登录之后可查看解析

Total risk is composed of frequency and severity. A Poisson distribution is frequently assumed for the distribution of operational risk event frequency. The severity of each event often follows a different distribution, such as a lognormal. These distributions can be combined into a single operational loss distribution in a process known as convolution. Separating losses into likelihood and severity allows the analyst to determine whether operational losses stem from low-frequency, high-severity events or high-frequency, low-severity events and aids diagnosis. If this analysis is repeated for multiple processes in the firm, which is necessary to estimate total operational risk using a bottom-up approach, then the analyst must consider correlations among the different processes.