Measurement Of Operational Risk And Systemic Risk Of Commercial Bank

According to the actual requirements of the commercial bank risk management, this thesis is devoted to the investigation of risk quantification, including risk modeling, estimation and analysis from the perspective of the refinement of risk management. Risk models are beyond those techniques used for measuring the standard risk types market risk and credit risk, therefore, we consider the quantification of operational risk and systemic risk. Operational risk is a risk that bank must allocate capital in Basel Ⅱ, and Basel Ⅲ highlights the importance of reducing systemic risk to achieve the goal of overall financial stability.Firstly, the procedure and the general framework of the risk quantification are given by abstracting and summarizing previous project experience and paper writing experience. This general framework has four linked loops:foundation, key point, realization and optimization, and it can be divided into eight steps. The law and the general framework can provide a theoretical basis and guidance for risk measurement.Secondly, operational risk measurement and capital allocated are researched considering fat tail, calculation accuracy, result instability, data bias, old data, and dependence of operational risk.(1) For operational risk loss fat tail (which in practice they are) and poor accuracy of first-order closed-form approximation algorithm, a second-order closed-form approximation algorithm is proposed to calculate the operational risk and operational risk capital. The second-order closed-form approximation for operational risk Value-at-Risk can be obtained when sub-exponential distribution is used, and in this paper the generalized Pareto distribution (one of sub-exponential distribution) is used. We find that the second-order closed-form approximation is the most accuracy among first-order closed-form approximation, refinement first-order closed-form approximation by mean correction, and second-order closed-form approximation by comparing these results with enough large simulation results.(2) We propose a combination model to estimate operational risk in order to integrate characteristics of different heavy-tailed distributions and to reduce uncertainty of operational risk model. Operational risk estimation is divided into two levels: operational risk estimation by single model and belief of operational risk estimation by single model, which is similar to Bayesian estimation model. Measured variable has the linear relationship with model variables, empirical analysis shows the combinational model performs better than any single model and can even outperform the best single model. So the combination model can integrate characteristics of different heavy-tailed distributions and reduce uncertainty of operational risk.(3) We propose a risk integrated method to resolve problems of data bias, old data, and dependence among operational risk units in operational quantification. We measure operational risk and operational risk capital using a multivariate model under the loss distribution approach framework. Firstly, conditional distribution is used to model left-truncated data. Secondly, general linear model is employed to forecast the frequency of next year’s operational risk. Thirdly, copula functions are employed to model the dependence between the operational risk units. We use the integrated approach to measure operational risk and operational risk capital of Chinese commercial banks by our designed simulation procedures. Empirical analysis shows that the approach allows the allocation of capital in a more efficient way than the standard approach.Thirdly, bank systemic risk measurement is researched considering bank association, risk dependence, risk pro-cyclical and endogeneity of risk based on the mechanism and characteristics of the analysis of bank systemic risk.(1) Network analysis method is empoyed to measure bank systemic risk by using inter-bank market data from the perspective of bank association. This study includes: the mechanism and characteristics analysis of systemic risk, which are foundation of bank systemic risk measurement;"Domino Model" and "Beyond the Domino Model" both demonstrate that inter-bank market is the important channel of systemic risk and network analysis is feasibe method; the asset-debt matrix expression of the network structure of inter-bank market and the mathematical expression of contagion process are researched; the asset-debt matrix is sloved by the maximum entropy method; the influence of inter-bank market structure to systemic risk is researched, and we find inter-bank market structure can severely change systemic risk, and then three specific inter-bank market structure are elaborated, asset-debt matrix solution algorithms are also given in line with the actual banking market structure; finally, we analyse Chinese banking systemic risk using the network analysis approach, and find systemic risk increased when subprime mortgage crisis extended into the global financial crisis.(2) Top-down model is proposed to measure bank systemic risk. From the perspective of risk integration and risk dependence, top-down method is proposed to measure bank systemic risk considering risk denpendence among different risk types and separate risk measurement and management practice in banking. Variance/ covariance method and copula method are respectively used to integrate credit risk, market risk and operational risk to get systemic risk, and risk dependence is expressed by risk correlation matrix and copula function. Top-down method integrates different types risk in high level (for example, risk distribution or risk value), but top-down method does not consider the reason of risk dependence. It can be easily used in banking because it is in line with risk measurement and management practice of dividing bank risk into different type risks, and separately measuring and managing these risks.(3) Bottom-up model is also proposed to measure systemic risk. According to the mechanism of bank systemic risk (inter-bank market and external market), bottom-up method based on scenario analysis is used to measure bank systemic risk considering bank association, risk dependence, risk pro-cyclical, endogeneity and importance of liquidity risk in this crisis. The scenario analysis based bottom-up model takes into account credit risk, market risk, contagion risk and liquidity risk. We distinguish debt that is held between banks from debt held with parties outside of the banking system. While the value of interbank debt is determined in the network analysis, the potential losses from non-interbank debt are captured by a credit risk model. For risk pro-cyclical characteristics, a model translates macroeconomic risk factor changes to default probabilities for different industry sectors. The CreditRisk+model uses default probabilities, loss given default and exposure at default to estimate credit risk in this paper. With respect to market, we construct a mapping from market risk factors to portfolio positions in line with the same credit risk scenarios. We also consider liquidity risk in the model besides credit risk, market risk and contagion risk, and we introduce endogenous funding liquidity into the bottom-up systemic risk assessment framework.