【作者】 程功；

【导师】 张维；

【作者基本信息】 天津大学 ， 管理科学与工程， 2007， 博士

【摘要】 本文研究了如何将信用风险结构化模型基本原理与国内银行管理实践相结合的问题。现有的结构化模型主要是针对发达国家债券投资者建立的,与之相比,国内银行面临的信用风险本质上虽然一致,但所处的市场环境和信息环境却有很大的不同。因此,建立符合国内特点的结构化理论模型是需要解决的首要问题。在实践中,国内银行在信用风险定价和风险限额测算方面采用的技术较为落后,难以满足业务发展和风险管理的需要,迫切需要先进的技术解决方案。本文首先从概念框架、模型假设和基本模型三个方面建立了结构化模型基本分析框架,提出了具有一般性的违约概率基本模型(以下简称“基本模型”)。研究表明:结构化模型符合内部评级法的基本原理,可以作为银行信用风险度量与管理的工具。然后,本文根据国内银行面临的信息环境和信息获取机制,提出了一种新的信息噪音假设,建立了具有中国特点的结构化违约概率模型(以下简称“噪音模型”),并利用国内银行的实际数据进行了实证分析。结果表明:噪音模型预测的准确度高于基本模型和Z′评分模型;国内企业向银行提供的财务报表信息失真的现象较为严重,企业倾向于向银行夸大其信用实力。接下来,本文在结构化模型的基础上,利用解析分析和情景分析相结合的方法,全面分析了债务期限对边际违约概率和边际违约概率曲线的影响。分析发现:近期违约概率随企业风险程度的提高而增大,中远期违约概率先增后减;噪音对近期违约概率的影响较大,对中远期违约概率的影响较小。在此基础上,提出了一种新的中长期信用风险定价方法,它更加合理、精确的考虑了债务期限的影响。最后,分析了资本结构(公司债务率)对违约概率的影响,发现:违约概率随公司债务率的增加而单调递增。在此基础上,提出了一种新的客户风险限额测算方法。这种方法综合考虑了客户承贷能力和银行风险偏好,将客户风险限额和违约概率直接“挂钩”,具有科学合理的理论基础以及较强的直观性和操作性。更多还原

【Abstract】 This dissertation analyzed how to apply the basic rules of structural models to the credit risk management practices in Chinese domestic banks. The existing structural models were mainly built for bonds investors in well-developed financial markets. Although the nature of credit risk of the domestic banks is the same as that of foreign bonds investors, the market and information environment in China is quite different. Therefore, the first objective of this work is to build a theoretical structural model adaptive to Chinese market characters. In practice, the domestic banks’techniques used in credit risk pricing and credit limits calculations were out of date and can’t satisfy the banks’market development and risk management. So, the second objective is to bring forward advanced techniques resolvents.Firstly, this paper developed a basic analysis framework of a structural model including the conceptual framework, the model hypothesis and a generalized traditional model (called‘basic model’) to calculate probability of default (PD). It showed that the structural model is consistent with the basic rules of internal rating-based approaches and is applicable to the banks’credit risk management.Secondly, a specified structural model of probability of default (called‘noisy model’) was brought forward which had a new noisy information hypothesis consistent with Chinese loan markets’situation. The hypotheses were based on domestic banks’information environment and their ways to acquire infomation. The following is an empirical analysis based on Chinese banks’practical data. The result showed that: the predictions power of noisy model was better than those of basic model and Z’score model; the information distortion of companies’financial reports to the banks were seriously and the companies prefer to overstating their credit quality.Then, on the base of the structural model, the impacts of debt maturity on marginal probability of default and its curve were systematically analyzed through analytical approaches and scenario approaches. We found that: when the companies’risk increase, the short-term PD increases accordingly, but the medium and long-term PD increase first and then decrease; while the impacts of noise on short-term PD was very obviously, its impacts on medium and long term PD were mild. According to these findings, a new pricing method of medium-long term credit risk was derived, which reflects the impact of the debt maturity more precisely.Finally, the impact of capital structure on PD was analyzed and it is found out that: the PD increases in line with the componies’debt rate increase. Then, a new calculating approach of credit limit was presented, which considered both the companies’debt capacity and banks’risk attitude and linked the credit limits with the PD directly. It has a sounder theoretical foundation and is more intuitionistic and operational.更多还原