The Federal Reserve has made its bank stress test more predictable, and it has told us the price: about half the year-over-year swing in capital requirements, with no material change in the total. I think that is a good trade on its own terms. I also think it moves the whole argument onto one question the final rule does not answer, which is whether the scenarios stay severe once banks can see most of what drives them.
What the Board finalized
On September 30, the Board finalized two rules. The first requires the Board to invite public input every year on the stress test scenarios and on any material model changes, and it adopts the models for the 2027 test. Banks with large trading books will be tested against two global market shock components each year, and the Board will use whichever produces the largest losses for each firm. The second rule averages the results of the two most recent annual tests when it sets stress capital buffer requirements, starting in 2028.
The Board’s own estimate of the effect is that the changes “are likely to reduce year-over-year volatility in capital requirements by approximately 50 percent and are not expected to materially affect aggregate capital requirements.”
Advertisement
The case for the rule
I start with the strongest argument on the other side, because it is a serious one. Vice Chair for Supervision Michelle W. Bowman said that “the stress test is an essential component of our regulatory capital framework” and that the changes preserve its resilience “by ensuring that it is transparent, granular, and risk-sensitive.” Averaging two years removes some year-to-year noise from the capital requirement. Public comment on scenarios gives banks and outside experts a chance to say when a scenario makes no sense.
Governor Lisa D. Cook, who supported the package, made the same point about trust: she wrote that “the framework finalized today preserves the Board’s capacity to administer a trustworthy, effective stress testing regime.” On the volatility goal specifically, I agree with the Board. A capital requirement that moves by half as much is easier to plan around, and planning is where capital becomes lending.
Where I think the rule leaves a gap
The objection comes from inside the Board. Governor Michael S. Barr, who voted against the final rule, wrote that “calcified models will also allow banks to optimize their balance sheets to the test, rather than focusing on underlying risk.” He added that using the same models “could increase concentrated risks in the financial system.”
A test that is published, commented on and averaged is a test that banks can forecast. A forecastable test rewards the bank that fits the test best, which is not always the bank that is safest. The rule’s answer is in its design: Cook says it has policy design features meant to reduce “gaming” and “window-dressing” by banks.
What the rule does not contain is a commitment on severity. Cook says so directly: “Should stress tests become less severe or overly predictable over time, we may need to contemplate other options to maintain resilience.” That sentence is a promise to act later, not a requirement now. She also says she sees “significant benefits” in exploratory stress scenarios that carry no direct effect on regulatory capital but would show supervisors alternative conditions. Those scenarios are a good idea. They are not in the final rule, and nothing in the Board’s announcement says when they will arrive.
My position
The Board has delivered the measurable half of its promise. Volatility is a number, the Board has put a figure on it, and next year’s results will show whether the figure held. Severity cannot be checked in advance, because it depends on scenario choices made year by year.
So the test of this rule is not the 50 percent. It is whether, in the first two or three cycles, the Board publishes at least one scenario that banks did not expect and that costs them real capital, and whether it brings exploratory scenarios into the supervisory process on a stated schedule. If the scenarios stay hard, the volatility gain is a clean win for planning. If they soften, Barr’s concern about a test that no longer assesses the largest banks will have been borne out, and the predictability will have been bought with the one thing the test exists to provide.
Cook has already described the principle. A date and a trigger would turn it into a rule.
Related: our coverage of the living-wills feedback letters and our opinion on the Board’s stablecoin rule.
Source: Federal Reserve Board
