How do stress tests assess bank resilience?

The short answer

Stress testing applies macro and market shocks — recession, equity crash, real estate decline, credit losses — to bank balance sheets and projects how capital ratios evolve under those conditions. The objective is to verify whether banks have sufficient loss-absorbing capital to survive a severe scenario without taxpayer support.

The Federal Reserve’s Dodd-Frank Act Stress Test (DFAST) is the US flagship: large banks submit projections under a Fed-designed severely adverse scenario over nine quarters. Banks failing to maintain minimum capital ratios face restrictions on dividends and share buybacks. The 2024 results showed aggregate CET1 declining from 12.7% to a minimum of 9.9% — a 2.8 percentage point drop, larger than the 2.5 point decline in 2023.

The 2023 regional bank episode exposed structural limitations: SVB-class banks were below the post-EGRRCPA threshold, and the standard scenarios did not test rapid digital deposit outflow combined with HTM mark-to-market losses.

→ New to bank supervision? Systemic fragilities pillar

What the data shows

The DFAST framework and its evolution (Federal Reserve, BPI, public bank disclosures):

• 2024 DFAST aggregate result: CET1 ratio projected to decline from 12.7% to a minimum of 9.9% over nine quarters (Q1 2024 through Q1 2026)
• 2024 severity: 2.8 percentage point CET1 decline, larger than 2023’s 2.5 point projection, attributed to credit card delinquencies and corporate credit deterioration
• 2024 severely adverse scenario assumptions: US unemployment rising 6.3 percentage points to peak 10% in Q3 2025, severe global recession with commercial and residential real estate declines
• EGRRCPA threshold (2018-2023): banks below $250bn in assets faced lighter stress-testing requirements. SVB ($209bn) sat just below this threshold
• 2024 exploratory scenarios: Fed introduced four new exploratory scenarios assessing funding stress combined with rising rates and severe global recession — direct response to 2023 episode
• Coverage scope: roughly 32 large US bank holding companies subject to DFAST, accounting for the bulk of US banking assets

The exception that complicates the framework: stress tests are designed for capital adequacy under losses, not for liquidity adequacy under deposit runs. The 2023 failures were liquidity events, not capital events — the tests were structurally not designed to catch them.

→ Dataset: Credit spreads dataset

Why it happens — the macro mechanism

Stress testing addresses bank fragility through three reinforcing mechanisms.

The capital adequacy channel. By projecting losses under severe scenarios, stress tests verify that capital buffers can absorb shocks. The Fed’s stress capital buffer mechanism translates DFAST results into binding capital requirements. Banks that fail face restrictions on capital distributions until ratios are restored.

The behavioral incentive channel. The prospect of failing the test incentivizes banks to build capital pre-emptively, hedge duration risk, and limit risky exposures. Critics argue that this drives banks to optimize against the test scenarios rather than against real-world risks. Contrary to the assumption that stress tests reduce risk uniformly, they sometimes shift risk to areas not covered by scenarios — a form of regulatory arbitrage.

Bridging point: the third mechanism concerns scenario design and its limits.

The scenario design channel. Pre-2024 DFAST scenarios focused on credit losses, market shocks, and gradual macro stress over nine quarters. They did not test rapid liquidity drain combined with held-to-maturity mark-to-market losses. SVB failure details how this scenario gap, combined with the EGRRCPA supervisory exemption, left SVB-class banks essentially un-stress-tested for the conditions they actually faced. This is the angle distinctive — even G-SIBs would not have been caught by pre-2024 scenarios for a 36-hour digital run.

Synthesis by regime: in the post-Dodd-Frank regime (2010-2017), DFAST applied broadly to banks above $50bn in assets and focused on credit and market risk under recession scenarios. In the post-EGRRCPA regime (2018-2023), the threshold rose to $250bn, exempting mid-sized banks like SVB and Signature. In the post-SVB regime (2024 onward), exploratory scenarios add funding stress and accelerated deposit outflow assumptions, beginning to address the 2023 gap — but the formal capital requirements still derive from the traditional severely adverse scenario.

Stress tests measure resilience against the scenarios they design — and pre-2024 scenarios did not include the digital-run-plus-HTM-losses combination that brought down SVB.

→ Framework: Systemic fragilities

What it means for different economic actors

Bank shareholders. DFAST results determine the stress capital buffer, which directly limits dividend and buyback capacity. Banks passing comfortably can return more capital; banks with thinner buffers face tighter restrictions. Stress test outcomes are now major equity-price catalysts at the annual June release.

Bank creditors. Senior debt and uninsured depositors benefit from stress-test-driven capital buffers, which provide a first line of defense before their claims are at risk. The 2024 results showing 9.9% minimum aggregate CET1 indicate substantial buffer capacity at large banks under severe scenarios.

Macro analysts. The Fed’s exploratory scenarios introduced in 2024 signal a regulatory shift toward incorporating liquidity-driven stress. This is informative about which fragilities supervisors now consider material — funding stress, deposit outflow speed, and HTM duration losses are now formally on the assessment menu.

A common error is to treat stress tests as a comprehensive resilience check. They cover the scenarios designed; gaps in scenario design produce gaps in resilience verification. The 2023 episode demonstrated that the scenario-design dimension is at least as important as the headline result.

Practical observation

Go deeper

Related questions

• Why did SVB fail so quickly in March 2023?
• What are too-big-to-fail banks and the moral hazard?
• How did the 2023 regional bank crisis differ from 2008?
• What is the discount window stigma problem?

Frequently asked questions

Why didn’t stress tests catch SVB’s vulnerability?

Two reasons. First, SVB was below the EGRRCPA $250bn threshold and faced lighter testing requirements. Second, even at the G-SIB level, pre-2024 stress scenarios were designed around traditional recession dynamics — credit losses, market crashes, gradual macro deterioration — and did not include rapid digital deposit outflow combined with held-to-maturity duration losses. The combination that brought down SVB was not in the scenario design. The 2024 exploratory scenarios began addressing this gap.

What is the angle distinctive about scenario design as a resilience determinant?

Most discussion of stress tests focuses on whether banks pass or fail. The deeper question is what scenarios test against. A bank passing a scenario that excludes its dominant fragility provides false comfort. The 2023 episode showed that the scenario-design dimension can be as important as the capital adequacy dimension — a bank with a 12% CET1 ratio under traditional scenarios can still fail in 36 hours if its dominant risk is rapid uninsured deposit outflow that the scenarios do not test. This is the structural lesson supervisors are still incorporating.

How does the European stress test framework compare?

The European Banking Authority conducts a biennial EU-wide stress test on the largest banks, with the European Central Bank running parallel exercises for euro-area institutions. The framework is methodologically similar to DFAST but applied less frequently and with somewhat different scenario assumptions. The Credit Suisse failure in March 2023 — a bank that had passed multiple stress tests — illustrated that European frameworks face the same scenario-design limitations as US ones, particularly around liquidity risk and rapid funding withdrawal.

Last updated — 1 June 2026