Measurement Of Commercial Bank Operational Risk Based On Bootstrap Transform Kernel Density

Since the 1990s,major operational risk events have occurred frequently in domestic and international commercial banks.These cases have caused huge losses to financial institutions and have also caused major damage their reputations,causing serious negative impacts on the bank's steady operations.The operational risk has begun to receive widespread attention.In June 2004,the "New Basel Capital Accord"was published,which incorporated operational risks into the risk management framework and required commercial banks to allocate corresponding capital for operational risks.In order to improve the effectiveness of commercial bank operational risk management,the "New Basel Capital Accord" proposes three methods,the basic indicator method,the standard method,and the advanced measurement method,to measure the capital of operational risk.The advanced measurement method mainly includes the loss distribution method,extreme value theory,scorecards,etc,and encourage commercial banks to develop advanced measurement methods to measure operational risks in light of their own circumstances.However,due to the late start of internal commercial bank data collection,and the fact that even some banks have not yet put on the agenda,this has led to few research on the operational risk measurement of commercial banks in China,and the measurement of operational risk in China is still at an exploratory stage.Therefore,the Chinese banking industry should speed up the pace of researching on operational measurement and management,and improve China's banking industry's resistance to risk.The loss distribution method is currently the most sophisticated,most accurate,and risk-sensitive method among the accepted advanced measurement methods.However,this method requires a higher amount of data because of the low-frequency high-loss operation risk loss caused by domestic and foreign commercial banks.The lack of event data has brought great difficulties to modeling.Therefore,this article will take the entire Chinese banking industry as a whole and collect operational risk loss events from 2005 to 2015 as sample data to conduct empirical research,which can greatly increase the sample data volume;in order to better fit the operational risk loss intensity Due to the excessive assumptions of the intensity distribution in the parametric method,we first studied the numerical simulation and empirical research of the semi-parametric method of transforming nuclear density estimation in bank operation risk.Bootstrap Transform Kernel Density Method This new semi-parametric method replaces parametric methods to fit loss intensity distributions.In the article,numerical simulations are used to compare several non-parametric and parametric methods.Comparisons include classical kernel density estimation,extraction Nucleus density estimation,kernel density estimation and four different parametric models.KS test results and image comparison show that the Bootstrap transform kernel density method has obvious advantages;then the Poisson distribution method is used to fit the loss frequency and KS test results show that the Poisson distribution is more accurate.After fitting the loss distribution and loss strength,we used the ARM algorithm to improve the traditional Monte Carlo method to calculate and compare the VaR values under different risk models under each quantile.In summary,this paper innovatively proposes a loss intensity estimation model based on the bootstrap kernel density method.It performs well on both numerical simulations and empirical data on traditional parameter models and semi-parametric models.Based on this method,we input the past several years of bank risk data,the predicted VaR values were calculated.In the calculation of VaR values,we added an ARM algorithm to accelerate the algorithm and achieved good results.However,there are still some deficiencies in this article.Due to the limitations of the data itself,we only consider all risks as one whole category for analysis.We fail to specifically classify the types of risks and the relevance of risks in accordance with the Basel Accord.To find a suitable internal data set,this paper only considers the external data of bank operation risk,and the VaR value obtained only for external data is often not accurate enough.