Research On Operational Risk Measurement Of Bank

Over the past few decades, the financial crisis happened frequently. Scholars deeply research operational mechanism in the financial markets while also strengthened financial supervision. Modern financial theory is made of Financial Risk Measurement theory, Portfolio theory and Asset pricing theory. Turmoil in the financial markets occurs frequently, and the impact of financial cases was increasing significantly. These are made of the financial risk management challenges; there is an urgent call for a more suitable risk model to deal with these situations.Evidence shows that financial assets mostly have "thick tail distribution" feature. The traditional assumptions of normal distribution model would be underestimating the risks. In order to more accurately measure the risk that some scholars have used extreme value theory to the bank's operational risk measurement. The extreme value theory (EVT) was widespread applied in engineering. There is not any assumption on the distribution of losses about extreme value theory. The "thick tail distribution" could be fitted by actual data only. So, EVT is suitable to measure excess loss for banks. But there are many of high-frequency /low-severity events caused by the careless staff's personal reasons. One side, if use EVT to measure bank's risk only, we maybe miss high-frequency /low-severity events; on the other hand, if use only Delta-normal model, we could not get good fitting on "thick tail". Therefore, we use the threshold to filter operational losses, and then we use Delta-method and EVT respectively in the two dimensions of operational risk.This paper sums up the definition; characteristics and classification on operational risk. Then, I compare existing frameworks of operation risk measurement summarily; and introduce a reasonable operational risk measurement framework. The framework is based on business processes, including measurement method. In the modeling process, the paper quotes Bayesian decision theory. It makes me obtain a good operational risk measurement model. In this paper, we assume that the high-frequency /low-severity losses obey normal distribution. Also, we assume that the low-frequency high-severity loss obeys generalized Pareto distribution, and the frequency of large losses obeys Poisson distribution. So we obtain the excess loss distribution by Monte Carlo simulation. By using threshold, we combined operating losses and excess loss distribution, and ultimately calculated the value at operational risk of "A" bank. In the last article, the author lay out complete operational risk measurement example, "A" bank's Empirical Analysis, with Delta-EVT model.