Sunday, August 4, 2013

IIF Meeting in Paris (I) – Banking Regulatory Stress Testing: Fit for purpose?

On June 25 and 26, 2013, the Institute of International Finance held its spring membership meeting in Paris. Unfortunately, I wasn't invited. To be honest, I wasn't even aware of the meeting before it took place. Anyway, the content is on-line and this is actually enough to write a short summary about it.

The first discussion was about regulatory stress testing.

Jean-Paul Chifflet, CEO of Crédit Agricole

Jean-Paul starts off with some history: He describes that stress tests were initially an internal tool for banks to measure their risk. It is only since the financial crisis that stress tests have become a major tool of communication for regulators with the markets.

In the main part of his speech, he insists on two stress test concepts that have crystallized:

First, the use of stress tests is useful for banking management: It helps understanding the interaction between risks. In addition, stress testing involves analyzing portfolio correlation on the basis of vast historical data and some form of forecasting. Thus, it leads to a proactive risk management policy.

Second, as is the case with any complex communication tool, its difficulties and limitations should not be underestimated. Jean-Paul Chifflet says that stressed times are characterized by sudden, non-linear effects such as contamination. This makes forecasting difficult, if not impossible. The remedies are twofold: On the one hand, assumptions of scenarios must be perfectly clear and communicated as such. On the other hand, right calibration of scenarios is also important: Crédit Agricole's CEO insists that assumptions should not be directed by short-term considerations but look at the whole business cycle, especially if the cycle itself is under stress. Finally, assumptions must take into account existing prudential filters.

Towards the end of his speech, Jean-Paul touches upon the question to what extend stress tests should provide information. This part is a bit confusing to me. Although transparency through full disclosure of information is king, he also says that stress tests should not convey too much information or replace standard financial communication.


“The credibility of the communication of stress tests depends on the communicator's credibility and ability to act and take on the complexity of stress test information provided.”

“Transparency through full disclosure of information.”

“If you provide too much information [when communicating stress tests], this might be counterproductive. Don't jump ahead of the gun!”

“Regulatory stress tests shouldn't replace standard financial communication.”

Timothy Clark, Board of Governors, Federal Reserve Board

The main feature of Timothy's intervention is to distinguish three functions of stress testing:

  • As a risk management tool, stress tests ensure that bank managers understand risks and dispose of reliable data – to that extent available, on a group level.
  • From a corporate governance viewpoint, they allow people making adequate decisions for the capital adequacy of the firm.
  • Finally, stress tests remind financial managers about the bad things that can happen.

Besides, he outlines the relationship between stress testing and comprehensive capital analysis and review: Although they constitute separate processes, Timothy Clark holds that, together, they work much better. A final thought goes to the credibility of stress tests: Transparency and disclosure (of scenarios as well as results) are critical here.


“Capture the risks better on a firm-wide basis globally is a major objective of stress testing.”

“Don't forget that bad things can happen!”

“The conversations that the firms have [about stress testing] have moved up to a higher level and that's critical.”

Adam M. Gilbert, Managing Director at JP Morgan Chase & Co.

Adam begins his speech by focusing on the interaction between finance and risk. He says that JP Morgan tries to bring these disciplines closer to each-other. As a consequence, stress testing should become more relevant for processes such as budgeting, forecasting, and capital planning.

To be successful at firm level, stress tests should meet six conditions:

  • Consistent use of economic and management assumptions;
  • Product centric forecasting;
  • Match idiosyncratic risk and have sufficiently granular data;
  • Use multiple macro scenarios
  • Challenge your processes, businesses and their assumptions and modeling
  • Legal entity stress testing: According to Adam, stress testing should, today, focus on the consolidated firm. However, in the future, they should also apply to the each legal entity of a group.

Adam then shares four questions and concerns with his audience:

  • How do stress tests interact with other supervisory policies and rules such as the leverage ratio?
  • Can stress testing become another tool for supervisors to address the question of “too big to fail”?
  • What do you do if you have firms in different countries subject to different stress testing rules? What is the effect of having stress tests in multiple jurisdictions? Should and, if yes, can stress testing rules be harmonized?
  • How do stress tests on capital interact with stress tests on liquidity?

Finally, Mr. Gilbert talks about two aspects of stress test disclosure:

  • Timing of stress test publication: In the US, they coincide roughly with the publication of banks' first quarter earnings. (I don't really understand whether Adam finds this good or bad.) Stress tests by supervisors and by firms should respect the right sequencing. (What the right sequencing is, remains unanswered.)
  • Level and detail of disclosure: The speaker calls this a continuing balancing act and says that “too much detail can get overwhelming and confusing.” On the other hand, “it's important for the marketplace to understand what these tests are telling us.”


“Stress testing has come quite a long way.”

“We really use it [stress testing] as an instrument for capital planning.”

“Stress testing has usurped what has traditionally been borne by management judgment.”

“[Stress testing] is not only a compliance exercise. It is meant to be something that enhances the financial and risk management practices with the firm.”

Piers Haben, Director Cluster Oversight, European Banking Authority

Piers Haben focuses on the importance to clarify the objectives and circumstances of stress tests. Although he ideally wants to use them as a means of supervisory tools, he recognizes that, in practice stress tests are subject to three contingencies:

  • They are often carried out in crisis times and, thus, undertaken in the public eye.
  • Only their result (pass of fail) counts.
  • They focus on one particular solution – often capital enhancement.

With respect to Europe's condemned stress test of 2011, Piers summarizes six lessons learned:

  • Be clear in your communication about your objectives, expectations, and proposed solutions.
  • Get the starting point the ending point right (“Rubbish in, rubbish out!”).
  • Use clear and consistent definitions and data.
  • Choose the right methodology in light of your objectives.
  • Enhance the systems and resources for supervisors to challenge the banks' outcomes.
  • Embellish the transparency of actual figures.


In my nirvana, we would be doing supervisory stress testing as an analytical tool […] and use the outcomes as part of a sweet of supervisory measures.”

Rubbish in, rubbish out!”

There is no one right methodology for supervisory stress testing.”

Dr. Til Schuermann, Partner, Oliver Wyman

Til concentrates on four stress test criteria. In this opinion, stress tests must be

  • easy to understand,
  • transparent (so that outsiders can for themselves figure out what is going on),
  • specific to the situation at hand, and
  • vested with a credible backstop.

On the subject of transparency, the speaker mentions possible risks of publishing too much information. As a matter of fact, the more information is published about stress tests, the more banks will have incentives to match such information.

The solution can be for the supervisor to publish only aggregates, for example average loss rates. This avoids banks playing the game of matching them but, at the same time, gives the market an idea about how bad in can in the given scenario.


“Fundamentally, [stress testing] is supposed to generate information that didn't exit before.”

“[For the US stress test in early 2011,] there was no disclosure at all – and nobody cared!”

“You can only have a view on what you can see.”

“The success [of stress testing] is intimately tight to generous disclosures.”

“It's no good showing how sick the patient is if you don't have medicine to cure it.”

“Rich transparency is not a free lunch – not without costs and not without unintended consequences.”


If you work for a member institution and register with the Institute of International Finance, you can watch or download the speeches here.