商业银行操作风险与系统性风险度量研究

发布时间：2018-06-07 06:34

本文选题：商业银行 + 风险度量　；参考：《中国科学技术大学》2012年博士论文

【摘要】：本文从商业银行风险管理的实际需求出发,从精细化风险管理的角度着手,采用风险度量的技术对风险进行建模、估计和分析。研究的风险度量主体不是传统商业银行的信用风险和市场风险,而是新资本协议要求必须分配资本金的操作风险和第三版巴塞尔资本协议关注的银行系统性风险。本文研究的主要内容和创新点包括以下几个方面：首先,本文对风险度量的理论和方法进行了抽象、归纳和总结,给出了风险度量的流程和思路。其包括了基础、关键、实现和优化四个环节,并将风险建模细分为八个步骤,该风险度量思路的提出是对以往项目经历和论文写作经验的总结和凝练,为风险度量提供理论基础和指导方法。其次,对商业银行操作风险的度量及其资本金分配进行了研究。对影响操作风险度量结果的损失厚尾性、计算精度问题、模型稳定性问题、数据有偏性问题、数据陈旧性问题、操作风险的相关性问题进行了研究： (1)针对操作风险损失厚尾性的特点和一阶近似算法准确性较差的问题,提出采用二阶近似的算法计算操作风险及其资本金。根据次指数分布的特点推导出基于二阶近似的操作风险解析解,针对厚尾性的问题采用广义帕累托分布拟合极值损失,并在此基础上得到操作风险。并将结果与一阶近似模型、均值修正模型和模拟结果进行了对比分析,发现二阶近似模型提高了结果的计算精度。 (2)针对操作风险参数方法结果的不稳定性问题,提出了操作风险度量的组合估计模型。组合估计模型把贝叶斯的思想引进到操作风险的估计模型中,将风险估计分为两个层次：单个模型的估计和估计的信念。组合估计方法中估计结果是模型结果的线性函数,该方法集成不同厚尾性分布的特点、同时增强模型结果的稳定性和鲁棒性。 (3)针对数据有偏性与陈旧性问题、操作风险的相关性问题,提出了基于条件分布—广义线性模型—copula函数的多维集成模型来度量操作风险。该模型在损失分布的框架下,采用条件分布对有偏的左截尾数据进行建模解决数据有偏性问题；通过建立操作风险与宏观经济和银行自身变量的稳定关系来预测未来操作风险的发生情况,来解决数据陈旧的问题；通过copula函数来对操作风险之间的相关性进行建模,解决操作风险的相关性问题；并综合这些方法建立了基于该集成方法的模拟程序。再次,对银行系统性风险度量进行了研究。在分析银行系统性风险机理和特征的基础上,对系统性风险度量中考虑银行关联性问题、风险相关性问题、风险的亲周期问题、流动性风险的内生性问题： (1)基于银行之间关联性的视角,采用银行间市场的网络分析方法度量了银行系统性风险。研究内容包括了：银行业系统性风险的机理分析,其包括了银行间市场和外部市场两个方面,为度量研究工作奠定了基础；银行间市场和风险传染的模型实例研究,通过实例说明银行市场是系统性风险的重要来源,论证了采用该方法的可行性；银行间市场的网络结构的矩阵表达和传染过程分析的数学表达；基于最大熵方法的银行间资产负债矩阵的求解；银行间市场模式对系统性风险的影响,发现银行间市场结构对银行系统性风险的影响较大；然后,设计三种特定银行间市场模式下的矩阵求解算法,使得求解结果更加符合实际银行市场间结构；最后,采用中国银行业的数据进行了实证分析,发现在次贷危机转变为全球金融危机的过程中中国银行业系统性风险也出现了增大的趋势。 (2)提出了基于自上而下的系统性风险的度量方法。基于银行系统性风险是银行面临的整体风险的视角,在研究风险之间相关性表达基础上,采用自上而下方法度量银行系统性风险。在考虑信用风险、市场风险和操作风险相关性的基础上,分别采用方差／协方差方法和copula函数方法利用相关性矩阵和相依函数刻画相关性的基础上来集成多种风险,最后得到银行系统性风险。自上而下方法基于实际的各种风险的分布或者风险值,比较符合目前中国商业管理各种风险分开度量和管理的实际,容易在银行业进行运用和推广 (3)提出了自下而上的系统性风险的度量方法。根据银行系统性风险的机理,考虑银行间市场与外部市场两种市场,综合了本次金融危机暴露的系统性风险的时间维度上亲周期性、空间维度上的传染性和流动性风险的重要性特征,研究基于情景分析的至下而上方法度量银行系统性风险。该模型考虑了信用风险、市场风险、传染风险和流动性风险。对于信用风险的亲周期性的特征通过建立违约概率与宏观经济变量和金融变量的关系来对信用风险违约率进行建模,在设定情景的基础上得到不同情境下的信用风险违约率。然后结合信用风险的违约损失率和风险暴露因子,采用CreditRisk+模型得到信用风险损失。对于市场风险,在与信用风险同样的情景下,分析不同情景下市场风险因子对表内外风险暴露价值的影响。对于传染风险主要是将网络分析方法应运到综合模型中。信用风险、市场风险、传染风险是清算风险,银行不存在清算风险并不代表银行具有稳定性,流动性风险是银行面临的又一重要的风险,因此在模型中将银行融资风险纳入到整体框架中来,而且融资流动性是银行信用风险的内生风险。
[Abstract]:Starting from the actual demand of the risk management of commercial banks, this paper sets out from the perspective of fine risk management and uses the technology of risk measurement to model, estimate and analyze the risk. The subject of the risk measurement is not the credit risk and the market risk of the traditional commercial banks, but the operation of the new capital agreement that requires the allocation of capital. Risks and Banking Systemic Risks in the third edition of the Basel Capital Accord.
The main contents and innovations of this paper include the following aspects:
Firstly, the theory and method of risk measurement are abstracted, summed up and summarized, and the process and ideas of risk measurement are given. It includes four links, including basic, key, realization and optimization, and subdivides the risk modeling into eight steps. The proposed risk measurement idea is the summary of past project experience and paper writing experience. Concise and provide theoretical basis and guidance for risk measurement.
Secondly, the measurement of operational risk and the capital allocation of commercial banks are studied. The heavy tailing of the loss of operational risk, the problem of calculation accuracy, the stability of the model, the bias of data, the obsolete data and the relevance of operation risk are studied.
(1) in view of the characteristics of heavy tailing of operational risk loss and the poor accuracy of the first order approximation algorithm, the operation risk and its capital are calculated by the two order approximation algorithm. According to the characteristics of the sub exponential distribution, an analytical solution of operation risk based on the two order approximation is derived. The generalized Pareto distribution fitting is adopted for the thick tail problem. The operation risk is obtained on the basis of the extreme value loss, and the results are compared with the first order approximation model, the mean correction model and the simulation results. It is found that the two order approximation model improves the calculation precision of the result.
(2) in view of the instability of the results of the operational risk parameter method, a combined estimation model of operational risk measurement is proposed. The combined estimation model introduces Bias's thought into the estimation model of operational risk, and divides the risk estimates into two levels: the estimation and estimation of a single model. The estimated results in the combined estimation method are the results. It is a linear function of the model result, which integrates the characteristics of different thick tail distributions, and enhances the stability and robustness of the model results.
