This post examines the stress test results recently released by the Federal Reserve.[1] While the outcome was a reduction in projected capital depletion under severely adverse conditions relative to last year, the stress test results continue to demonstrate year-over-year volatility. Because these results are directly linked to capital requirements, this volatility undermines banks’ efficient capital management and planning. The Federal Reserve has recently acknowledged the limitations of its stress testing framework that contribute to this volatility and is working to reduce it through averaging and other steps.
Overview of the 2025 Stress Test Results
On June 27, 2025, the Federal Reserve released the annual stress test results for the 22 large banks participating in this year’s exercise.[2] All of these institutions were found to remain well capitalized through the hypothetical severely adverse scenario and Global Market Shock posited by the Federal Reserve.
The stress test projected an aggregate 1.8-percentage-point peak-to-trough decline in the common equity tier 1 (CET1) capital ratio over the nine-quarter severe stress horizon, significantly smaller than the decline in 2024 (2.8 percentage points). Broadly consistent with BPI’s analysis of the 2025 stress test scenarios, this lower CET1 depletion was driven by increases in pre-provision net revenue (PPNR) and decreases in loan losses, trading and counterparty losses and private equity losses.
These aggregate results mask some variation across participating banks. Exhibit 1 shows the weighted average peak-to-trough declines in CET1 under the stress test overall and by bank category for the past five annual exercises, excluding Category IV institutions.[3] Note that the figures for “all banks” are different from the Federal Reserve’s aggregate numbers above as they are rolled up from the bank-specific results.[4] Category I banks saw the largest weighted average peak-to-trough declines in CET1 under the stress test, while Category II and III institutions also experienced a material improvement in their outcomes on a weighted average basis.
Exhibit 2 shows the change in peak-to-trough declines in CET1 produced by this year’s stress test versus those from 2024 for the 22 banks that participated in both exercises. All but one institution saw a smaller decline in CET1 and four banks saw peak-to-trough declines greater than 2 percentage points over the two exercises.
Two of the key drivers of the smaller decline in CET1 capital ratios produced by the 2025 stress test were significantly higher projected pre-provision net revenue (PPNR) and lower projected loan losses owing to a less severe scenario.[5]
Income generation over the nine-quarter stress test horizon significantly buffers banks against the losses incurred through the severely adverse scenario. PPNR can be broadly decomposed into net interest income, non-interest income and non-interest expense. Net interest income is naturally dependent on the path of the term spread, or the difference in yields between 10-year and three-month Treasury securities. Statistical models used to project non-interest income and non-interest expense (including operational losses) are specified in a way that recent bank performance is heavily weighted.[6] This approach creates a “momentum effect” that results in volatility in stress test projections within these PPNR categories and ultimately bank stress capital buffers (SCBs).[7]
As shown in Exhibit 3, for the 22 banks participating in both the 2024 and 2025 stress test exercises, the Federal Reserve’s PPNR projections under the severely adverse scenario increased significantly. Consistent with the “momentum effect”, the Fed primarily attributes the improved PPNR projections to improved bank profitability in 2024.[8]
Loan losses represent a large fraction of total losses in the stress test and are driven by the paths for macroeconomic factors posited by the severely adverse macroeconomic scenario, including output growth, unemployment, home prices and corporate bond spreads. Total loan losses under the stress test scenario were projected to be $472 billion, of which $157 billion (33 percent) were from credit cards and $124 billion (26 percent) from commercial and industrial loans.
Exhibit 4 presents projected loss rates for different types of loans in the 2025 stress test as compared with 2024 for the 22 banks participating in both exercises. These projected loss rates declined overall, with a particularly notable decrease in commercial real estate exposures owing to a less severe path for underlying asset values in the severely adverse scenario.
Stress Test Results Demonstrate Continued Volatility
While the 2025 stress test results projected less capital depletion under severely adverse conditions relative to last year, they illustrate the continued volatility in annual outcomes. Because stress capital projections feed into bank capital requirements, such volatility undermines efficient capital management and planning. The Federal Reserve has acknowledged the limitations of its stress testing framework that contribute to this volatility and is working to reduce it.
