Liquidity stress testing is a simulation a bank runs on itself to see whether it would still have enough cash, and enough assets it could quickly turn into cash, if a funding crisis hit over the next 30 days, 90 days, or year. The exercise applies assumed withdrawal rates to every funding source and discount percentages to every liquid asset, then checks whether the shrunken buffer still covers the projected outflows. In the United States, every bank holding company with at least $100 billion in total consolidated assets has to do this on a defined schedule, and smaller banks are expected to run proportional versions under interagency guidance.
The Problem It Is Designed to Solve
A bank can be solvent on paper and still fail within days. Solvency means assets exceed liabilities; it says nothing about whether those assets can be converted to cash fast enough to pay depositors and creditors who all want their money at once. Before 2008, bank supervision leaned heavily on capital adequacy and gave less attention to that conversion problem. The crisis showed that well-capitalized institutions could collapse when wholesale funding dried up, and the 2023 failures of Silicon Valley Bank and Signature Bank made the point again. Liquidity stress testing exists to force an answer to a blunt question: if conditions deteriorated sharply, could you still pay everyone with a claim on your cash?
The Two Numbers That Drive the Math
Every liquidity stress test runs on two sets of assumptions. One governs how fast money leaves. The other governs how much your assets are actually worth when you try to sell them under pressure.
Run-Off Rates
Run-off rates are the percentage of each funding source assumed to disappear during the stress period. The Basel III framework sets floors that national supervisors can tighten, and the gradient reflects how sticky different kinds of money tend to be.1Bank for International Settlements. LCR40 – Cash Inflows and Outflows
- Stable retail deposits, insured and tied to an established banking relationship: as low as 3% to 5% run-off over 30 days.
- Less stable retail deposits, uninsured or without established relationships: 10% or higher at the supervisor’s discretion.
- Operational wholesale deposits, where the customer depends on the bank for payment services: 25%.
- Non-operational unsecured wholesale funding: up to 100%, meaning the model assumes every dollar walks.
The spread between 3% and 100% explains why regulators pay such close attention to a bank’s funding mix. A bank funded largely by insured retail deposits has a very different stress profile from one that lives on short-term institutional money.
Haircuts
Haircuts are discounts applied to the market value of a bank’s liquid assets, reflecting what those assets would really fetch in a distressed sale. The U.S. Liquidity Coverage Ratio rule sorts high-quality liquid assets into three tiers:2eCFR. 12 CFR 249.20 – High-Quality Liquid Assets
- Level 1: cash, Federal Reserve balances, and U.S. Treasury securities. No haircut.
- Level 2A: debt of U.S. government-sponsored enterprises and certain sovereign or multilateral development bank securities with a 20% or lower risk weight. 15% haircut.
- Level 2B: investment-grade corporate debt, certain municipal securities, and other qualifying assets. 50% haircut.
A $1 billion Level 2B corporate bond holding counts as only $500 million of liquidity buffer. The combination of run-off assumptions on the liability side and haircuts on the asset side determines whether the bank shows a positive or negative liquidity position across the stress horizon.
How a Bank Actually Runs the Test
Federal Reserve rules require internal liquidity stress testing across multiple time horizons: overnight, 30 days, 90 days, one year, and any additional horizons that match the firm’s risk profile.3eCFR. 12 CFR 252.35 – Liquidity Stress Testing and Buffer Requirements The 30-day horizon carries the most weight because it drives the size of the required liquidity buffer under the LCR.
The sequence is straightforward. Analysts start with a complete inventory of liquid assets, broken down by HQLA tier at current market valuations, and apply the appropriate haircut to each. A $10 billion nominal portfolio might shrink to $7 or $8 billion depending on the mix. Then they apply run-off rates to every liability and off-balance-sheet commitment, projecting total cash outflows over the chosen horizon. Expected cash inflows, such as loan repayments and maturing securities, come off that total to yield net cash outflows.
The last step compares the stressed buffer against those net outflows. A positive number means the bank clears the scenario. A negative number pinpoints the day the buffer runs out. That specificity is the point. Knowing you might fail is worth very little; knowing you run out of cash on day twelve because of a $2 billion collateral call tied to a ratings downgrade tells management exactly where to intervene.
Data Behind the Numbers
The test is only as good as the inputs. On the asset side, the bank needs every liquid position classified by HQLA level with current valuations. On the liability side, every funding source needs a maturity schedule showing exactly when cash leaves: deposits by type (insured versus uninsured, retail versus wholesale), secured borrowings, long-term debt maturities, interbank obligations. Off-balance-sheet commitments matter because unused credit lines extended to corporate clients tend to get drawn down during a crisis, at the worst possible moment. Historical deposit behavior helps calibrate how customer segments move under pressure, though 2023 showed those historical patterns can badly understate how fast modern deposit runs happen.
Scenarios Banks Are Expected to Run
Three broad categories cover the forward-looking tests, and most banks are expected to run all three. The interactions are usually more informative than any one scenario in isolation.
- Idiosyncratic scenarios: trouble specific to one bank, such as a ratings downgrade that triggers collateral calls and chokes off unsecured funding, while the broader market works normally.
- Market-wide scenarios: systemic shocks that hit everyone, such as a credit market freeze, a sharp asset-price drop, or a rate spike that devalues bond portfolios.
- Combined scenarios: firm-specific trouble coinciding with a market-wide crisis. The most demanding, and the most revealing, because a bank losing confidence during a panic has nowhere to turn.
