Pavel Shevchenko, Principal Research Scientist from CSIRO CMIS's Financial Risk Management Group, gave a Statistics seminar on 'Modelling operational risk', Oct 18 2006.
Download slides here
A significant part of uncertainty in a firm's earnings is related to the way it operates business in terms of operational risk that cannot be explained by market fluctuations. Under Basel II requirements, the banks should hold a capital against operational risk. The banks are allowed to use internal models for operational risk quantification. However, they should demonstrate the accuracy of internal models within 56 Basel II risk cells (eight business lines times seven risk types). The models should make use of internal data, relevant external data, scenario analysis and factors reflecting business environment and internal control systems. Emerging best practices share a common view that Loss Distribution Approach (based on modelling of frequency and severity of operational risk losses) is the soundest method. Quantification of operational risk is a relatively new area when compared to quantification and understanding of credit and market risks. Creation of meaningful quantitative approach to operational risk management is a difficult task with many challenges emerging. Different opinions on measurement techniques and methodologies are hotly debated. In this talk we address several important aspects of operational risk modelling using Loss Distribution Approach such as: combining expert opinions with historical data; modelling data thresholds; dependence between risks.