Liquidity Stress Testing Explained: Scenarios, Assumptions and Survival Horizons

A full walkthrough of liquidity stress testing for UK banks: how the three scenario types work, how to translate shocks into cash flow numbers, how to calculate the survival horizon, and how the results connect to the ILAAP and your liquidity buffer.

Liquidity stress testing is not a box to tick before an ILAAP submission. It is the process that tells a bank what it actually needs to survive a crisis, and that answer drives buffer sizing, funding structure, and where the risk appetite line gets drawn. If your stress testing is calibrated too loosely, your buffers are too thin. Too conservatively and you are holding liquidity that drags on returns for no good reason. Getting it right is a judgement exercise, not a formula, and that is what this post works through.

The Purpose of Liquidity Stress Testing

There is a distinction worth stating clearly before anything else. The contingency funding plan (CFP) tells a bank what it will do when a crisis arrives. Stress testing tells a bank what crisis it should be preparing for in the first place.

That is not a semantic point. It determines how the two tools are governed, who owns them, and when they matter. Stress testing is a planning input. The CFP is an operational response framework. Conflating them produces both a weak stress test and a CFP that is not calibrated to the risks that actually face the firm.

The practical purpose of liquidity stress testing is to answer one question: under a defined set of adverse but plausible conditions, for how long can this firm survive without accessing new external funding? Everything else, the scenario design, the assumption setting, the cash flow modelling, serves that question.

The Three Scenario Types

Banks run three types of liquidity stress scenario, and they run all three for a reason. Each one isolates a different risk, and the combined scenario reveals something neither of the others can show on its own.

Idiosyncratic Stress

This scenario assumes the problem is specific to the firm. A credit rating downgrade. A reputational event that triggers deposit outflows. A large operational loss that becomes public. A failed audit or regulatory censure.

The rest of the market is functioning normally. Wholesale funding markets are open. Counterparties are willing to transact. The stress is entirely the firm's own problem. This scenario tests whether the bank can survive a crisis of confidence in itself without support from the wider system.

Systemic Market Stress

Here the problem is the opposite. The firm is fine but the market is not. Think of a broad sovereign debt crisis, a central bank policy shock, or a rapid repricing of credit risk across the system. Wholesale funding markets close or become prohibitively expensive. Asset values fall. Collateral that was HQLA yesterday is suddenly subject to much larger haircuts.

The firm's own depositors may be calm but its ability to roll wholesale funding with short maturities is impaired. This scenario tests how dependent the bank is on market access and what happens when that access disappears.

Combined Stress

This is the most severe and the most realistic scenario for stress testing purposes. It layers a shock specific to the firm on top of a broad market dislocation, because in practice the two tend to arrive together or accelerate each other. A rating downgrade triggers deposit outflows at exactly the moment wholesale markets are already closed. That interaction is what makes combined stress the scenario most regulators and internal risk functions use to size the buffer.

Running all three is not redundant. Each one reveals different vulnerabilities. A firm that survives idiosyncratic and systemic stress individually but fails in the combined scenario has a concentration problem in its funding structure that only the combined test exposes.

Turning Shocks into Numbers

Scenario design is the conceptual work. Translating scenarios into numbers is where most of the effort sits.

Each shock needs a quantified impact on cash flows. The common ones look like this:

• Deposit run off: retail deposits, SME deposits, and wholesale deposits each carry different assumed outflow rates. Operational deposits behave differently to non operational ones. A stressed assumption might assume 10% of retail deposits leave in the first week, rising to a higher cumulative rate over a 30 day horizon. These rates are not plucked from air; they are informed by the LCR framework, internal historical data, and judgement about the firm's depositor base.
• Funding market closure: assume no new wholesale term funding is raised and no existing facilities can be rolled. All maturing wholesale funding becomes a cash outflow with no corresponding inflow.
• Rating downgrade: a one or two notch downgrade triggers contractual collateral calls under Credit Support Annexes and other bilateral agreements, and may activate additional termination rights under ISDA master agreements. These impacts need to be quantified by the treasury team from actual contract terms, not estimated generically.
• Collateral calls: separate from calls arising from a rating downgrade, market moves can generate variation margin calls on derivatives books. A 200 basis points parallel shift in rates (purely illustrative) might generate a significant cash outflow on a day one basis.
• HQLA haircuts: in a systemic stress, the haircut applied to assets you plan to sell or repo may widen materially. An asset that generates 95p in the pound under normal conditions might generate 80p under stress, reducing the effective size of your buffer.

