What Is Bank Stress Testing? Scenarios, Capital Math & the SCB

A stress test asks whether your capital could survive a really bad year. Value at Risk, or VaR, asks how much you could lose on a normal bad day, based on recent market history. A stress test uses a severe but plausible scenario and plays it out over nine quarters. This post walks through the spirit of the rules, the basic math, and what it means for a regional bank, using a fictional example: Ridgeway Regional Bank.

<iframe width="560" height="315" src="https://www.youtube.com/embed/jqbS44TobUY" title="What Is Bank Stress Testing? Scenarios, Capital Math & the SCB" frameborder="0" allowfullscreen></iframe>

Watch on YouTube

Meet Ridgeway, and why stress tests exist

Ridgeway Regional Bank is fictional, and we use it for teaching. It has about $9 billion in assets and $6 billion of loans, half of it commercial real estate. Its risk-weighted assets are $6.6 billion, and it holds $792 million of common equity tier 1 capital, or CET1, a CET1 ratio of 12%.

Why do stress tests exist? In 2009, during the crisis, U.S. regulators tested the 19 largest banks, and told 10 of them to raise $74.6 billion. In 2010, the Dodd-Frank Act wrote stress tests into law. In 2020, the Fed began using results to set each big bank's capital buffer. And in 2023, Silicon Valley Bank failed after bond losses and a one-day run of $42 billion. The spirit: prove capital survives a severe downturn before it happens, so the bank can keep lending.

Base, bad, and brutal

Most banks run three scenarios. Remember them as base, bad, and brutal. Baseline is what forecasters expect. Adverse is a moderate recession. Severely adverse is a deep recession, severe but plausible. For scale, the Fed's 2026 severely adverse scenario had unemployment peaking at 10%, house prices down 30%, and commercial real estate down 39%.

What goes in is the scenario: unemployment and GDP, interest rates, and house and commercial property prices. What you project is loan losses by type of loan, revenue (interest income minus costs), and finally the capital ratio, quarter by quarter.

The math: start, earn, burn

End capital equals start, plus earnings, minus losses.

Ridgeway starts with $792 million of CET1. Over nine brutal quarters it still earns $132 million. But it burns $429 million: $396 million of loan losses, plus $33 million of other losses. That leaves $495 million, or 7.5% of risk-weighted assets.

The earn part has a name: pre-provision net revenue. It's net interest income plus fees, minus expenses, before any loan losses. In a normal economy, Ridgeway would earn about $340 million over nine quarters. In the brutal scenario that falls to $132 million, because rates drop, fewer customers borrow, and fees weaken. That missing $208 million never shows up as a loss, but it costs capital just the same.

Where do the loan losses come from?

• Commercial real estate: $3 billion at a 9% loss rate is $270 million.
• Business loans: $1.4 billion at 5% is $70 million.
• Home mortgages: $1 billion at 2% is $20 million.
• Consumer loans: $600 million at 6% is $36 million.

The total is $396 million, and commercial real estate is 68% of it.

Does Ridgeway pass?

Its low point is 7.5%, a drop of 4.5 points. The regulatory minimum for CET1 is 4.5%. Add the 2.5% conservation buffer and you get 7%. Below that line, dividends and bonuses get limited. Ridgeway stays above both lines, but only half a point above the buffer line. A pass, with little room to spare.

The stress capital buffer

For banks with $100 billion or more, the Fed turns the test into a requirement: the stress capital buffer, or SCB. It's the CET1 drop from start to low point, plus four quarters of planned dividends, with a floor of 2.5%. Ridgeway is too small for this rule, but if it were tested, its 4.5 point drop plus dividends of about 0.7% would give a buffer of 5.2%. Its CET1 requirement would be 4.5% plus 5.2%, or 9.7%.

Who gets tested

As of mid-November 2026, banks with $100 billion or more take the Fed's test: the largest every year, Category IV banks every other year. In June 2026, the Fed tested 32 banks, and all of them stayed above their CET1 minimums. Together they absorbed more than $708 billion of hypothetical losses, about $200 billion of it from credit cards. Yet their combined CET1 ratio fell only 1.6 points, from 12.8% to 11.2%. Compare that with Ridgeway's illustrative drop of 4.5 points. A bank packed into commercial real estate can lose far more capital in a similar storm.

On September 30, 2026, the Fed finalized rules to publish its scenarios and models for comment, and to average two years of results.

And banks under $10 billion, like Ridgeway, have no required Dodd-Frank test, but supervisors still expect some stress testing.

How a community bank can run the test

For a bank Ridgeway's size, the OCC's community bank guidance describes four methods. Transaction level: re-run your biggest loans with lower cash flow and lower property values. Portfolio level: shock a whole segment, like office loans. Enterprise-wide: credit, rates, and liquidity together, for the whole bank. And reverse stress testing, which works backward from broke. The OCC says even a single spreadsheet can work, if it looks out at least two years.

2023 added a lesson. Rates rose fast, and long-dated bonds lost value. At Silicon Valley Bank, depositors pulled about $42 billion on March 9. The next day, about $100 billion more was queued, and the bank was closed. So in your own test, add a rate shock and a deposit run to the recession, at the same time.

A reverse stress test asks what it would take to hit the minimum. Ridgeway can lose $792 million minus 4.5% of $6.6 billion, or $495 million. Add back $132 million of earnings, subtract $33 million of other losses, and the breaking point is $594 million of loan losses, about 9.9% of all loans. Then ask what story would cause that. A fifty percent drop in office values?

Here's how Ridgeway runs its own test. One, pick three scenarios: base, bad, and brutal. Two, turn them into drivers like unemployment, rates, and property prices. Three, project losses, revenue, and capital for nine quarters. Four, compare the low point with 4.5% and 7%. And five, act: set limits, update the capital plan, and keep a contingency funding plan ready.

Who owns the test, and five ways it goes wrong

The board sets the risk appetite and approves the scenarios and the capital floor. Management runs it, with treasury, credit, and finance working together. Model risk management challenges the models and the assumptions, independently. And internal audit checks that the process was followed and written down. If one of those four is missing, the numbers get much easier to bend.

And here are five ways stress tests go wrong:

1. The scenario is too mild to hurt.
2. The balance sheet stays frozen, with no deposit outflows.
3. It stresses credit only, with no rate shock and no liquidity squeeze.
4. Nobody challenged the models.
5. The results never changed a single decision.

What this means for a regional bank

First, a stress test is a management tool before it's a compliance exercise. Second, concentrations drive results. At Ridgeway, commercial real estate is 68% of stressed losses. And third, stress capital, liquidity, and interest-rate risk together, not in silos.

Five questions to ask your team

• What's our CET1 low point in the brutal scenario?
• How close does it come to the 7% buffer line?
• Which portfolio drives most of the losses?
• Did we stress rates and deposits along with credit?
• And what loss would break us, and how likely is that story?

Recap

VaR is a bad day, and a stress test is a bad year. There are three scenarios: base, bad, and brutal. The math is start, earn, burn, and Ridgeway goes from 12% to 7.5%. For big banks, the result sets the stress capital buffer. And a reverse stress test works backward from broke.

Watch more Bank Risk Decoded videos on the Bank Risk Decoded YouTube channel.