Stress testing: Current problems and the path ahead

Stress testing: Current problems and the path ahead

The Center for Financial Professionals (Cefpro) held its annual CCAR and DFAST conference in New York on November the 3rd and 4th, 2016. Over 250 participants from across the US met to discuss the status of US stress testing, with the conference in its 4th year. The following recaps one of the discussions that cut across the many talks.

With two days split into two tracks each, Cefpro’s Stress Testing conference had very many different and detailed problems and solution ideas covered. Participants discussed everything from Dynamic CoVariance Models to team management across old organizational silos; all attempting to answer where stress testing stands, where it will go and how best to handle current and future challenges.

Many of the challenges covered by two themes: First, there were many discussions on data infrastructure, model practices and model governance. A large number of institutions are still building policies and procedures to handle the stress testing challenges. Secondly, participants were trying to chart the path ahead. Everyone had Governor Tarullo’s stress testing speech from September in the back of their minds. Tarullo’s key message was that the stress testing exercises would be dynamic and adapt to changing financial conditions. The nature of that adaption is still uncertain, but there are clues and these were discussed vigorously during the conference.

A cornerstone

“Stress testing has become a cornerstone of post crisis financial regulation”. That’s the premise of a lot of the Fed’s CCAR and DFAST work. It makes sense as stress testing is a way of marrying capital controls from Basel III’s first pillar with the second pillar’s prudential oversight. It’s a lever for a more holistic approach to financial regulation. With it, the Fed becomes an active player in banks’ risk management. Both quantitatively and qualitatively, and on an ongoing basis. Everyone agrees that  CCAR and DFAST procedures slowly morphed from yearly fire drills towards ongoing relationships – the questions are around the road ahead and the uncertainties this brings forward.

The Fed’s proposals for changes reflect this consolidation move. First of all CCAR and pillar-one capital requirements are being blended with Tarullo’s ”Stress Capital Buffer” (SCB). It formally makes the stressed scenario capital loss in CCAR the binding constraint for banks as their capital planning would be capped if they fall below the base CET1 requirement plus the maximum of the capital conservation buffer and the capital loss in the stressed scenario plus (crucially) any GSIB surcharge. Along with a drop in qualitative requirements for large but less complex institutions,[1] the SCB is the main immediate change going forward a year or two.  The two changes present no new requirements to CCAR/DFAST infrastructure in banks, but the SCB tightens the capital wiggle room.

A marriage between liquidity and capital

The more interesting, uncertain and potentially impactful changes lie in the ”medium term”, which could be anything from 4-7 years. On top of ”adapting stress testing to financial structure changes”, the Fed is trying to absorb various regulatory agendas into stress testing. Most remarkably, macroprudential regulation. ”Macropru” has been the hot topic in regulatory circles ever since Basel III was initiated. Regulators always knew they had to address interdependency risks beyond the individual institutions but they only produced the hard-to-implement ”Counter Cyclical Capital Buffer”. Until now, that is. Stress testing might be the golden lever that actually enables macroprudential financial regulation. But it is uncharted territory. Systemic risk is often associated with various shocks to liquidity. Different forms of ”liquidity” make up the linkages between individual institutions’ balance sheets that cause systemic crises. The Fed aims to do research that will at least implement funding and market liquidity shocks into the stress testing framework. The vicious cycle between funding crunches and firesales pressures on market prices played a large part in the 2007-08 crisis so it makes good sense.

The problem is that nobody knows how to model these effects. Vicious cycles can be modeled but it’s hard to have them end endogenously. In practice, they usually end because Central Banks step in and support the systems under their control. With a never ending liquidity crunch your stress test will easily wipe out all your capital and that doesn’t seem viable. Many participating institutions had thought about marrying capital and liquidity stress tests. It makes sense theoretically but the nature of liquidity crunches makes the modeling too hard. Nevertheless, the Fed is working to implement these liquidity issues and it will probably end up in some form of link between CLAR and CCAR stress results. It is hard to be more precise and it sounds as if the Fed have a lot to consider and many challenges ahead. It will be crucial to watch how the regulator decides to conceptualize these risks, and how these will be implemented.


Article by Frederik Vitting Hermann


[1] ($50—$250 billion in total consolidated assets, less than $10 billion in on-balance sheet foreign exposure, and less than $75 billion in total nonbank assets

Our 6th annual Risk Americas 2017 Convention will be taking place on May 23-24 and will feature a dedicated track on stress testing and model risk. Register before the agenda is announced for the heavily reduced rate of only $999.