【作者】 刘家鹏；

【导师】 詹原瑞；

【作者基本信息】 天津大学 ， 金融工程与金融管理， 2009， 博士

【摘要】 金融机构的风险管理状况不仅对银行自身的生存与发展至关重要,而且关系到整个社会经济的安全与稳定,当前的金融海啸足以说明这一点。商业银行构建有效的风险管理模型既是企业发展、提高银行核心竞争力的需要,也是监管部门的监管要求,巴塞尔新资本协议明确提出大型的国际化商业银行应逐步建立自己的内部模型。信用组合是商业银行资产的主体,信用风险管理是银行风险管理最重要的内容。然而,由于信用风险的特殊性,信用组合风险建模非常困难:信用风险收益或损失分布是非正态的,不但用正态分布难以拟合,就是用其它的有偏分布也难以较好的近似,违约事件是非连续的、离散的概率事件,加之相关性、违约损失的易变性等要素,最终的损失分布可能非常怪异;信用风险要素本身的复杂性,暴露的不确定性、收益和损失的非线性、回收率的动态性等都使问题变得相当复杂;信用组合资产数目巨大,数学上的多元性,高维积分无法实现;同时也存在信用管理数据的缺失、极端事件数据稀少。基于以上因素,信用组合风险建模用解析方法很难完成,只能使用模拟方法。因此我们构建基于蒙特卡罗模拟技术的商业银行信用组合的度量框架。构建了一个基于蒙特卡洛模拟技术的信贷组合管理模型。模型能够产生组合的损失分布,并计算不同置信水平上的风险值(VaR)和期望短缺(ES)。模型中回收率为随机抽样,并采用了更具灵活性与现实性的双峰分布。模型中采用了多种与实际银行业务相适应的不同信用品质的信贷组合,同时也对贷款承诺等随机暴露形式进行了探讨。模型使用了混合分布抽样,移动窗口等技术,也采用了重要性抽样、低差异序列等现代蒙特卡洛模拟技术。模型既可生产单纯违约造成的损失分布,也可生产考虑信用迁移风险的价值分布,并对不同信用组合进行两种方法计算的风险进行比较。我们也把模型的计算结果和新巴塞尔资本协议中的标准法和初级IRB法计算的资本要求进行了比较。我们发现使用内部模型在高品质信用的条件下可以节约银行资本,实现资本套利;而低信用品质组合做不到这一点,模型计算的经济资本高于根据协议计算的监管资本。这也说明巴塞尔委员会的政策导向,鼓励大型的活跃银行采用内部模型法,同时引导这些银行放贷于高信用品质的信贷客户。引入随机利率和随机信用价差将模型扩展。使模型集成处理信用风险和利率风险。模型的构建同时使用了结构型信用风险模型和简化型信用风险模型。通过的模拟框架对信用组合的相关性、利率对风险的敏感性进行了模拟与分析,相关的结论对资产组合的构建和风险管理都具有重要意义。我们发现,利差风险和利率风险是很重要的因素,尤其是高信用品质的金融工具利率风险和利差风险更加重要,而且在大组合条件下也不能被分散化。在以上模型的基础上构建集成的压力测试框架。提出新的抗压能力度量指标,用来衡量组合抵御极端状况风险的能力。压力测试能够用来测量设定的意外事件发生所导致的风险因素变化给金融机构带来的潜在影响,这对当前金融海啸的经济背景下的我国银行业非常有意义。首先,进行了单一因素的压力测试,分别对信用恶化、违约损失率剧增、相关性明显增强三种压力情境进行了压力测试。接下来,又进行了集成压力测试,同时考虑了上述三种压力条件同时发生的情境下的不同品质信用组合的抗压情况。压力测试框架采用混合分布进行压力测试;使用了马尔科夫机制转换模型;最后讨论了信用评级的评级方法。信用评级是本模型的基础,因此评级的可靠性非常重要,但实际的信用级别又无法观察,因此我们提出了基于蒙特卡罗模拟的评级评级方法,给出了的四个评估评级系统的度量指标,并验证了Bootstrap方法的有效性;最后,将上述方法用于实际模型的评估和改进。更多还原

【Abstract】 Financial institutions risk management of banks not only their own survival and development is critical, and the entire society’s economic security and stability, but before the financial tsunami illustrates this point. Commercial banks to build effective risk management model is the enterprise development, enhance core competitiveness of the banking needs, but also the regulatory requirements of regulatory authorities, the New Basel Capital Accord explicitly put forward the internationalization of large commercial banks should gradually establish its own internal model.Credit portfolio is the main commercial bank assets, credit risk management banking risk management is the most important elements. However, due to the special nature of credit risk, credit portfolio risk modeling is very difficult: gains or losses on the credit risk of non-normal distribution, not only difficult to fit with the normal distribution, that is, with other partial distribution can hardly have a better approximation, and ultimately the loss of the distribution may be very strange; credit risk elements of its own complexity, exposure to uncertainty, gains and losses non-linear, dynamic recovery have become very complex problem; tremendous number of credit portfolio, the diversity of mathematics, high-dimensional integral can not be realized; At the same time, there are credit management data is missing, the extreme scarcity of event data. Based on the above factors, the credit portfolio risk modeling using analytical method very difficult to accomplish, can only use the simulation method. Therefore, we construct Monte Carlo simulation technique based on the commercial bank’s credit portfolio measurement framework.Construct a Monte Carlo simulation technique based model on the risk management of credit portfolio that is difficult to solve because of non-linear, non-normality, diversity factors. Model can produce a simple breach of contract caused the loss distribution, but also can produce consider the value of the credit migration risk distribution, and calculate the different confidence level on the Value-at-Risk (VaR) and expectations of a shortage of (ES). We use a more flexible and realistic bimodal distribution to sample Recovery. Model using a mixed distribution of the sample, and mobile window technology, also used the importance of sampling, low-difference sequence simulation of modern Monte Carlo technology.The model is extended to the stochastic interest rate and stochastic spreads, and credit risk achieves integration of interest rate risk. Under the framework of the integrated portfolio of assets on the relevance, interest rates, asset number of factors are sensitivity analysis, which concluded on portfolio construction and risk management are of great significance.Construct an integrated framework for stress testing based on the model. Develop new metric indicators for compressive capacity, used to measure the portfolio against extreme risks. Stress tests used to measure can set accident caused by changes in risk factors give financial institutions the potential impact of this on the current financial tsunami economic background of China’s banking industry is very meaningful. Mixed distribution of stress tests; set up a variety of situational stress testing framework integration. Stress-testing framework for the use of mixed distribution of stress tests; the use of a Markov switching model mechanism;We proposed a new method to evaluation credit rating system. Credit ratings are the basis of this model, so the reliability of ratings is very important, but the actual level of credit can not be observed, so we have made Monte Carlo simulation based on the ratings of rating methods, give a rating system to assess the four metrics indicators, and verify the validity of the Bootstrap method; Finally, the above method has been applied to the actual model assessment and improvement.更多还原