求清华博士论文一部-经管之家官网！

Mean-CVaR 模型与商业银行破产风险管理
http://www.yidu.edu.cn/educhina/ShowPaper.do?mid=479323&svalue=%22CVaR%22&ssort=2&sscope=0&skey=0&hase=0&stype=3
http://res4.nlc.gov.cn/home/pdfRead.trs?marcid=MDAzMzE3ODQ4&channelid=3&bookid=01608151_71&dataid=drpaper&currentpage=1&pathinfo=742_200&top=a&type=first&subchannel=0&ifPdfReader=0&jumpType=0&filenameSize=5&part=0

Mean-CVaR 模型与商业银行破产风险管理
【外文题目】：	Mean-CVaR Model and the Bankrupt Risk Management of Commercial Banks
【答辩日期】：	2006	【页数】：	未知
【作者】：	潘霁	【学院名称及院系】：	清华大学/清华大学
【作者专业】：	清华大学	【学科分类】：
【导师】：	李子奈	【导师单位】：	清华大学
【学位级别】：	博士	【保密级别】：	公开
【关键词】：	破产风险; 资本储备; CVaR; 绩效研究; bankrupt risk; capital reserve; performance research;	【语种/类型】：	汉语
【中文摘要】：	本文以银行整体的破产风险（经济资本）为视角，从信用风险管理工具出发，通过数量经济模型系统地分析“规制、竞争、风险和收益”之间的关系。具体而言就是在风险管理技术CVaR（条件在险值）基础上提出Mean-CVaR（均值-条件在险值）模型，并通过这一模型力图分析在给定规制环境中，商业银行如何配置自身的破产风险和资产（或负债）以及银行之间的互动会对银行带来怎样的影响。本文通过严谨地数学证明，一定条件下Mean-CVaR模型秉承了CVaR模型解的良好性质；而且线性和非线性两种模型在满足一定条件时都可以有广泛的理论研究和实证分析价值。基于可获美国实际商业银行业的数据和计算机模拟，在对Mean-CVaR模型数值解进行分析的基础上，我们得到如下主要结论：（1）Mean-CVaR模型比不考虑收益约束的CVaR框架在破产风险和收益的配置上更加有效；（2）Mean-CVaR模型对破产风险和收益的配置结果对数据的长短非常敏感；（3）Mean-CVaR模型框架下，风险资本储备和风险经营收益不是正向对应的，他们之间的关系取决于银行的经营目标；（4）Mean-CVaR模型框架下的寡头垄断银行产业存在最低的资本规模要求和最优的资本规模；同时竞争会显著影响银行风险经营的效率并且大银行由于拥有更高的退出壁垒从而具有战略优势。（5）Mean-CVaR模型可以看作是一个系统的风险管理框架，其他基础性的风险管理措施都可以被整合到其中。在银行实务中，采用线性Mean-CVaR模型对资产（负债）和资本进行配置时，关键的问题就是如何按照实际资产收益率（成本率）对资产（负债）进行细分。利用Mean-CVaR模型，我们还提出风险调节的资本收益率以及边际资本收益率来测算和比较考虑了资本规模影响后的银行风险经营效率。在银行绩效实证研究领域，Mean-CVaR模型为实证模型识别出重要的变量：权益资本和竞争环境；同时经验研究中融入经济资本不仅可以克服数据偏误，而且还能够提高模型设定的准确性以及合理程度。最后也必须明确，对于非线性Mean-CVaR模型而言，如果一些条件无法满足，则此框架也无法提供一个合适的参照系.
【外文摘要】：	From the viewpoint of the bankrupt risk of a typical bank, this dissertation studied the relationships among regulation, competition or interaction, risk and revenue in a systematic method and quantitative economic model on the base of some credit risk management tools. In fact, we developed Mean-CVaR model from the new tail risk management tool, CVaR, and tried to do some research on how a bank allocates its economic capital and assets(or debts) and the influence on banks from their interactions under the given regulation environment. We proved that the model of Mean-CVaR not only has some good properties as CVaR but also can be widely used in theoretical and empirical research on the form of linear or non-linear model under some regular conditions. On the available data of American commercial banks and computer simulation, we drove some important conclusions from the numberical results of Mean-CVaR model. Some main conclusions were described as following: (1) the result of the bankrupt risk and revenue allocation from the model of Mean-CVaR can be more efficient than the model of CVaR; (2) the result of bankrupt risk and revenue allocation from the model of Mean-CVaR is highly sensitive to the length of sample data; (3) the demand for risk capital reserve doesn’t have positive relationship with the revenue from risk operations and the relationship between them de facto depends on the objection of each bank operation; (4) there are minimal and optimal demands for capital scale in the oligopoly banking under Mean-CVaR model. At the same time, the competition among banks will make a significant effect on the efficiency of bank risk operations and those big banks have a special strategic advantage because of higher exit barrier than small banks; (5) Mean-CVaR model can be seemed as a systematic risk management framework, into which other basic risk management tools could be included. If we put Mean-CVaR model in bank daily operations to get the best allocation of assets (debts) and economic capital, the key problem is to how to classify and re-construct assets (debts) by the real rate of return on assets (rate of cost on debts). We also put forward risk-adjusted rate of return on capital and risk-adjusted marginal rate of return on capital by the model of Mean-CVaR to measure and compare the efficiency of bank risk operation after taking account of capital scale. In the field of empirical research on bank performance, Mean-CVaR not only specifies two important variables: capital scale and competition environment, but also overcomes the selection bias of sample data and improves the accuracy and rationality of model specification. Finally, we must point out if some regular conditions can’t be satisfied, Mean-CVaR model won’t be a good benchmark for the allocation of bankrupt risk capital reserve and revenue.