Mixing Internal and External Data for Managing Operational Risk

摘要: The Loss Distribution Approach has many appealing features since it is expected to be much more risk - sensitive than any other methods taken into consideration by the last proposals Basel Committee. Thus this approach provide significantly lower capital charges for banks whose track record particularly good relatively their exposures and compared with industry wide benchmarks. Unfortunately LDA when calibrated only on internal data far from being satisfactory a regulatory perspective as could likely underestimate necessary charge. This happens two reasons. First if bank experienced average number of events, will benefit charge even though its happened chance does not result better management practices. As consequence, acceptable long frequency are tempered references. such, immediately raises issue how cope both external paper proposes solution based credibility theory which widely used in insurance tackle analogous problems. result, we show make statistical adjustment temper information conveyed use Similarly calibration severity parameters ignores data, then distribution biased towards low losses typically those recorded databases. Again cannot accepted unless merged database process. Here again regarding best way merge these data. Obviously done without care databases directly fuelled distributions strongly high losses. also comparable one another order permit safe unbiased merging.