Stress testing used to be an obscure corner of a specialised branch of economics and banking. No longer. Urged on by the IMF and others (Hardy and Hesse 2013), it has become a central piece in the banking world. For most of the large global banks in the US and Europe, meeting the standard to pass their annual supervisory stress tests is the binding regulatory constraint for capital and even strategic planning.
To gather views from a wide range of perspectives on stress testing, Vox is today posting a new CEPR Press eBook on the remarkable development of a tool which ten years ago was little known, apart from among a small fraternity of banks’ risk modellers and their supervisory counterparts.
Stress Testing and Macroprudential Regulation: A Transatlantic Assessment
The eBook’s ten chapters were written by policymakers, stress test designers and academics.
The chapters recount how the steps taken by the Federal Reserve System in the US and by the EU have progressively led to the formalisation of periodic macro stress tests of large banks. As detailed by Rochelle Edge and Andreas Lehnert, a decisive step was taken in 2009 when the Fed used stress tests designed around a common macroeconomic scenario to assess the solvency of large US banks with the result that some of the banks were required to make substantial capital increases. This strong step taken at the height of the crisis was subsequently judged to be crucial in restoring confidence in the overall stability of the banking system.
The form of the tests has continued evolve in the US. However, a constant feature is that the vulnerabilities that are brought to light under stressed scenarios are being used by supervisors to require banks to suspend dividends or share buybacks or otherwise take steps to raise capital.
The development of macro stress testing in Europe has followed along somewhat similar lines. However, as discussed in Vítor Constâcio’s contribution, the institutional framework of the tests has changed substantially over time. Only with the fourth round of testing in 2014 was responsibility for the tests placed squarely within the mandate of the European Central Bank. Furthermore, the degree of centralisation of supervision is not uniform within the EU, with the important distinction between large banks within the Eurozone that fall within the mandate of the Single Supervisory Mechanism, which commenced operations in 2015, and non-Eurozone banks for which the national supervisors retain a high degree of autonomy. Nevertheless, a common feature in the US and Europe is that the outcomes of stress tests result in changes in bank capital plans to address supervisory concerns.
Thomas Huertas points out in his piece that the effect of these developments in some cases has been to supersede the regulatory capital standards in Basel III as implemented in the relevant jurisdictions. Generally, this possible inconsistency between two major policy instruments has not been resolved by policymakers. However, as discussed by Alex Brazier in his contribution, the Bank of England has made an effort integrate its own stress testing regime into a coherent regulatory capital framework. The framework consists of regulatory minima (the Pillar 1 standard for all banks plus the Pillar 2a minimum that is bank specific) plus buffers including two that are to be calibrated using the BOE’s macro stress tests. One is the counter-cyclical capital buffer that applies to all banks, and the second is a buffer set by the UK’s Prudential Regulatory Authority on a bank specific basis.
A recurrent theme: Transparency matters
One recurrent theme throughout the book is the importance of the transparency of stress tests. Traditionally, the results of supervisors’ monitoring of specific banks are treated as sensitive and confidential. In contrast, the supervisory stress tests in the US and Europe have been relatively transparent and, in some cases, highly so. The appropriate degree of transparency to apply to supervisory stress testing is an open question, and a number of different views on this question are expressed in the volume. By publishing stress test methodology and results, macro stress testing is able build confidence that the banking system is strong enough to withstand an adverse shock to the economic system. As reviewed by my introductory first chapter to the eBook, this was what convinced many that the 2009 US stress tests contributed very positively to restoring financial stability. And it was on the grounds that they revealed too little information that the first and second rounds of European stress tests were faulted and deemed by many to lack credibility.
In this spirit, Viral Acharya has argued in favour of checking the results of the regulatory stress tests against an alternative measure of systemic risk based on publicly available information. His contribution written with Sascha Steffin applies a systemic measure called SRISK to listed banks included in the EU’s 2014 comprehensive assessment and finds numerous cases where banks passed the 2014 test but would be predict by SRISK to be insolvent when exposed to a large stress.
An alternative perspective of transparency is offered by Til Schuermann, who argues that a high level of transparency may be required at some times while at others it may actually be counter-productive. When faced with an unfolding crisis, as in 2009, the primary question for all was whether the major banks had sufficient capital to survive a recession were it to occur. In that context, transparency of stress tests is crucial. In contrast, in normal times there is no single clear threat to financial stability, but instead, there are many potential threats. In that case, using stress scenarios to explore banks’ vulnerabilities allows supervisors to learn something about the ability of an individual bank to survive a conceivable hazard and also about its ability to probe its own potential weaknesses. In these circumstances, full transparency of stress tests may not be an essential feature.
Yet another perspective on the transparency issue is to note that more transparency does not necessarily contribute to financial stability. The US 2009 stress tests seemed to calm the markets, but this may have been because the tests were transparent and it was clear that there was a plan in place to recapitalise banks if necessary using public means. As pointed out by Thomas Huertas, it was the doubts about the fiscal capacity in some European countries deal with a major bank insolvency and not the lack of transparency per se that may have given rise to the credibility problem in the early EU stress tests. Europe has moved to deal with this credibility problem with Banking Union including the SSM, but also with recovery and resolution planning (so-called Living Wills) and with the Single Resolution Mechanism. In this regard, Charles Goodhart argues that stress testing can make one of its most valuable contributions to financial stability by pointing to vulnerabilities while banks are well away from the point of insolvency. In such circumstances bank management will be naturally reluctant to build up the bank’s capital for fear of diluting existing shareholders and putting their continued tenure in jeopardy. Worrying stress test results can provide the trigger to force action on their part as set out in their own recovery plans.
Another theme that recurs in the eBook is whether stress testing by the public authorities is a tool of microprudential or macroprudential regulation. Traditional banking supervision is the primary tool used to implement microprudential regulation. With the development of supervisory stress testing, prudential actions (e.g. required capital increases) are being based on a forward-looking basis, under stress scenarios set by the supervisor. Using stress tests in this way on a single institution is still in the nature of microprudential regulation. But the fact that the scenarios can change from one round of testing to the next means that they can be made explicitly countercyclical in nature. Thus supervisory stress testing can also serve a macroprudential purpose.
Edge and Lehnert point out that US stress testing, which has designed a one test as an increase in unemployment to a given high level, is implicitly countercyclical. As presented by Brazier, the intent in the next rounds of UK stress testing is to become more explicitly countercyclical by designing more severe stresses in upswings of the cycle and smaller stresses in downturn. In his contribution, Philipp Hartmann recognises that the planned stress testing approach of the Bank of England is breaking new ground in making stress testing a genuine tool of macroprudential regulation. However, he points out some of the new challenges that this step implies for the design of scenarios. The contribution of Udaibir Das gives further insights into what is required to use stress testing as a tool of macroprudential policy. He notes that the International Monetary Fund has been using stress testing for more than 15 years. Based on this experience, the IMF has developed a set of in-house models, generally of the macro top-down variety, that can be used in conjunction with bottom-up analyses provided by local institutions in the country being assessed. The effectiveness of stress testing approaches can depend upon governance conditions including the clarity of the mandates of supervisory and monetary authorities as well as the institutional coverage of stress testing.
Author’s note: The support of the Economic and Social Research Council (ESRC) in funding the SRC is gratefully acknowledged [grant number ES/K002309/1].
Anderson, R.W. (ed.) (2016) Stress Testing and Macroprudential Regulation: A Transatlantic Assessment. CEPR Press.
Hardy, D. and H. Hesse (2013), “The future of Europe-wide stress testing”, 20 April.