Reverse Stress Testing
A traditional stress test starts with a scenario and calculates the damage. Reverse stress testing flips that: it starts with the catastrophic outcome, usually the bank being unable to meet obligations, and works backward to identify what combination of events could produce it.4Federal Reserve. Interagency Supervisory Guidance on Stress Testing for Banking Organizations It catches blind spots a forward-looking test misses, like a hidden concentration in uninsured deposits from a single industry or an unnoticed correlation between two funding sources. Interagency guidance treats it as a complement to traditional testing, not a replacement.
What Happens With the Results
The Contingency Funding Plan
A stress test number sitting on a page accomplishes nothing. Results feed into the bank’s Contingency Funding Plan, which spells out specific actions if a liquidity crisis materializes: which assets get sold first, which borrowing facilities get tapped (including the Federal Reserve’s discount window), and when new lending gets restricted. Regulators expect these plans to be actionable rather than theoretical, and the 2023 addendum to the Interagency Policy Statement on Funding and Liquidity Risk Management emphasized that plans must be periodically tested in practice.5FDIC. Updated Guidance: Interagency Policy Statement on Funding and Liquidity Risk Management Pre-positioning collateral at the discount window is one specific push, so a bank can actually borrow in an emergency instead of discovering operational barriers mid-crisis.
Reporting
Large banks submit liquidity data to the Federal Reserve on the FR 2052a Complex Institution Liquidity Monitoring Report, which captures assets, liabilities, funding activities, and contingent liabilities on a consolidated basis and by material entity.6Federal Reserve Board. FR 2052a Covered institutions also publish quarterly LCR disclosures, including HQLA composition, cash outflows and inflows by category, and a qualitative discussion of what moved the ratio.7Federal Register. Liquidity Coverage Ratio: Public Disclosure Requirements Results go through internal governance review by senior management and the board before submission, and that review is supposed to pressure-test the assumptions rather than rubber-stamp the output.
Who Has to Do It
The core U.S. framework comes from Section 165 of the Dodd-Frank Act, which directs the Federal Reserve to impose enhanced prudential standards on large financial institutions.8FDIC. Selected Sections of the Dodd-Frank Wall Street Reform and Consumer Protection Act In 2019, the Fed finalized a tailoring framework that sorts large banks into four categories and scales the obligations accordingly.
- Category I: U.S. global systemically important banks. Full LCR, internal liquidity stress testing, and full reporting.
- Category II: $700 billion or more in assets, or $100 billion in assets with $75 billion or more in cross-jurisdictional activity. Same full requirements as Category I.
- Category III: $250 billion or more in assets, or $100 billion in assets with $75 billion or more in short-term wholesale funding, nonbank assets, or off-balance-sheet exposure. Full LCR if short-term wholesale funding exceeds $75 billion; otherwise a reduced LCR calibrated at 85%.
- Category IV: $100 billion or more in assets, not meeting higher-category criteria. Reduced LCR at 70% if short-term wholesale funding is at least $50 billion, no LCR below that, and still required to run internal liquidity stress tests.
All four categories file the FR 2052a. The OCC enforces parallel stress testing requirements for national banks, with a $250 billion threshold for company-run stress tests.9Office of the Comptroller of the Currency. OCC Bulletin 2019-47 – Amendments to the Stress Testing Rule for National Banks and Federal Savings Associations: Final Rule10Bank for International Settlements. Basel III: The Liquidity Coverage Ratio and Liquidity Risk Monitoring Tools11Bank for International Settlements. Basel III: The Net Stable Funding Ratio
Banks below $100 billion are outside the formal LCR and tailoring categories, but they are not exempt from the concept. The Interagency Policy Statement on Funding and Liquidity Risk Management, with its 2023 addendum, expects every supervised bank to conduct liquidity stress testing proportional to its size and complexity and to keep an actionable contingency funding plan.5FDIC. Updated Guidance: Interagency Policy Statement on Funding and Liquidity Risk Management A community bank’s version looks lighter than a trillion-dollar institution’s, but the underlying question is the same.
When the Test Is Done Badly: Silicon Valley Bank
Silicon Valley Bank’s collapse in March 2023 worked as a live demonstration of what stress testing is supposed to catch and what happens when it does not. The Federal Reserve’s post-mortem found that SVB’s primary internal stress scenario did not adequately stress its liquidity exposures, borrowed deposit behavior assumptions from dissimilar peer institutions, and modeled a slow deterioration rather than a sudden run.12Federal Reserve. Review of the Federal Reserve’s Supervision and Regulation of Silicon Valley Bank
The deposit segmentation also missed the attributes that mattered. SVB held a very high share of large, uninsured deposits concentrated in venture-capital-backed technology companies, a tightly networked client base that coordinated withdrawals at a speed no historical model had contemplated. Supervisors had flagged some of these issues through Matters Requiring Attention but still characterized the bank’s overall liquidity position as sufficient well into 2022.
The review called for reconsidering stability assumptions for uninsured deposits, revisiting how held-to-maturity securities are treated in standardized liquidity rules, and potentially extending liquidity requirements to a broader set of firms. The underlying lesson for anyone trying to understand how liquidity stress testing works: the mechanics are only as good as the assumptions fed into them, and assumptions calibrated to calm periods can fail dramatically when the panic they were meant to simulate actually arrives.