Measuring Operational Risk Based On The Bavesian-Buhlmann Methods

In recent years,due to the operational risk loss events occur frequently and make a huge amountof loss in the field of domestic and foreign commercial banks,making the operational risk becomeone of the main risks facing the banking industry.Following the2004New Basel Capital Accordtake operational risk into the framework of the provision for capital grant,international and domesticBanks pay more attention of the operating risk measurement and management. Operational riskmeasurement is the basis for operational risk management, only accurately measure of operatingrisk can be effective risk management. At present, the banking sector's credit risk and market riskmeasurement methods and management has reached a relatively mature stage, the operational riskmeasurement and management has just started.BaselⅡ Accord put forward Basic IndicatorApproach, Standardized Approach, Advanced Measurement Approach. the Advanced MeasurementApproach mainly include Internal Measurement Approaches,Loss Distribution Approach,ExtremeValue Theory,Scorecard Approach,etc. However, due to the lack of China 's commercial banksoperational risk loss data accumulated, resulting in the application of these advanced measurementmethod may result in deviations of economic capital allocation, which led to reduced ability toprevent and control risks. Solve these problems, this paper introduces a Bayesian estimation ofreliability theory, in order to provide a way for the improvement of operational risk measurement.At present, the common operation risk evaluation model is loss distribution method, but thismethod require a higher loss of data,in particular the determination of loss frequency and lossseverity distribution parameters.This article is to solve the lack of loss data,estimate the operationalrisk losses,,and further make bank to reasonably configuration operation risk capital. This paperwith the four major state-owned Banks overall as the research object, collecting the1994-2009years to operation risk loss data as sample, and of the state-owned Banks internal fraud losses totallosses events do empirical analysis.This article through the two aspects to solve the problem ofinsufficient data in operation risk measure.First, the introduction of Bayesian inference to estimatethe parameters in the distribution of loss frequency and loss, combined with Monte Carlosimulation VaR technique to estimate the operational risk of internal fraud, and secondly, in theconfiguration of the operational risk capital, the introduction of the Buhlmann credibility models,mixed internal data (state-owned banks 'own data), and external data (data of non-state-ownedbanks), by increasing the sample data to accurately estimate and reduce operational risk economiccapital, to further improve the liquidity of commercial banks' capital.