The AOCI elephant in the DFAST room

After March’s banking crisis, Fed stress tests should adopt harsher and wider ranging rate scenarios

As analysts from the US Federal Reserve finalised the results of the 2023 Dodd-Frank Act stress test (DFAST), March’s banking crisis – and the factors that precipitated it – may have felt uncomfortably close.

As in every round of the prudential exercise since 2020, most participating banks did not have to reckon with how the recession envisioned by the Fed would affect the value of mark-to-market bonds or other rate-sensitive instruments, such as cashflow hedges. This is thanks to a waiver the Fed introduced in late 2019, which allowed all but the most systemically intertwined banks – meaning all but nine dealers – to smooth such mark-to-market vagaries out of regulatory capital.

Outside the DFAST wargaming room, however, there were growing calls for banks to go back to reflecting these kinds of losses – recorded as accumulated other comprehensive income (AOCI) – in their capital positions. Such losses are widely seen as having precipitated the implosion of Silicon Valley Bank and the wider banking crisis that followed.

The Fed is bound by regulation to conduct DFAST simulations according to the capital standards currently in place. It could not, faced with issues resulting from the crisis, surreptitiously reinclude AOCI losses in a simulation if a bank would not be required to do so in real life. This has put the 2023 exercise out of tempo with the Fed’s vice-chair for supervision Michael Barr, who said the AOCI filter should be curtailed even further than had been the case before 2020, and with lobby group the Bank Policy Institute, which believes the filter should be brought back at “certain” banks.

Another glaring point of divergence from the macroeconomic zeitgeist was DFAST girding banks for a scenario in which the longer end of the US Treasuries yield curve plummets to 1.25%, thereby boosting the value of investment bonds and reducing AOCI losses at banks where such losses affect capital. The assumption of a cut in interest rates that would trigger such a plummeting in the yield curve is in keeping with other elements of DFAST 2023’s severely adverse scenario, which envisions a painful consumption and demand crunch leading to a strangling of inflation. However, it would be far from reflective of the current monetary environment, where even a 0.25 percentage point cut in the Fed’s policy rate is a distant mirage.

DFAST 2023 did contain an alternative exploratory scenario envisioning a milder recession in which rates would be kept lofty by stubborn inflation – a state of affairs closer to banking executives’ present fears. However, the scenario was only used to assess potential trading and counterparty losses for the eight US global systemically important banks, and there was no public-facing analysis on how it might affect the G-Sibs’ AOCI losses. Although the test’s baseline simulation, based on an averaging of actual macroeconomic forecasts for the global economy, does include AOCI computations, its results are not released to the public.

In its current form DFAST is, on the one hand, a top-down drill for how the banking system might fare in the direst of contingencies and, on the other, a capital planning exercise. It is not meant to be taken as an X-ray of a particular bank’s salient vulnerabilities nor of its ability to survive under alternative prudential regimes.

Nevertheless, it is as comprehensive and transparent a tool the Fed and the public have to periodically check up on the health of the banking system. Not to complement its main severely adverse simulation with alternative, speculative scenarios, including ones in which AOCI is tested for different rate paths, seems like a waste of its exploratory potential.

Editing by Daniel Blackburn

Only users who have a paid subscription or are part of a corporate subscription are able to print or copy content.

To access these options, along with all other subscription benefits, please contact or view our subscription options here:

You are currently unable to copy this content. Please contact to find out more.

You need to sign in to use this feature. If you don’t have a account, please register for a trial.

Sign in
You are currently on corporate access.

To use this feature you will need an individual account. If you have one already please sign in.

Sign in.

Alternatively you can request an individual account here