This thesis measures operational risk from internal fraud in a Korean bank using Loss Distribution Approach(LDA). Loss distribution approach is the methodology for measuring operational risk, which is emerging, preferred to other methodologies, and almost set into standard methodology among many financial institutions.
We first estimate frequency and severity distribution for internal fraud event in the bank, execute Monte Carlo simulation using the estimation result, and get expected loss and unexpected loss. Secondly, we evaluate sensitivity analysis with varying parameter of poisson distribution to find out the significance of the distribution assumption. Finally, we observe the effect of external data in measuring operational risk, which is used when internal data is not sufficient and the forecast of loss event needs to be improved.
The analysis results show that operational risk measurement using LDA seems to be appropriate for financial institutions that accumulate loss data to some degree. In addition, to make the best use of external data, we need to build some methodology to adjust it according to the characteristics of individual institutions.