The output of this modelling is a stressed cash flow projection, produced on a daily or weekly basis, over the survival horizon period. The cumulative net outflow at each point in time is what determines whether the buffer holds.

Survival Horizon and HQLA Buffer Sizing

The survival horizon is the primary output metric of a liquidity stress test. It is the number of days the bank can survive under the stress scenario before its liquidity buffer is exhausted and it is unable to meet its obligations.

A firm looks at that number and asks: is it consistent with our risk appetite? If the risk appetite states the firm should survive at least 30 days of combined stress, and the model shows a 22 day horizon, the firm has a liquidity shortfall. It needs either more HQLA, a better funding structure, or a narrower risk appetite, and those are all real strategic decisions.

This is how the stress test connects to buffer sizing. The minimum HQLA requirement is not simply the LCR minimum times a safety margin. It is the quantity of liquid assets needed to maintain the firm's target survival horizon under its most severe plausible scenario.

From Test Results to Real Decisions

A stress test result that sits in a report and goes nowhere has failed. The point is to change something: the buffer, the funding mix, the concentration limit, or the risk appetite.

In practice the results should feed into three places:

1. Buffer sizing: the stressed cash flow model tells you the minimum HQLA you need to hold to meet your survival horizon target. That feeds directly into the liquidity buffer policy.
2. Funding strategy: if the combined scenario reveals that closure of a specific funding market (say, unsecured wholesale) generates the largest outflow, that tells you something about concentration risk in your funding mix. Diversifying funding sources or extending the tenor of wholesale liabilities is a direct response to what the stress test found.
3. Risk appetite statement: the survival horizon target is a risk appetite parameter. It is the number the board has agreed represents an acceptable level of resilience. Stress testing is what tells the board whether the firm is meeting that standard.

If the output of your stress test is a table of numbers and a conclusion that reads "the bank is sufficiently resilient", you are describing a compliance exercise, not a risk management tool.

Liquidity Stress Testing and the ILAAP

The ILAAP is the internal framework through which a bank demonstrates to the PRA that it holds adequate liquidity (the ILAAP covers liquidity adequacy specifically; capital adequacy sits in the separate ICAAP process). Stress testing is central to it. In fact, the ILAAP stress testing section is where the PRA expects to see whether the firm's internal scenarios are genuinely informative or simply a reformatting of standardised LCR assumptions.

The PRA expects internal scenarios to be specific to the firm, to reflect the actual depositor base and funding structure of the business, and to be more severe than the standardised LCR floor in meaningful ways. A stress test that reproduces LCR run off rates exactly and concludes the firm is fine at its LCR minimum provides no additional information. That is not what the ILAAP is asking for.

For a detailed walkthrough of how the ILAAP fits together and what good liquidity adequacy assessment looks like in practice, see our post on the ILAAP and liquidity adequacy assessment.

Where Judgement Comes In

This is the part that makes liquidity stress testing difficult to industrialise and impossible to fully automate.

The run off rate you apply to a specific deposit segment is not prescribed by a formula for internal stress purposes. You might reference the LCR rates as a floor, look at actual outflow experience from historical stress periods at comparable firms, and then apply judgement about how your depositor base differs. Two firms with similar deposit books might reach different assumptions, and both could be defensible.

That is fine. What matters is that the assumptions are:

• Documented with a clear rationale
• Reviewed and challenged by someone independent of the team that produced them
• Tested for sensitivity (what happens to the survival horizon if the retail run off rate is 15% rather than 10%?)
• Revisited regularly and updated when the balance sheet or market conditions change

Assumption governance is where models become credible or lose credibility with regulators. A stress test with assumptions that were set three years ago, have never been back tested, and are defended with "this is how we have always done it" is a governance problem as much as a risk management one.

Sensitivity analysis matters here because it shows where the model is brittle. If a small change in one assumption flips the outcome from comfortable to stressed, that assumption deserves much more rigorous scrutiny and probably a more conservative calibration.

The practical takeaway is this: document the judgement, challenge the assumptions, and make sure the results are connected to something real. A stress test that changes the buffer, informs the funding plan, and sits in the risk appetite statement is doing its job. One that produces a number and then waits for the next ILAAP cycle is not.

If you want to go deeper on this and related topics, the Academy course catalogue covers liquidity risk, treasury management, and the regulatory frameworks in structured format. The newsletter is a free weekly read covering new content and industry developments. Join free here.