How to Read the 2026 Results Without Being Misled

The Federal Reserve’s 2026 stress-test report offers several numbers that seem to invite a quick verdict: an aggregate CET1 minimum of 11.2 percent, nearly $708 billion in losses, and bank-specific minima ranging from 6.7 percent to 32.2 percent. This chapter applies the earlier framework to those figures. The practical question is not whether the numbers are “good” or “bad” in isolation, but which scenario, channel, exposure, sample, date, and assumption make each number true.

That discipline matters because a stress-test report is dense with quantities that describe different things. A capital ratio is not a loss total. Pre-provision net revenue is not net income. A bank-specific minimum is not a forecast of that bank’s future capital. And a year-over-year change is not automatically a trend when the sample, scenario, exposures, and models have changed.

Read the headline as a decomposition

Start with the capital path. Across the 32 banks in the 2026 exercise, the aggregate CET1 ratio fell from 12.8 percent at 2025:Q4 to a projected minimum of 11.2 percent, then rose to 12.7 percent by 2028:Q1. The Federal Reserve reports that the aggregate and individual post-stress CET1 ratios remained above required minimum regulatory levels throughout the projection horizon under the exercise’s assumptions.

That is a meaningful statement about conditional resilience. It says that the tested banks, beginning with their actual starting conditions and evaluated under the severely adverse scenario, retained a modeled capital cushion above the specified regulatory floors. It does not say that the banks will actually move from 12.8 percent to 11.2 percent and back to 12.7 percent. The path belongs to the modelled counterfactual, not to an unconditional forecast.

Next, open the loss total rather than treating it as the explanation by itself. The report projects nearly $708 billion in aggregate losses. About $625 billion comes from loan losses, $37 billion from trading and counterparty losses, $7 billion from securities losses, and $39.5 billion from other losses or gains. These categories do not all enter the calculation in the same way. They reflect different exposures and different routes through pre-tax income and capital.

Now add the income side. Aggregate pre-provision net revenue is projected at $719 billion over the nine quarters, but aggregate cumulative pre-tax net income is only $1.2 billion. That gap is the point of the decomposition. Revenue provides a substantial offset, while provisions and other losses consume most of it. A reader who sees $719 billion and concludes that banks remain highly profitable has ignored the loss channels. A reader who sees $708 billion and concludes that banks are destined to lose that amount has turned a conditional projection into a forecast.

The distinction between pre-provision net revenue and pre-tax net income is especially important. PPNR describes the revenue and expense capacity before provisions and several loss categories are subtracted. It is a cushion within the stress calculation, not the final result. The Federal Reserve also includes items such as operational-risk expenses in the relevant revenue and expense projections. The measure is therefore useful for understanding how much income can absorb stress, but it cannot be used alone to describe profitability or safety.

Other effects on capital complicate the picture further. Unrealized gains and losses on available-for-sale securities affect capital for the banks to which that treatment applies. Other comprehensive income is reported separately. Trading and counterparty losses apply to the banks subject to the relevant market-shock and counterparty components. The capital ratio is the combined outcome after these channels are brought together.