On Feb. 5, 2025, the Federal Reserve released the severely adverse scenario and the global market shock (GMS) component for the supervisory stress tests, which will be used to calculate the stress capital charge imposed on large banks. The severely adverse scenario is used to assess a bank’s ability to withstand a severe macroeconomic recession and to set a bank-specific capital buffer, while the GMS is imposed on banks with significant trading operations, resulting in additional stress losses that contribute to the stress capital charge.
This charge is determined by the decline in each bank’s common equity tier 1 capital ratio under the severely adverse scenario, with a larger decline resulting in a higher charge. Failure to maintain the required level of minimum capital, including the stress capital charge, severely restricts a bank’s ability to distribute capital to shareholders.
In aggregate, the 2025 severely adverse scenario is slightly less severe than last year’s scenario, particularly regarding the assumed trajectories of commercial real estate prices, equity prices and corporate bond spreads. Based on our models, we anticipate a modest decline in loan losses. In addition, we project significant improvements in pre-provision net revenue projections relative to the 2024 stress test. While we expect the improvements in revenue to come primarily from investment banking fees, expected changes by the Fed staff to PPNR models may moderate the increase in this year’s projections.[1]
Furthermore, the global market shock component includes less severe risk factors compared to last year’s scenario, such as equity price shocks and interest rate shocks. The Federal Reserve also announced that private equity shocks will not be included in the GMS; instead, they will be stressed using the severely adverse macroeconomic scenario.[2] This change might result in slightly lower capital depletion under stress for some large banks.
Our top-down models indicate that banks participating in this year’s stress test will experience a lower depletion in their projected bank capital ratios compared to last year.[3] Exhibit 1 presents the projected decrease in CET1 capital ratios, broken down into Category I banks, other banks and all banks combined. The weighted average common equity tier 1 capital ratio declines by 2.7 percentage points from the start of the projection horizon to its minimum in the 2025 stress test. Last year the decline was 3.2 percentage points. The expected improvement in this year’s results can be attributed to better PPNR projections, slightly lower provisions for loan losses and reduced severity of market shocks.
A new aspect of this year’s stress tests is the Fed’s approach to modeling PPNR. We anticipate significant improvements in net interest income projections due to the inclusion of the full period of higher interest rates in the Fed’s annual model updates. However, there is a risk to our forecast if the Fed decides to reduce the dependence of PPNR projections on recent performance, particularly for net interest income and noninterest revenues. This adjustment could be made to reduce year-over-year volatility in stress test projections. As a result, while we expect overall improvements in PPNR projections, the Fed’s potential model changes could temper the magnitude of these improvements and their subsequent impact on banks’ capital charges.
In addition to the severely adverse scenario, the Fed also announced two additional “exploratory” scenarios to evaluate the resilience of banks to a wider range of risks. These scenarios include shocks to the nonbank financial sector during a severe recession and an exploratory market shock with the default of the largest five hedge funds. The results of these exploratory scenarios will not be used to calculate the stress capital buffer.
The 2025 severely adverse scenario is less severe compared to 2024…
This year’s severely adverse scenario includes, on a start-to-stress basis:
- A 5.9-percentage-point increase in the unemployment rate (6.3 p.p. in 2024).
- A 30 percent drop in commercial real estate prices (40 percent in 2024).
- A 33 percent decline in house prices (36 percent in 2024).
- A 3.9-percentage-point increase in corporate bond spreads (4.1 p.p. in 2024).
- A 50 percent drop in the stock market (55 percent in 2024).
- A 7.8 percent fall in real GDP (8.2 percent in 2024).
- A 65 peak value in the volatility index (70 in 2024).
The Scenario Design Framework stipulates that the unemployment rate should increase by 3 to 5 percentage points, or at minimum, to a level sufficient to reach a peak of at least 10 percent. For the 2025 severely adverse scenario, this results in a 5.9-percentage-point rise in the unemployment rate, compared to a 6.3-percentage-point increase in the previous year’s scenario. This slightly smaller increase in the unemployment rate is consistent with the less severe declines in real estate prices, real GDP, bond spreads and the stock market.