(3) aiming at the problem of data biased and obsolete and the relevance of operational risk, a multidimensional integration model based on conditional distribution generalized linear model copula function is proposed to measure operational risk. Under the framework of loss distribution, the model uses conditional distribution to model the biased left truncated data to solve data biased questions. By establishing a stable relationship between operational risk and macroeconomic and bank's own variables to predict the occurrence of operational risk in the future, to solve the problem of data obsolescence, to model the correlation between operational risks by copula function, and to solve the related problems of operational risk; and the integration of these methods is built on the basis of these methods. The simulation program of the integrated method.
Thirdly, the systematic risk measurement of bank is studied. On the basis of analyzing the mechanism and characteristics of the systemic risk of the bank, the problem of bank relevance, risk relevance, the relative cycle of risk, the endogenous question of liquidity risk are considered in the systematic risk measurement.
(1) based on the perspective of interbank relevance, the systematic risk of the bank is measured by the network analysis method of interbank market. The research contents include: the mechanism analysis of the systemic risk of the banking industry, which includes two aspects of the interbank market and the external market, which lays the foundation for the research work; the interbank market and the risk transmission A case study of the dyed model shows that the bank market is an important source of systemic risk and demonstrates the feasibility of using this method; the matrix expression of the interbank market structure and the mathematical expression of the analysis of the process of infection; the solution of the inter bank asset liability matrix based on the maximum entropy method; the interbank market model The influence of systemic risk is that the inter bank market structure has a great influence on the systemic risk of the bank. Then, the matrix solution algorithm under three specific interbank market models is designed to make the result more consistent with the actual bank market structure. Finally, the data of the Chinese silver industry is empirically analyzed and found in the subprime mortgage. In the process of crisis turning into the global financial crisis, the systemic risk of China's banking industry has also increased.
(2) a measurement method based on top-down systematic risk is proposed. Based on the view of the overall risk facing the bank, the bank systemic risk is based on the study of the correlation expression between the risks and the top-down method to measure the systemic risk of the bank. The variance / covariance method and the copula function method are used to integrate a variety of risks based on the correlation matrix and the dependence function. Finally, the systematic risk of the bank is obtained. The top-down method is based on the actual distribution of various risks or the risk values, which are more consistent with the various risk points of the current Chinese business management. The practice of measurement and management is easy to apply and popularize in the banking industry.
(3) a measurement method of systematic risk from bottom to top is put forward. According to the mechanism of bank systematic risk, two kinds of markets are considered in the interbank market and external market, and the time dimension of systemic risk exposed by this financial crisis, the importance of the contagious and liquidity risk in the spatial dimension, the research base This model takes account of credit risk, market risk, contagious risk and liquidity risk. The characteristics of the pro cyclical characteristics of credit risk are set up by establishing the relationship between default probability and macroeconomic variables and financial variables to establish the default rate of credit risk. On the basis of the situation, we get the default rate of credit risk under different situations, and then combine the default loss rate and risk exposure factor of the credit risk and use the CreditRisk+ model to get the credit risk loss. The effect of the value is mainly to carry the network analysis method into the comprehensive model. The credit risk, the market risk, the contagious risk are the liquidation risk, the bank does not exist the liquidation risk and does not represent the bank's stability. The liquidity risk is another important risk that the bank faces. Therefore, the bank financing risk is in the model. In the overall framework, financing liquidity is the endogenous risk of bank credit risk.
【学位授予单位】：中国科学技术大学
【学位级别】：博士
【学位授予年份】：2012
【分类号】：F832.33;F224

【引证文献】