Stress capital buffers are set annually for each bank as the peak-to-trough decline in the CET1 ratio produced by the stress test plus planned dividends (as a percentage of risk-weighted assets) over the next four quarters, subject to a 2.5 percent floor. In April 2025, the Federal Reserve proposed to average stress test results over two years to reduce the volatility of capital requirements from year to year.[9] The Fed has not confirmed whether the averaging proposal, when finalized, would apply prospectively, making it unclear whether final 2025 SCBs will be determined under new or existing rules.[10] Arguably, the proposed averaging should apply prospectively as preliminary SCB requirements and anticipated dividends have historically been disclosed to investors and the public within two business days of the stress test results. A bank similarly cannot reasonably adjust anticipated capital distributions if it is unclear how the final SCB requirement will be calculated when the distributions are to be paid out.
The Federal Reserve proposal to average stress test results over two years to reduce the volatility of capital requirements, while helpful, does not address the underlying cause of the problem. As noted above, PPNR projections are an important source of volatility of annual stress test outcomes owing to their dependency on recent bank performance. The Fed has indicated that it plans to revisit the suite of stress test models it uses to project PPNR to reduce the sensitivity to recent periods and decrease year-over-year volatility.
More broadly, the Federal Reserve’s historical approach to the mechanics of the stress testing framework creates additional uncertainty for participating banks. The Fed has committed “to improve the transparency of the stress test process by disclosing and seeking comment on the models and scenario design framework that determine banks hypothetical losses and revenues under stress.”[11] Greater transparency will enable banks to make more informed and efficient decisions when managing their balance sheets.
Conclusion
The 2025 DFAST stress test results again found that participating banks would remain well capitalized through the hypothetical severely adverse scenario and Global Market Shock posited by the Federal Reserve. Moreover, the aggregate reduction in bank capital projected by the 2025 stress test was significantly lower than that produced last year. Nonetheless, the resulting year-to-year volatility of stress test outcomes and required capital buffers undermines bank efforts to efficiently allocate capital.
The Federal Reserve has recognized the limitations of its stress testing framework that contribute to this volatility. To that end, it proposed averaging stress test results and has committed to reevaluating its PPNR models. The Fed has also promised to disclose and seek public comment on the stress test models and scenario design framework. Taken together, these changes should result in more stable and predictable bank capital requirements in the future.
[1] Board of Governors of the Federal Reserve System (2025). “2025 Federal Reserve Stress Test Results.” https://www.federalreserve.gov/newsevents/pressreleases/bcreg20250627b.htm
[2] The number of banks participating in the 2025 stress tests is lower than in 2024 since Category IV institutions are only subject to the exercise every other year. (For 2025, two Category IV banks elected to participate.) As a result, the aggregate results are not fully comparable across the two years.
[3] Category IV banks are excluded from this analysis as only two participated in the 2025 stress testing cycle.
[4] The difference arises as the Federal Reserve’s aggregate measurement results in the minimum value, or trough, being captured at a single point in time over the nine-quarter stress test horizon. Rolling up firm-level results accounts for the fact that minimum values for individual banks occur at different times.
[5] Two other drivers of the smaller aggregate CET1 decline were lower private equity losses and trading and counterparty losses. Private equity exposures were removed from the Global Market Shock component of the 2025 stress test and instead projected under the severely adverse macroeconomic scenario. The Federal Reserve attributes its lower projection of trading and counterparty losses to “atypical client behavior at certain banks in early October 2024, when positions were measured for the 2025 stress test”. See Board of Governors of the Federal Reserve System (2025) at 4.
[6] See Board of Governors of the Federal Reserve System (2025) at 4.
[7] See Board of Governors of the Federal Reserve (2025) at 4, FN3.
[8] The Federal Reserve also notes that it adjusted the historical data for non-interest expense models to exclude certain “non-recurring” expenses.
[9] See Modifications to the Capital Plan Rule and Stress Capital Buffer Requirement, 90 Fed. Reg. 16,843 (April 22, 2025), https://www.federalregister.gov/documents/2025/04/22/2025-06863/modifications-to-the-capitalplan-rule-and-stress-capital-buffer-requirement.
[10] If implemented as proposed, the Federal Reserve estimates that the averaged aggregate decline in CET1 would be 2.3 percentage points (compared with 1.8 percentage points using only the 2025 decline).
[11] Board of Governors of the Federal Reserve (2025) at 4.