Exhibit 2 displays four macroeconomic variables that drive bank performance in the stress tests. All four variables – the unemployment rate, the CRE price index, the stock market index and the corporate bond spread – show lower severity levels compared to last year. The CRE price index declines 30 percent in this year’s scenario, versus 40 percent last year (top-right panel). Equity prices fall by about 50 percent, compared to 55 percent previously (lower-left panel). Bond spreads widen by 3.9 percentage points, which is slightly less severe than last year (lower-right panel).
Exhibit 3 shows the path of key variables that drive the projections of net interest income in the Federal Reserve’s stress testing models. In the severely adverse scenario, the higher term spread—the difference between the 10-year and three-month yields of Treasury securities—creates more of a tailwind for net interest income projections in this year’s stress test. In addition, the performance of banks in the year preceding the start of the stress test planning horizon significantly affects PPNR projections. Specifically, the net interest margin at the jump-off date of this year’s stress tests is modestly lower than at the beginning of last year’s, as shown in the right panel of Exhibit 3. However, the recent decline in net interest margins is not sufficient to counteract the positive effect of the higher term spread over the projection period.
Loan losses and provisions are projected to decrease…
Observing the paths of the macroeconomic variables over the stress horizon provides only a partial view of how the severely adverse scenario impacts bank performance in stress tests. A bank’s initial balance sheet at the start of the stress tests also plays a crucial role in determining its performance. For instance, if a bank disposes of assets between two stress tests, its losses would decrease even if the macroeconomic variables remain relatively unchanged.
To overcome this limitation and establish a more robust connection between macroeconomic variables and bank-level performance, we utilize a series of top-down, time-series models. These models help us estimate the severely adverse scenario’s impact on loan losses, provisions and pre-provision net revenue projections throughout the stress planning horizon.
Our analysis indicates that loan loss projections for the 2025 stress scenarios are expected to be lower than the previous year, mainly because of reduced projected losses in real estate and commercial and industrial loans. For the 22 firms participating in this year’s stress tests, total estimated loan losses are projected to reach $453 billion, which is approximately $25 billion less than in 2024 for the same group of banks.
The lower projected losses are driven by the associated paths of macroeconomic variables. Specifically, losses on CRE loans are expected to be lower because of a less severe path of CRE prices. Losses on residential real estate loans are projected to decrease as a result of a slightly smaller decline in house prices. Additionally, losses on commercial and industrial loans are anticipated to be lower because of a smaller increase in corporate bond spreads.
However, despite these improvements on loan losses, total losses remain about 27 percent higher compared to the severely adverse scenario in the 2020 stress tests. This comparison shows the increased severity of stress scenarios in the post-pandemic period.
Exhibit 4 plots the projected loan loss rates over the past four stress testing exercises. The aggregate loss rate is projected to decline 40 basis points to 6.3 percent. The loss rate for CRE loans is expected to decrease from 8.6 percent to 6.2 percent compared to last year’s test, primarily because of a less severe trajectory for CRE prices. Similarly, the loss rate for commercial and industrial loans is predicted to decline from 8.1 percent to 7.4 percent because of a less severe outlook for the unemployment rate, real GDP and corporate bond spreads. The loss rate for the remaining loan categories is expected to remain little changed relative to 2024.
Over the nine quarters of the stress planning horizon, projected provisions are expected to decrease cumulatively by approximately $35 billion compared to 2024. The decline in provisions exceeds the decrease in loan losses because banks have been building reserves as credit quality normalizes. As a result, banks need smaller reserve builds over the stress test planning horizon.
PPNR is projected to increase… but the Fed could start reducing YoY PPNR volatility in this year’s stress tests.
Based on our models, we expect PPNR projections to increase relative to the 2024 stress test. In aggregate, the 22 firms that participate in this year’s stress tests are projected to generate $403 billion in net revenues over the nine quarters of the planning horizon, exceeding last year’s projections by more than $50 billion, as shown in the left panel of Exhibit 5.
This improvement is primarily driven by enhanced projections of noninterest income and net interest income, partially offset by slightly higher noninterest expenses. Exhibit 5’s right panel shows that the noninterest income boost stems mainly from higher investment banking fees, partly because of the Fed’s models’ backward-looking component and a less severe stress scenario.
