Fed finalizes its stress test overhaul 6-1; its own staff put the capital cut at about 1%

The Federal Reserve Board on Wednesday finalized two rules that change how it stress tests the largest banks and how those results turn into capital requirements. The Fed will now invite public comment every year on its stress scenarios and on material model changes, and it will set each bank's stress capital buffer from the average of its two most recent tests rather than the latest one alone. The Board also proposed a new model for how banks' fee income holds up under stress, with comments due 60 days after publication.

Bloomberg framed the vote as a major win for Wall Street banks. The Fed's own numbers suggest the win is mostly about predictability, not about freeing up a large amount of capital.

The number behind the headline

The Board staff memo behind the vote gives the figure the press release leaves vague. Measured against the 2024, 2025 and 2026 stress test cycles, staff estimate the changes would have lowered aggregate stress capital buffer requirements by about 1% of required common equity tier 1 capital. The same memo says the package should cut the volatility of year-to-year changes in those buffers by about 50%, and that it is "not expected to materially change aggregate capital requirements."

So the change that matters for bank investors is the smoothing. Under the current system a single bad test can raise a bank's buffer sharply in one year, which forces it to hold back buybacks and dividends. Averaging two years dilutes any one result. The buffer still has a 2.5% floor, and it still adds four quarters of planned dividends on top of the modeled capital decline.

When it actually bites

The timing is later than a first read suggests:

• 2027 test: run on models that already went through public comment. The requirement to take comment on scenarios every year starts with the 2028 test, because the rule arrived after the August 31 cutoff for next year's cycle.
• Buffer effective date: moves to January 1 of the year after each test, from October 1. Banks get three more months to adjust, starting with the buffer that takes effect January 1, 2028.
• Averaging: the release says it "begins in 2028." In practice that is the buffer calculated from the 2028 test, which takes effect January 1, 2029. The Fed delayed it so that only models that went through public comment feed the average.

Banks with big trading books also face a tougher market-shock piece: the Fed will run two global market shock scenarios each year and apply whichever produces the larger loss for each firm.

The dissent

The Board voted 6-1, with Governor Michael Barr against. In his statement he backed the two-scenario market shock but said publishing the models and taking annual comment will make the test "less responsive to emerging risks." He also warned it lets banks optimize their balance sheets to the test. Governor Lisa Cook voted yes but said in her statement that if the tests become "less severe or overly predictable" over time, the Fed "may need to contemplate other options." Vice Chair for Supervision Michelle Bowman said the changes keep the test "transparent, granular, and risk-sensitive."

Who it hits

For large-bank shareholders, steadier buffers make capital return easier to plan, since a rough stress test year no longer forces a sudden cut. For business borrowers, a change worth about 1% of required capital across the system is not the kind of shift that loosens credit on its own. The Fed's memo argues that clearer capital rules should help banks lend, but it attaches no dollar figure to that. Loan pricing for a company with a floating-rate line still depends mostly on the prime rate and on each bank's own credit appetite.

The results of the next test, the first run on the newly adopted models, are due in 2027. Our Fed and rates coverage tracks the rest of the Board's regulatory calendar.

Sources: Federal Reserve Board press release and Board staff memo; statements by Governor Barr and Governor Cook; Board votes; Bloomberg. This is market information, not investment advice.