Skip to main content

The Effect of the Choice of the Loss Severity Distribution and the Parameter Estimation Method on Operational Risk Measurement*1

Analysis Using Sample Data

  • *1This is an English translation of the Japanese original released in June 2007.

December 26, 2007
Financial Systems and Bank Examination Department
Bank of Japan

Click on ron0712c.pdf to download the full text.

Abstract

A number of financial institutions in Japan and overseas employ the loss distribution approach as an operational risk measurement technique. However, as yet, there is no standard practice. There are wide variations, especially in the specifications of the models used, the assumed loss severity distribution and the parameter estimation methods.

In this paper we introduce a series of processes for the measurement of operational risk: estimation of the loss severity distribution: estimation of the loss distribution and assessment of the results. For that purpose, we present an example of operational risk quantification for a sample data set that has the characteristics summarized below.

We use a sample data set extracted and processed from operational risk loss data for Japanese financial institutions. The sample data set is characterized as having 'stronger tail heaviness' than data drawn from a lognormal distribution, which is often used as a loss severity distribution.

By using this data set, we analyzed the effect on risk measurement of assumptions about the loss severity distributions and the effect of the parameter estimation methods used.

We could not find any distribution or parameter estimation method that is generally best suited. However, by analyzing the measurement results, we found that a more reasonable result could be obtained by: 1) estimating the loss severity distribution separately for low-severity and high-severity loss portions; and 2) selecting an appropriate parameter estimation method.

This paper is prepared to present points and issues relating to the measures taken by the Financial Systems and Bank Examination Department of the Bank of Japan.

It only outlines preliminary results for the purpose of inviting comments from the parties concerned and does not necessarily express established views or policies of the Department.

Notice

Please contact below in advance to request permission when reproducing or copying the content of this paper for commercial purposes.

Risk Assessment Section, Financial Systems and Bank Examination Department, Bank of Japan
E-mail: post.fsbe65ra@boj.or.jp
Please credit the source when reproducing or copying the content of this paper.