27 November 2018

BIS: Stress-testing banks – a comparative analysis

This paper reviews stress tests carried out in exercises since the Great Financial Crisis, drawing on the literature and a comparative analysis carried out by the Financial Stability Institute on system-wide stress tests for banks in the euro area, Japan, Switzerland and the United States.

Sector-wide stress tests differ widely in their design and implementation. A comparison of country practices, complemented by a review of the literature, shows that authorities design stress tests in different ways, with some employing more than one type of test. To better understand these differences and their drivers, the Financial Stability Institute (FSI) conducted a comparative analysis, for which extensive information was gathered from selected authorities across the globe. In particular, the paper covers practices on system-wide stress tests for banks, ie stress tests that cover a significant part of the banking sector in the euro area, Japan, Switzerland and the United States. It presents information in a consistent way across exercises on how these are conducted by central banks and supervisory authorities in these jurisdictions.

Stress tests can have a microprudential or macroprudential policy objective. In a stress test with a microprudential objective, the exercise, albeit system-wide, is focused on assessing the resilience of individual banks, providing authorities with information on whether these banks should take remedial actions (such as increasing regulatory capital, reducing risk exposures or improving their capital planning processes). In turn, stress tests with a macroprudential objective focus on system-wide risks and their aggregate impacts. They may also be used by authorities as an input to calibrate macroprudential measures. In exceptional circumstances, ie at times of systemic financial crises, stress tests can be and have been used to provide information about recapitalisation needs for both individual banks and the banking system, and they can also help restore market confidence.

Stress tests are most effective when their design is fully aligned with the policy objectives associated with them. It is crucially important that authorities make an early decision about why they would like to run a stress test, and how they plan to use the results. For given operational constraints, authorities’ policy objectives are the most effective criterion to select the preferred approach within each of the three above-mentioned building blocks. This also brings internal consistency within the exercise, as all decisions will be driven by the same goal. Consistency with the high-level principles for stress testing recently published by the Basel Committee on Banking Supervision (BCBS (2018)), which also cover objectives, governance, technical implementation and communication, is an important step in this regard. 

However, stress tests are no panacea, and are best used in combination with other tools. Although stress tests are a powerful tool to understand conditions in the banking sector, or of individual banks, the validity of their results is affected by a number of factors, such as data quality and availability, model risk and models’ capacity to capture contagion effects and interlinkages, both within the banking sector and beyond. Moreover, since the results are conditional on assumptions in the methodology and the scenario design, a stress test should not be expected to accurately predict the impact of a specific, forthcoming crisis. It is rather a hypothetical exercise intended to assess the resilience of a bank or the banking sector against various potential shocks. Stress tests are therefore best complemented by other tools available to the authorities to achieve their policy objectives, such as systemic risk monitoring or capital planning reviews. 

Stress testing is being continually improved, and further developments could help to enhance the implementation and the policy use of stress tests. There are several areas where stress tests could be improved, such as, on the implementation side, the joint treatment of solvency and liquidity risks or the specification of second-round, spillover and contagion effects. On the policy side, more authorities could use stress tests as an input to the calibration of macroprudential measures, and stress tests could be further integrated into regular supervisory reviews. Some of these changes will be driven by progress in research or advances in technology, while others will be dependent on gaining enough practical experience, especially in the macroprudential sphere. From a global perspective, a dialogue among relevant authorities regarding a common scenario design for large and cross-border active banks would be a helpful addition to the stress testing landscape.

Full publication

© BIS - Bank for International Settlements