Stress testing at major financial institutions: survey results and practice

CGFS Papers  |  No 24  | 
10 January 2005

 Introduction

In May 2004, the Committee on the Global Financial System (CGFS) initiated an exercise on stress tests undertaken by banks and securities firms. The exercise had two main aims. The first was to conduct a review of what financial institutions perceived to be the main risk scenarios for them at that time, based on the type of enterprise-wide stress tests that they were running. The second aim was to explore some of the more structural aspects of stress testing and examine how practices had evolved, particularly over the period since the previous CGFS survey, the results of which were published in A survey of stress tests and current practice at major financial institutions in April 2001.

There were two main parts to the latest exercise. The first part involved a survey of stress tests being conducted at banks and securities firms. The survey asked respondents to list details of the stress test scenarios and associated risk factors that were in use as at the end of May 2004. Sixty four banks and securities firms from 16 countries participated in the survey, with the reporting institutions selected by their national central banks.2 Firms participating in the survey reported to their national central bank; the data were then submitted on a no-name basis to the BIS-based CGFS secretariat and entered into a database. Around 960 stress tests were reported and more than 5,000 risk factors were listed.3 Reflecting a desire to focus on the range of scenarios being employed and confidentiality concerns, survey respondents were not asked to report the results of any scenario runs.

In the second stage, national central banks conducted follow-up meetings with institutions that had participated in the stress test survey. National central banks met to discuss the results of the survey and these interviews. As part of this meeting, senior risk managers from several large complex financial firms were invited to discuss stress test practice. Both the follow-up meetings and the group discussion with risk managers made clear that risk measurement and the role of stress testing in risk management vary widely across firms, reflecting differences in both the complexity of risks faced by firms and the breadth and scale of the different businesses.

The output of the group is a synthesis of observations based on the survey, interviews with respondent firms and discussion with market participants. The exercise illustrated the wide range of practices and risk management frameworks at firms. This reflected, inter alia, the heterogeneous business models that are being employed by firms. The use of stress tests has expanded from the exploration of exceptional but plausible events, to encompass a range of applications. It has met with wider acceptance within firms because it is a flexible tool which can adapt quickly and efficiently to the changing environment and specific needs of a firm and provide important information on the risk exposures of firms. Notwithstanding this positive development, a number of challenges remain, most notably in the areas of stress testing credit risks, integrated stress testing and the treatment of market liquidity in stress situations.