Net interest income projections are expected to strengthen slightly in this year’s stress tests, attributed to higher term spreads. Noninterest expense projections show a modest increase.
In response to a reconsideration request last year, the Board instructed Fed staff to explore potential refinements to the PPNR model components. These adjustments aim to reduce the sensitivity of projections to recent periods and decrease year-over-year volatility in PPNR projections. Should the Fed implement these changes this year, we anticipate that improvements in PPNR projections would be significantly tempered. As result, improvements in the maximum decline of capital ratios under stress would be lower than those projected in Exhibit 1.
The GMS is less severe compared to last year’s test…
Banks with significant trading operations are subject to the GMS component in the stress tests. The GMS consists of thousands of large movements in market prices and rates, which are generally calibrated to extreme market moves seen in the second half of 2008 and the period following the Global Financial Crisis.
Exhibit 6 shows the shocks to a set of risk factors across various important asset classes, including corporate bond spreads, mortgage-backed securities spreads, commodity prices, interest rates and the S&P 500 Index. The GMS scenario for 2025 differs from last year’s, particularly regarding the shocks to oil prices, 2-year Treasury rates and 10-year Treasury rates. The shocks to both 2-year and 10-year Treasury rates are less severe than in the prior year’s scenario. Similarly, the shock on equity prices is also milder. Given the significance of these assets classes in banks’ trading portfolios, we anticipate GMS losses to decline further, in addition to the reduction resulting from the removal of private equity shocks from the GMS.
Large banks with substantial trading and custodial operations are also required to incorporate the default of their largest counterparty. For this year’s stress tests, the Fed has excluded certain types of counterparties, including multilateral development banks, which may reduce counterparty losses relative to the previous year’s scenario.
Finally, it is also worth noting that the path of interest rates in the macroeconomic scenario and the global market shock is coherent, which is important to avoid having non-intuitive outcomes in terms of changes in banks’ capital requirements.
We project lower aggregate declines in stressed common equity tier 1 ratios…
Our analysis employs the projections described here to estimate the impact of the severely adverse scenario on the peak decline in each bank’s common equity tier 1 capital ratio under the supervisory stress test. In making these estimates, we must make further assumptions about other significant components of the stress tests, such as trading and counterparty losses, operational risk losses and changes in accumulated other comprehensive income. For the purposes of this analysis, we assume the following:
- Operational risk losses remain unchanged from last year.
- Trading and counterparty losses decrease 5 percent in the severely adverse scenario because of the reduction in shock severity.
- Unrealized gains on available-for-sale securities are reduced by 30 percent because the decline in long-term interest rates is smaller compared to the previous year’s severely adverse scenario.
Under these assumptions, the weighted average common equity tier 1 capital ratio declines by 2.7 percentage points from the start of the projection horizon to its minimum in the 2025 stress test. Last year the decline was 3.2 percentage points.
Conclusion
In summary, the 2025 DFAST stress test scenarios are modestly less severe compared to the previous year, potentially leading to lower projected loan losses and improved pre-provision net revenue for participating banks. The reduced severity in key macroeconomic variables such as commercial real estate prices, equity prices, and corporate bond spreads are the main drivers of lower loan losses.
However, there remains significant uncertainty in these projections, particularly because of potential changes in the Federal Reserve’s approach to modeling pre-provision net revenue. The Fed’s consideration of reducing the dependence on recent performance in PPNR projections, especially for net interest income and noninterest revenues, could substantially moderate the anticipated improvements in stress test results. This adjustment, while aimed at reducing year-over-year volatility in stress test projections, could limit the reduction in banks’ stress capital buffer charges.
[1] In response to one reconsideration request last year, the Federal Reserve Board directed Federal Reserve staff to explore possible refinements to the PPNR model subcomponents to address possible weaknesses, including model over-sensitivity to most recent periods and the impact of non-recurring expenses.
[2] The Fed also made some changes to the set of counterparties subject to the largest counterparty default component by excluding certain sovereign entities and certain multilateral development banks and supranational entities.
[3] Two Category IV banks elected to opt into the 2025 stress tests.
