Liquidity Risk Management: Stress Testing and Contingency Planning
7/7/2026
Interest rate risk damages earnings. Liquidity risk closes institutions. That asymmetry justifies treating liquidity as a discipline in its own right rather than as a section of the asset/liability review — which is how most community banks handle it, and where our companion post on asset/liability management treats it.
The events of 2023 changed the operative assumption in this discipline, and many plans have not been updated to reflect it.
What Changed: Speed
The traditional liquidity plan assumed that a deposit outflow takes time. Customers learn something, consider it, visit a branch or call, and move money over days or weeks. That assumption underpinned the scenario severities in most contingency funding plans.
It is no longer true. Money moves instantly, information moves faster, and depositor behavior is now correlated in ways that pre-date-digital models did not capture. An institution whose plan assumes a ten percent outflow over thirty days is modeling a world that no longer exists.
Three consequences worth building into the plan:
Compress the horizon. Scenarios should include what happens in a day and in a week, not only over a month.
Assume correlation. Depositors in the same segment, learning the same information, act together. A plan assuming independent behavior across a segment understates the outflow.
Assume the response window is short. Actions that take a week to execute are not available in a scenario that develops in two days.
Sources of Liquidity, Honestly Assessed
The exercise that matters is not listing sources. It is stating what is genuinely available, how fast, and at what cost.
Cash and due from banks. Immediate, and typically small relative to a stress.
Unencumbered securities. The key word is unencumbered. Securities pledged for borrowings or public deposits are not available, and securities classified as held-to-maturity are practically constrained even where legally saleable. As covered in the interest rate risk post, institutions frequently count the same securities twice — once as a liquidity source and once as collateral.
Loan cash flows. Scheduled principal and interest, reliable in normal conditions and slower in stress as borrowers draw lines and delay.
Loan sales and participations. Real but slow, and the market for what an institution wants to sell is worst exactly when it needs to sell.
Borrowing capacity. Federal Home Loan Bank advances, correspondent lines, and the Federal Reserve's discount window. Two disciplines: capacity should be tested by actually drawing, not assumed from a line agreement, and collateral should be pre-positioned, since pledging collateral during a stress takes time the institution will not have.
Wholesale and brokered deposits. Available, priced, and the least reliable in a stress — access frequently disappears precisely when needed, and heavy reliance is itself a supervisory concern.
The output should be a single number the board can hold: counterbalancing capacity available within one day, one week, and one month, net of what is encumbered.
Deposit Segmentation
The most valuable liquidity analysis a community bank can perform, and the one most often skipped.
Segment the deposit base by:
Insured versus uninsured. Uninsured balances leave fastest and in the largest amounts. The proportion and concentration of uninsured deposits is the single most informative liquidity statistic an institution has.
Operating versus rate-sensitive. A business's payroll account behaves nothing like a balance that arrived chasing a promotional rate.
Concentration. How much of the deposit base sits with the ten largest relationships, and what industry or sector are they in.
Channel. Branch-acquired retail behaves differently from digitally acquired or rate-comparison-sourced balances.
Sector correlation. Deposits from a single industry — as the crypto-sector episode demonstrated — can move together for reasons unrelated to the institution.
The purpose is to be able to say, with evidence, what proportion of the deposit base would plausibly leave in a stress and how quickly. An institution that cannot segment its deposits cannot stress test them meaningfully.
Stress Scenarios
Three families, and a plan needs all three.
Idiosyncratic. Something about this institution — a loss announcement, an enforcement action, negative local publicity, a downgrade, a rumor. Characteristically fast and concentrated in uninsured and rate-sensitive balances.
Market-wide. A sector event, a rate shock, or a general flight to perceived safety. Wholesale access tightens for everyone simultaneously.
Combined. Both at once, which is the scenario that actually damages institutions and the one most often omitted because it looks unfair.
For each, specify the outflow assumption by deposit segment rather than as a single percentage, the horizon in days, what counterbalancing capacity is genuinely available in that window, and where the institution runs out.
The output should identify the survival horizon: how many days the institution can meet obligations under each scenario. That single number is more useful to a board than any ratio.
The Contingency Funding Plan
Most contingency funding plans fail the only test that matters: could someone execute this under pressure?
What makes one usable:
Triggers, quantified. Not "significant deposit outflow" but specific thresholds on named indicators that move the institution between defined stages.
Stages with escalating actions. Normal, heightened monitoring, contingency, and crisis — with the actions available at each stage listed in the order they would be taken.
An action menu with real detail. Which securities would be sold, in what order, at what expected loss. Which borrowings would be drawn, in what sequence, and what collateral supports them. What deposit pricing action would be taken. Which loans could be sold, to whom.
Named roles, including who declares a stage change, who contacts the regulator, who speaks to the media, and who communicates with large depositors.
Contact lists for the FHLB, correspondents, the Federal Reserve, brokers, counsel, and the primary regulator — with after-hours numbers.
Regulator communication planned in advance. Notifying the primary federal regulator early in a developing liquidity situation is materially better than being discovered, and the plan should say who does it and when.
Testing. A tabletop exercise at least annually. Institutions that run one routinely discover that the plan's first action depends on a system, a person, or a facility that is not actually available.
