In the nine quarters between September 2007 and December 2009, Citigroup reported pre-tax losses of $77 billion, burning through its Tier 1 capital and requiring a US government bailout. This is the historical narrative behind the format of the Dodd-Frank Act Stress Tests (DFAST). Every year, the big American banks are required to model the impact of a 2007-9 scenario on their balance sheets, over a nine-quarter time horizon. The Federal Reserve tweaks the scenarios every year but the result is the same: the banks’ capital ratios dip down towards the regulatory minimum, and as long as they don’t go too close to it, the Fed gives them a pass grade.
While the DFAST cycle appears quite different to Basel capital requirements, we argue that the narrative of something happening two years in the future is a red herring. The main output is a stressed capital ratio, and that is a number which is important today. In effect DFAST is a parallel capital requirement on the banks, even though it is calculated differently. To show this, Risky Finance has transformed the results into a format that looks more like a capital requirement.
The implied stressed risk-weighted assets for six large US banks are available as an interactive visualisation for Risky Finance subscribers, and can be viewed alongside our four-year database of Basel advanced Pillar 3 RWA disclosures for the same banks. Subscribers can access this data and read our detailed analysis here.
Perhaps the biggest takeaway from our exercise is that DFAST is no longer a real stress test. After showing scepticism about the banks’ modelling a few years ago (and failing Citigroup twice in its CCAR review), the Fed is now waving the banks through. While the scenarios look impressive – a 6.5 percent GDP decline, a 5 percent unemployment increase, equity prices plunging by 50 percent and US house prices by 25 percent – the banks have learned how to jump through the Fed’s hoops.
Although it is more stringent than Basel III rules in limiting banks’ leverage, ultimately DFAST is no more than a internal modelling exercise. Rather than showing that the banks are getting safer, DFAST might just be telling us that the banks are becoming better at doing the modelling. There are so many unpredictable links and nonlinearities in a real deep recession or market shock, and no-one should be fooled that DFAST can account for them.