【作者】 王箭；

【导师】 王怒立；

【作者基本信息】 武汉理工大学 ， 产业经济学， 2008， 博士

【摘要】 随着2006年我国金融市场的全面开放,商业银行面临着前所未有的激烈竞争。自1999年开始的不良贷款资产剥离以来,在各大商业银行完善自身信贷风险控制和外部监管控力度逐步加强基础上,我国银行的信贷资产质量有一定程度的提高。但由于我国银行业的信贷风险控制技术比较落后,在很多方面已经跟不上日益复杂的宏观经济和信贷运营环境的变化。所以对我国商业银行而言,大力加强银行信贷风险控制研究迫在眉睫,尤其是信用风险度量技术的运用和内部控制管理的强化。本文在大量信贷业务风险控制的实践基础上,结合银行风险控制理论,把信贷风险控制系统分为:信贷风险评估、信贷活动风险控制、信贷风险事后监督和处理、环境控制四大子系统,运用全程和全面风险控制的理念对上述每个子系统进行了分析研究。本文所做的主要研究工作如下:(1)为了克服古典和现代信用风险度量方法中需要大量的原始数据且要求满足各种分布形态的局限,建立了短期信用风险指标体系和长期信用风险指标体系,引入了灰色关联度模型来对ST公司和非ST公司进行信用风险度量。实证结果表明,基于灰关联度的信用评估方法对非ST公司得出了良好的关联区分度,而对于ST股票类型得到的关联度区分并不很好,这是因为ST股票公司的某些财务指标出现了极大值或极小值有关,导致了算法的区分度不高。(2)本文对KMV在我国的应用情况进行了研究,分析了我国上市公司的信用状况,计算出我国上市公司的违约距离。针对KMV公司提出的发现违约发生最频繁的分界点在公司价值大约等于流动负债加50%的长期负债,但是,我国上市公司失信状况比较严重,与美国的信用状况不同,所以不能直接套用其违约点。本文设立七种违约点,分别计算出违约距离,然后通过对违约距离进行非参数检验,最终确立一个最适合中国股市的违约点计算模式,且在计算股权的市场价值考虑了非流通股的价值。实证分析表明,ST公司违约距离与非ST公司违约距离差距明显,从而说明KMV模型在我国使用是有效的,并发现对总资产与流动负债、长期负债之间的回归取得的带截距项的回归方程是最适合我国情况的违约点设置。(3)本文通过选取反映短期和长期信用风险两种情形的财务指标,建立了适合短期和长期违约风险监测的两个Logistic回归模型,并使用检验样本对得到的两个Logistic回归模型进行了信贷风险预警效果检验。在此基础上,本文还利用多层次评估的全面性,建立了比Logistic回归模型更加符合客观信贷实践的基于灰色层次评估的信贷风险预警体系,并用实证结果加以论证其有效性。(4)本文在信贷风险的内控制度的建设上,针对我国商业银行在内控制度上的缺陷,提出了基于银行实践的改进措施,如完善贷审会制度和建立风险经理制度等,并根据自身的信贷工作实践从微观信贷管理的视角提出了较好地对信贷风险实施过程控制的对策。更多还原

【Abstract】 With financial market of our country opened completely in 2006, the commercial banks faced fierce unprecedented competition. Since the start of the non-performing loans divestiture in 1999, the quality of banks’credit assets has been improved to a certain extent basing on the improvement of their own credit risk control and the gradual enhancement of exterior supervision and control. But because of relative lag in the credit risk control technology, the banking could not keep up with the variety of increasingly complicated macro economy and credit operation environment in many respects. So it is imminent to energetically reinforce the research on bank credit risk control to our country commercial banks, especially the operation of credit risk measurement technology and the enhancement of inner control management. On the practice of large number of credit business risk control and combining with bank risk control theories, this topic divides credit risk control system into four subsystems: credit risk assessment, credit activity risk control, credit risk supervision and processing afterwards and environment control, and applies complete and global risk control idea to analyze every subsystem above.The main research that this paper does as follows:(1) In order to overcome the limit of classic and modern credit risk measurement methods which need a great deal of original data and request satisfying various distribution appearance, the paper establishes short-term and long-term credit risk targets system, and introduces grey-connection degree model to measure the credit risk of ST and non-ST companies. The positive result indicates that, the credit risk assessment method basing on grey connection degree gets a good connection distinction degree to the non- ST companies, but not very good to the ST companies, because some financial targets of ST companies have maximum or minimum value, which causes the distinction degree of the calculation not high.(2)This subject studies the application of KMV model in our country, analyzes the credit condition of the listed companies and computes their default distance. The rule of thumb KVM Corporation proposed is empirically that default distance equal to the sum of value of short-term debt and fifty percent value of long-term debt. However, differing from American companies, credit condition of the Chinese listed companies is not satisfying. Default point cannot be indiscriminately imitated from it. This paper set seven kinds of default point, calculates each of default distance and examines them using no-parameter test. Finally, the most proper proportion for calculating default distance is gained. And non-circulated share is considered. Empirical analysis shows that the default distance between ST companies and non-ST companies is obviously and that KMV model in China is valid. It also finds that regression model with the intercept which regress total assets, current liabilities and long-term liabilities of the regression is the most effective.(3)This paper builds two Logistic regression models which are suitable to monitor short-term and long-term default risk, and uses test-sample to test credit risk early-warning effect of the two models through selecting financial targets that reflect short-term and long term credit risk. In this foundation, this paper also uses the complete of multilayer estimation to set up a credit risk early-warning system which is based on grey-layer estimation and accords with the objective credit practice much more than Logistic regression model, and makes use of the positive result to demonstrate the validity.Based on the construction of internal control system for credit risk, and combined with the deficiency of internal control system in our commercial banks, this subject proposal improved measures aiming at bank practice on human resources, for instance, improve the system of loan decision-making committee and mechanism of risk manager. At last, according to own credit practice, and from the perspective of micro-credit management, it brings forward a good practice on the implementation of the credit risk control measures.更多还原