How Bank Stress Tests Work

Every year, the Federal Reserve runs bank stress tests on large U.S. lenders to verify they can survive a financial crisis without collapsing. These tests use standardized economic scenarios—deep recessions, unemployment spikes, asset price crashes—to measure whether a bank’s capital would hold up. Failure means higher capital buffers, no dividends, no buybacks, and potential regulatory action.

The Origins and Dodd-Frank Mandate

Bank stress testing in its modern form was born during the 2008 financial crisis. In March 2009, the Federal Reserve announced that it would run “Supervisory Capital Assessment Program” (SCAP) tests on the 19 largest U.S. banks to model their losses if the recession worsened. The purpose was defensive: to show the public that banks could survive and avoid a second wave of panic and bank runs.

The SCAP succeeded in that limited goal. It gave banks and regulators a common language for thinking about solvency under duress. After the crisis, Congress embedded stress testing into the Dodd-Frank Act, making it mandatory for large banks going forward. The Federal Reserve converted SCAP into two permanent programs: CCAR (Comprehensive Capital Analysis and Review) and DFAST (Dodd-Frank Annual Stress Test), which run in tandem each spring.

What the Tests Measure

Both tests aim to answer one question: if a severe economic downturn occurs, will this bank retain enough capital to continue lending and avoid insolvency?

The tests work by simulating losses across a bank’s portfolio under hypothetical stress scenarios. A bank plugs its actual loan book, investment portfolio, trading positions, and operations into a model, then applies shock scenarios (recession, unemployment spike, asset price declines) and measures whether capital shrinks below critical thresholds.

CCAR is the more comprehensive assessment. It combines the stress test with a capital plan review, in which a bank proposes its intended dividend payments and share repurchases for the coming year. The Fed stress-tests the bank, then checks whether the bank’s planned capital actions are prudent given the stress results. If the bank is stressed to near-failure, it should not be paying out dividends and buying back stock; it should be hoarding capital.

DFAST is the statutory minimum: the Fed stress-tests all banks with $100 billion+ in assets using three scenarios and publishes the results, institution by institution, so the public can compare resilience.

The Scenarios

The Federal Reserve publishes three scenarios each year:

1. Baseline: A mild recession, maybe 1–2% GDP decline, unemployment rising slowly. This is the “nothing goes wrong” case.

2. Adverse: A sharper shock. GDP falls 3–5%, unemployment rises by 3–4 percentage points, house prices decline by 15–25%, long-term interest rates rise. This is a bad-but-not-unprecedented recession.

3. Severely adverse (only for CCAR): A crisis-level shock. GDP falls 5–7%, unemployment spikes by 4–5 points, house prices crash by 30%, stock indices fall by 40%, credit spreads widen sharply, short-term rates fall to zero. This is meant to approximate 2008 or worse.

Additionally, the Fed includes idiosyncratic risks tailored to each bank: higher loan loss rates for credit card issuers, greater trading losses for investment banks, larger deposit flight for regional lenders. A bank with heavy commercial real estate exposure faces worse real estate shocks; a bank with large trading books faces larger market-shock losses.

The bank runs its balance sheet and income statement through all three scenarios over a 9-quarter horizon (roughly 2 years). For each scenario, the model calculates:

• Loan loss provisions (reserves set aside for expected defaults)
• Trading losses (mark-to-market losses on derivatives and securities)
• Other income shocks (net interest margin compression, operational losses)
• Capital depletion (losses reduce equity)

Capital Ratios Under Stress

The tests focus on two capital ratios:

1. Tier-1 capital ratio: The ratio of core equity and retained earnings to risk-weighted assets. Minimum requirement is typically 4.5%, but banks must stay above 6% to avoid “capital conservation buffers.” Systemically important banks (the “too big to fail” firms) must maintain 8–10% Tier-1 capital to avoid being declared failing.

2. Common Equity Tier-1 (CET1) ratio: An even stricter metric—only the hardest equity counts. Minimum is 4.5%, but the conservation buffer puts the effective minimum at 5.5–6%.

During the severely adverse scenario, a large bank’s Tier-1 ratio might decline from 12% to 8%, meaning it is still solvent but has burned through half its excess capital. If the ratio drops below 5.5%, the bank fails the stress test.

Publication and Consequences of Failure

The Fed publishes DFAST results publicly in June, showing each bank’s stressed capital ratios under each scenario. This is a transparency measure: investors, regulators, and competitors can all assess banks’ resilience.

Failing CCAR has teeth. If a bank’s capital ratio falls below the minimum threshold under the severely adverse scenario, the Fed will:

1. Deny the bank’s proposed dividend increase
2. Reject or impose conditions on share repurchases
3. Require the bank to revise and resubmit a capital plan
4. In extreme cases, issue a formal enforcement action or restrict the bank’s expansion into new markets

Failing DFAST is technically less binding (DFAST is advisory), but a bank that fails DFAST typically also fails CCAR. The reputational damage is severe, and investors punish bank stocks that reveal low stress resilience.

Banks that pass comfortably can proceed with capital actions. Large banks that pass by narrow margins typically hold capital or increase it slightly (retaining earnings rather than paying dividends) as a cushion against next year’s test being harsher.

The Empirical Results: What Banks Have Learned

Since 2009, stress testing has reshaped bank behavior. Most large U.S. banks now maintain capital ratios well above the minimum, a shift toward conservatism that would have been unthinkable pre-2008. Banks have also reduced their holdings of the riskiest assets (e.g., off-balance-sheet securitized derivatives) and have simplified their business models to reduce tail risks.

The 2020 CCAR test was notable: the Fed ran a baseline, adverse, and severely adverse scenario including a COVID-19 shock (unemployment spiking to 10%, house prices falling, equity markets crashing 35%). All 33 banks passed, though some with single-digit excess capital. The results proved that post-2008 reforms had worked—even a pandemic could not trigger bank failures, though the Fed had to backstop credit markets to prevent a financial panic.

Criticisms and Limits

Critics argue that stress testing, while well-intentioned, has become rote and formulaic. The same three scenarios run each year; banks learn to optimize against them. A truly novel crisis (like a cyberattack or climate-driven insurance collapse) might not be captured in a standard stress test. Additionally, stress testing captures credit risk and market risk well, but operational risk and reputational risk are harder to model, so they receive less weight.

There is also debate over whether stress tests are stringent enough. A 5–7% GDP decline is painful but historically mild compared to recessions of the 1930s (GDP fell 25%) or even 1980s (double-digit unemployment). If the Fed calibrated a stress test to a true “tail risk” scenario (a 15% GDP decline), most banks would fail, which would make the test less credible.

Finally, stress testing focuses on solvency (can the bank survive losses?) but not liquidity (can the bank get cash when it needs it?). A bank with adequate capital but no access to funding can still fail. The 2008 crisis was largely a liquidity crisis: banks had assets but could not borrow. Stress testing partially addresses this via liquidity coverage ratios, but the focus remains on capital adequacy.

See also