基于证券交易信息的债券市场信用风险研究

发布时间：2018-11-15 08:16

【摘要】：近年来中国债券市场经历了快速的发展,但是信用风险监管的步伐却没有跟上。从2014年起国内信用债违约事件开始增加,2016年国内全年违约债79支,违约金额高达403亿元。从国内债券市场信用风险监管及预警机制来看,评级不客观,跟踪不及时以及评级方法落后的现象普遍存在。因此充分结合大数据等科技的发展,研究适合国内债券市场的信用风险监管方法迫在眉睫。在对国内债券市场的发展状况进行梳理和国内信用评级现状进行分析的基础上,本文结合国内外理论研究成果、国内市场可用信息、大数据挖掘及自动化技术,采用证券交易信息,构建信用风险度量和预警模型。基于债券交易信息和债券基础属性,构建DS模型。根据计算所得风险中性违约概率转换的信用分,度量和预测债券信用风险。基于上市公司发行主体的财务和股票交易信息,结合BS期权定价理论,构建KMV模型。根据计算所得违约距离,度量和预测上市公司发行主体的信用风险。本文解决了由于信息不完全造成的信用评级不客观甚至无评级信息的问题,打破了传统信用评级中所需财务信息不连续和时间滞后的局限。债券市场专业信用评级机构下调债券或者发行主体的信用级别后,会引起债券到期收益率的显著提高。但是在债券信用级别下调之前,一些债券的到期收益率已明显提高。部分投资者对债券信用风险的察觉早于专业信用评级机构。基于债券交易信息构建的DS模型在国内债券市场是有效的。DS模型对于信用状况恶化的债券,具有显著的预警功能。DS模型对信用状况趋好的债券的预测能力要弱于对信用状况恶化债券的预测能力。但是国外评级机构常用的经典KMV模型却在国内市场失效。根据KMV计算所得的发债主体违约距离不服从正态分布,所以不可以直接转换成违约概率。当模型参数变化时,违约距离随着发债主体信用状况的分布情况会发生巨大的变化。整体来看,随着发行主体信用状况级别变化,违约距离期望值的变化具有一定的规律性。当发行主体信用级别处于A层及其以上时,发债主体信用状况越好,违约距离越小。但是当发债主体信用状况处于A层及其以下时,发债主体信用状况越好,违约距离越大。
[Abstract]:China's bond market has experienced rapid development in recent years, but the pace of credit risk regulation has not kept pace. The number of defaults on domestic credit bonds has increased since 2014, with 79 defaults in 2016, totaling 40.3 billion yuan. Judging from the credit risk supervision and early warning mechanism of domestic bond market, the phenomenon of rating is not objective, tracking is not timely and the method of rating is backward. Therefore, with the development of big data and other science and technology, it is urgent to study the credit risk supervision method suitable for domestic bond market. On the basis of combing the development of domestic bond market and analyzing the present situation of domestic credit rating, this paper combines the theoretical research results at home and abroad, the information available in the domestic market, big data mining and automation technology. The model of credit risk measurement and early warning is constructed by using securities trading information. Based on bond trading information and bond basic properties, the DS model is constructed. The credit risk of bond is measured and forecasted according to the credit score of risk neutral probability conversion. Based on the financial and stock trading information of the issuers of listed companies, the KMV model is constructed based on the BS option pricing theory. According to the calculated default distance, the credit risk of the issuer of listed companies is measured and forecasted. This paper solves the problem that credit rating is not objective or even without rating information caused by incomplete information, and breaks the limitation of discontinuity of financial information and time lag in traditional credit rating. The bond market professional credit rating agencies reduce the credit rating of the bond or issuer, which will lead to a significant increase in bond maturity yield. But before the downgrade, some bonds had significantly higher yields on maturities. Some investors perceived bond credit risk earlier than professional credit rating agencies. The DS model based on bond trading information is effective in the domestic bond market. The prediction ability of DS model for bonds with better credit status is weaker than that for bonds with deteriorating credit status. However, the classical KMV model commonly used by foreign rating agencies fails in the domestic market. The default distance of the issuer calculated by KMV is not from normal distribution, so it can not be directly converted into default probability. When the model parameters change, the distance of default will change greatly with the distribution of the credit status of the issuer. As a whole, with the credit status of the issuer changing, the variation of the distance from default to expectation has certain regularity. When the credit grade of the issuer is above the A level, the better the credit condition of the issuer is, the smaller the default distance is. However, when the credit status of the issuer is in the A level and below, the better the credit condition of the issuer is, the greater the distance of default is.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F832.51

【参考文献】