Measuring Model On Operational Risk Management Of Commercial Bank

In recent years,loss in operational risk incidents happen frequently,and the operational risk is one of the three main risks which the banks face.Measuring operational risk is the basic and precondition of operational risk management. How to use appropriate models and methods to measure operational risk and distill corresponding capital is a key matter.Research on the operational risk management of commercial banks in China is just at the very outset.Therefore the study in this area will contribute to raise the operational risk management level of Chinese commercial banks and enhance the core competitiveness of Chinese commercial banks.However,since Chinese commercial banks do not attach importance to the operational risk management before,China does not really have the historical database for operational risk.This thesis discusses the relationship between LDA and Bayesian approach, constructs a two-stage loss severity distribution function and builds models in both random and fuzzy environment relatively.In random environment,a measuring model based on Bayesian estimation is given.In this model,prior distribution is obtained through conjugate law.When operational risk loss data is given,posterior distribution can be easily calculated. After posterior mean is exploited as Bayesian estimate,loss frequency distribution and severity distribution are gotten.Similarily,a measuring model based on fuzzy point estimation is given in fuzzy environment.In this model,prior membership function could be obtained through three laws.When operational risk loss data is given,posterior membership function can be easily calcaluated by using fuzzy point theorem.After posterior mean is exploited as fuzzy point estimate,loss frequency distribution and severity distribution are gotten.Finally,we simulate the operational risk loss of Chinese commercial bank using Monte Carlo simulation and derive the regulatory capital allotted for operational risks by China banking industry.