Early Warning Indicators
Indicators, with thresholds and owners, that precede a liquidity problem:
Deposit balance trends by segment; uninsured deposit proportion; the ten-largest-depositor concentration; loan-to-deposit ratio; reliance on wholesale funding; unencumbered securities as a proportion of assets; utilization of borrowing capacity; deposit rate paid relative to market; unusual outflow velocity or large withdrawal requests; credit rating or peer comparison movement; and — worth naming — local or social media sentiment, which now precedes outflow rather than following it.
Governance
Liquidity limits should be board-approved, expressed in terms the board understands, and include a minimum survival horizon under stress rather than only ratio floors.
Reporting should include the segmentation, the survival horizon by scenario, counterbalancing capacity net of encumbrance, indicator status, and any stage change under the contingency plan. A liquidity report that presents a loan-to-deposit ratio and an on-hand liquidity percentage has not told the board what it needs.
Structured coverage is available through the Certificate in Risk Management, Financial Risk Management: Interest Rate Risk, and the Asset Management Certificate Program.
Where Plans Fail
Double-counted collateral — securities treated as both a liquidity source and pledged.
Untested borrowing capacity — a line agreement is not capacity until it has been drawn.
Collateral not pre-positioned, so accessing capacity requires time the institution does not have.
Single-percentage outflow assumptions rather than segment-level analysis.
Horizons that are too long, reflecting pre-digital deposit behavior.
No combined scenario, because it seems unrealistic.
A plan nobody has read, with contact lists years out of date.
No regulator communication plan, which converts a manageable situation into a supervisory event.
Communication Is a Liquidity Control
Most contingency funding plans treat communication as an appendix. In a fast-developing situation it is one of the few levers that actually slows an outflow, and it has to be prepared in advance because there is no time to draft it under pressure.
Four audiences, each needing a different thing.
Large depositors. The relationships representing the largest uninsured balances will hear something and will call — or will not call and simply move. A named relationship officer for each, with a factual message and the authority to have the conversation, is worth more than any public statement. Institutions that have never identified who calls which depositor discover during a stress that nobody does.
Staff. Front-line employees will be asked by customers what is happening, and in the absence of guidance they will speculate or appear evasive — both of which accelerate the problem. A short, accurate, approved statement given to staff early is a control.
The regulator. Covered above, and worth repeating because the instinct is to wait until the picture is clear. Supervisors respond considerably better to early notification of an uncertain situation than to late notification of a clear one.
The public and media, where the operative discipline is that only named individuals speak, the message is factual, and nothing is said that cannot be substantiated. Reassurance that later proves wrong is worse than silence.
Two preparation items make this executable. Draft the holding statements in advance — for a rumor, for a genuine outflow, for a technology outage mistaken for insolvency — and have counsel review them while nobody is under pressure. And decide in advance what the institution will not say, because the most damaging communications in these situations are specific reassurances offered spontaneously by someone trying to help.
The underlying point is that a liquidity event is partly an information event. The balance sheet determines how much time the institution has; the communication determines whether the outflow accelerates or stabilizes within that time.
A final point on the relationship between liquidity and the rest of the balance sheet. Liquidity problems are usually the visible end of something that began elsewhere — credit deterioration that raised questions, an interest rate position that produced losses, a concentration that moved together, or an operational failure that damaged confidence. That is why the enterprise view described in the risk management framework post matters here specifically: an institution monitoring liquidity in isolation is watching the symptom. The useful discipline is to ask, whenever another risk deteriorates, what it would do to funding — because depositors do not distinguish between categories of bad news.
Frequently Asked Questions
What changed about liquidity risk in 2023?
Speed. Traditional plans assumed deposit outflows develop over days or weeks; money now moves instantly, information moves faster, and depositor behavior within a segment is correlated. Plans should include one-day and one-week horizons, assume correlated rather than independent behavior, and recognize that actions taking a week to execute are unavailable in a scenario developing in two days.
What is the most informative liquidity statistic for a community bank?
The proportion and concentration of uninsured deposits. Uninsured balances leave fastest and in the largest amounts, so an institution's exposure is largely a function of how much of its funding sits above the insurance limit and how concentrated those balances are among a small number of relationships.
Why must borrowing capacity be tested rather than assumed?
Because a line agreement is not capacity until it has been drawn. Institutions routinely discover during a stress that accessing capacity requires collateral pledging, documentation, or approvals that take time they do not have. Collateral should be pre-positioned in advance for the same reason.
What makes a contingency funding plan usable?
Quantified triggers on named indicators, defined stages with escalating actions listed in the order they would be taken, an action menu specifying which securities would be sold at what expected loss and which borrowings drawn in what sequence, named roles including who contacts the regulator, current after-hours contact lists, and an annual tabletop exercise.
Should a bank notify its regulator during a developing liquidity situation?
Yes, early. Notifying the primary federal regulator while a situation is developing is materially better than being discovered, and the contingency funding plan should name who makes that contact and at what stage. Institutions that wait convert a manageable position into a supervisory event.
What is a survival horizon and why does it matter?
The number of days the institution can meet its obligations under a given stress scenario, given genuinely available counterbalancing capacity. It is more useful to a board than any ratio because it translates the analysis into a single figure with an obvious interpretation — and because a minimum survival horizon makes a better policy limit than a ratio floor.
