基于修正KMV模型的我国商业银行信用风险测度

发布时间：2018-11-15 10:13

【摘要】：作为我国商业银行面临的最主要金融风险类型之一,信用风险对商业银行的整体发展起到至关重要的作用。与西方发达国家相比,我国商业银行的信用风险管理水平明显落后,信用风险管理体系有待完善,特别是风险管理工具及技术与国际上发达国家的商业银行相比存在较大的差距。因此我国商业银行若想稳健发展,并参与激烈的国际金融市场竞争,必须提高风险识别、评估、控制等能力,通过借鉴、吸收并应用国外一些成熟的风险管理技术,进而建立科学的风险识别、检测、度量控制体系,及时对面临的信用风险进行科学度量与有效管理,提高信用风险管理系统在风险控制过程中所发挥的关键效用。首先,本文多角度全面分析我国商业银行信用风险现状、信用风险管理中存在的问题并详细阐述度量在信用风险管理中的重要地位。在比较现代四种信用风险测度模型在我国适用性的基础上,选择KMV模型为基础模型。其次,为提高KMV模型在我国的适用性,对其参数进行修正:针对股票收益率序列尖峰厚尾及杠杆效应的特征,采用EGARCH(1,1)-M模型取代GARCH(1,1)模型计算股权价值波动率;针对我国存在非流通股问题,采用净资产定价法估值;在违约点的设置上,通过计算并检验3种不同违约点下正常公司与违约企业的违约距离是否存在显著性差异,从而选出最优违约点;无风险收益率采用上海同业拆借利率(Shibor)以更好反映利率市场化进程。最后,本文选取2014年商业银行贷款投向位居前5的行业共计14家上市公司财务数据与市场数据进行信用风险测度的实证分析和检验,验证了修正KMV模型在我国信用风险测度中的有效性。最后,本文从度量手段与应用环境两个方面对提高信用风险管理水平提出建议。研究结果表明:(1)运用EGARCH(1,1)-M模型计算股权价值波动率,可以反映出前收益率与未来波动间的杠杆效应,提高了计算精度。(2)通过独立样本T检验,选出最优违约点为DPT=0.75LTD+STD;(3)ST公司的违约距离均值在总体上是小于非ST公司的,这意味着总体上其违约风险也小于ST公司,这与现状相一致。而理论违约概率与现状并不完全相符,即理论违约概率无法有效识别公司的全部信用风险。(4)不同行业的上市公司信用风险水平差异明显,其中制造业上市公司的信用风险最大。(5)对实证结果进行K-S检验、Mann-Whitney U检验,结果表明修正后的KMV模型可以有效区分ST公司与非ST公司的风险水平。在此基础上,进一步利用ROC曲线表示修正后模型预测能力,预测能力准确率达到85.7%。
[Abstract]:As one of the most important types of financial risk faced by commercial banks in China, credit risk plays a vital role in the overall development of commercial banks. Compared with the western developed countries, the credit risk management level of our commercial banks is obviously backward, and the credit risk management system needs to be improved. Especially, there is a big gap between risk management tools and technology compared with commercial banks in developed countries. Therefore, if our commercial banks want to develop steadily and participate in the fierce international financial market competition, we must improve the ability of risk identification, evaluation, control, and absorb and apply some mature risk management techniques from abroad. Then a scientific risk identification, detection, measurement and control system is established to measure and effectively manage the credit risk in time, and to improve the key utility of the credit risk management system in the process of risk control. First of all, this paper comprehensively analyzes the current situation of credit risk of commercial banks in China, the existing problems in credit risk management and elaborates the important position of measurement in credit risk management. On the basis of comparing the applicability of four modern credit risk measurement models in China, KMV model is chosen as the basic model. Secondly, in order to improve the applicability of the KMV model in China, the parameters are revised: according to the characteristics of the sharp and thick tail of stock return series and leverage effect, the EGARCH (1K1) -M model is used to calculate the volatility of equity value instead of the GARCH (1K1) model; In view of the problem of non-tradable shares in China, the net asset pricing method is used to estimate the stock. In the setting of default point, the difference of default distance between normal company and defaulting enterprise under three different default points is calculated and tested, and the optimal default point is selected. The risk-free rate of return adopts Shanghai Interbank offered rate (Shibor) to better reflect the process of interest rate marketization. Finally, this paper analyzes and tests the credit risk measurement of 14 listed companies in the top 5 industries in 2014. The validity of the modified KMV model in China's credit risk measurement is verified. Finally, this paper puts forward some suggestions on how to improve the level of credit risk management from the aspects of measurement and application environment. The results show that: (1) using EGARCH (1) -M model to calculate the volatility of equity value can reflect the leverage effect between forward return and future volatility, and improve the accuracy of calculation. (2) through independent sample T test, The optimal default point is DPT=0.75LTD STD;. (3) the average default distance of ST is smaller than that of non-ST on the whole, which means that the default risk of ST is smaller than that of ST, which is consistent with the present situation. However, the theoretical default probability is not completely consistent with the present situation, that is, the theoretical default probability can not effectively identify the total credit risk of the company. (4) the level of credit risk of listed companies in different industries varies significantly. The credit risk of listed companies in manufacturing industry is the largest. (5) K-S test and Mann-Whitney U test are carried out on the empirical results. The results show that the modified KMV model can effectively distinguish the risk level between ST and non-ST companies. On this basis, the prediction ability of the modified model is further expressed by ROC curve, and the accuracy of prediction is 85.7%.
【学位授予单位】：南京财经大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：F832.33

【相似文献】