Research On The Internal Rating Validation Results Based On Block-Bootstrap Simulation

In 2004,the Basel Committee on Banking Supervision issued the "International Convergence of Capital Measurement and Capital Standards:A Revised Framework"(referred to as the New Basel Capital Accord).Based on the calculation of risk capital level,using the internal ratings-based approach to measure credit risk.After the New Basel Capital Accord released,many international banks have started to increase the credit risk evaluation model construction,use of internal ratings-based(IRB)for their own level of risk prevention measure.The minimum requirement of using Basel II IRB is validation.Whether it is for the bank's internal credit risk management process,or for the calculation of Basel II regulatory capital required,the bank must ensure that their IRB approach is effective,which requires the effectiveness of its internal rating system validation.Regardless of the use of the new Basel II IRB sample data requirements,or from the perspective of the statistical analysis,the data size,high data quality databases are an important foundation for the bank itself validation of internal ratings.China's commercial banks restructuring work started late,the historical data is not very rich,lacking of standard processes in the collection and management of data,making the sample data of internal rating model validation and subsequent maintenance,particularly related default sample data quantity and quality insufficient.Therefore,at this stage,China's banks need to find an effective way to increased data accumulation,improve data quality,expand the size of the database,which requires not only efforts of the whole banking sector,but also need the support and cooperation of the Government of China Banking Regulatory Commission and other regulatory agencies.From the IRB validation theory,this article introduced the distinguish ability test,prudential validation and stability validation,combined with IRB validation that appear in inadequate sample data quantity and quality issue,use Block-Bootstrap simulation based on original sample data to expand the number of default samples,through it to apply internal ratings validation of distinguish ability test empirical research,prove this method improved internal rating validation results effectiveness,and proposed feasible proposals to improve the quality of internal rating validation of required data.