Then, we choose 25 small-sized companies that deal with ST and non-ST in Shanghai and Shenzhen stock markets from 2006 to 2015 as samples. First, the GARCH model is used to estimate the volatility of the company’s equity return and the least squares method to estimate the company’s asset value and its volatility. Then, machine learning Random Searching is used to fit and optimize the special treatment of the company, determine the optimal default point, and calculate the default distance and expected probability of default. Finally, T test was used to test the default distance between Merton’s model and Moody’s model. This paper finds that the two models are effective in predicting whether a company will default or not. However, the default distance calculated by Merton’s KMV model is too large, that is, the expected probability of default is too small.