The March 2023 failures of Silicon Valley Bank and Credit Suisse were a reminder that liquidity risk operates on a timescale of hours, not quarters. Capital ratios did not save either institution once the outflow began. Liquidity risk is the risk category that most punishes complacency and most rewards internal frameworks that go beyond the regulatory minimum.
This article sets out how the Liquidity Coverage Ratio (LCR), the Net Stable Funding Ratio (NSFR) and the Internal Liquidity Adequacy Assessment Process (ILAAP) fit together, and what a robust internal liquidity model looks like.
The regulatory anchors
Basel III introduced two headline liquidity ratios:
- LCR requires a stock of high-quality liquid assets (HQLA) at least equal to net stressed outflows over a 30-day stress horizon. The stress calibration includes 3–10% retail deposit runoff, 25–100% wholesale deposit runoff by counterparty type, and prescribed drawdown rates on committed facilities.
- NSFR requires available stable funding (ASF) to be at least equal to required stable funding (RSF) over a one-year horizon. It targets funding structure rather than short-term stress.
Both ratios are minimum standards, not risk-management tools. The EBA and national regulators explicitly require an ILAAP — the liquidity analogue of ICAAP — that assesses the adequacy of the internal framework, going beyond the standardised ratios. The ECB's SREP guides and the EBA's liquidity SREP guidelines set out the components: governance, funding plan, intraday liquidity, stress testing, contingency planning and reverse stress.
The internal framework: five layers
An internal liquidity framework has to answer five questions, in order.
- Structural: is our funding profile stable enough to withstand normal market volatility?
- Behavioural: how do our cash flows actually behave, as opposed to how the contractual schedule says they behave?
- Stress: what happens under idiosyncratic and market-wide stress, and how long does the institution survive without corrective action?
- Intraday: can we meet payment obligations minute by minute during a bad day?
- Contingency: what actions can we take, in what order, and how much liquidity do they generate?
Behavioural cash-flow modelling
Contractual cash flows tell a story that does not match reality. Retail current-account deposits are contractually overnight but behaviourally have half-lives measured in years; committed facilities are contractually available but behaviourally drawn far less often; mortgages have contractual amortisation but behavioural prepayment (see IRRBB Behavioural Models for the interest-rate side of the same phenomenon).
Deposit-behaviour models typically split balances into:
- Core deposits: stable across rate and stress conditions, priced insensitively.
- Non-core stable: sensitive to rates and competitor pricing, but not stress-flighty.
- Non-core volatile: rate-sensitive and stress-flighty.
Empirical calibration uses account-level data: age of relationship, product type, primary-transactional flag, balance concentration, digital-channel usage. The behavioural runoff rate under stress is not the historical runoff rate under stress — the SVB episode showed that digital-native concentration can multiply stressed runoff rates far above Basel LCR calibrations.
Stress testing: three families of scenarios
ILAAP stress testing typically runs three scenario families in parallel.
Idiosyncratic stress assumes a name-specific event: rating downgrade, negative earnings surprise, adverse news. Retail deposit runoff accelerates; wholesale funding tenors shorten or disappear; secured funding haircuts widen. HQLA remain liquid because the market functions.
Market-wide stress assumes a systemic event: a funding-market seizure similar to 2008 or 2020. HQLA remain liquid but with wider haircuts and slower monetisation; central-bank facilities become the primary source of contingent funding.
Combined stress overlays the two. The survival horizon under combined stress is the key internal-model output. Regulators expect institutions to survive at least the LCR horizon; strong internal frameworks target multi-month survival under combined stress, with clear management actions at defined coverage triggers.
Reverse stress testing asks the opposite question: what combination of outflows and asset frictions makes the institution non-viable? The answer names the concentrations — a wholesale funding provider, a single large depositor segment, a specific HQLA class with a hidden correlation to the bank's own credit — that the standard scenarios did not stress hard enough.
Intraday liquidity: the underappreciated layer
Intraday liquidity is often treated as an operational rather than a risk-management topic — a mistake. The BCBS 248 monitoring tools formalised the intraday view, but real intraday management requires:
- Real-time visibility of nostro balances.
- Modelling of expected payment flows by hour, including known large-value payments.
- Identification of the intraday peak — the largest cumulative net-payment obligation during the day — and a buffer sized to cover it under stress.
- Contingent access to central-bank intraday credit and its collateral cost.
Institutions that manage intraday well tend not to have surprises at end-of-day; institutions that do not manage intraday find out during a crisis that their nominal daily liquidity is unavailable when they need it.
Survival horizon and the contingency plan
The internal liquidity model culminates in a survival-horizon number: the number of days the institution can meet all outflows under a defined stress before executing management actions. The contingency-funding plan (CFP) then defines those actions with three attributes:
- Availability: cash raised in day 1 versus day 30 versus day 90.
- Cost: haircuts, spread widening, reputational cost.
- Sequencing: which actions do not preclude others, and which actions signal weakness to the market.
An unsequenced list of actions is not a plan. The CFP should be tested annually with a dry run — a table-top or wargame — and the results should feed back into the ILAAP.
Governance implications
The board should own the risk-appetite statement for liquidity: LCR and NSFR floors well above regulatory minima, internal survival-horizon targets, intraday buffer requirements, and defined trigger levels for the CFP. The chief risk officer, the treasurer and internal audit should each have documented roles in the ILAAP cycle. Regulators pay particular attention to the governance chain during on-site inspections.
Validation and independent challenge
Validators should focus on:
- Behavioural assumptions: are runoff rates and drawdown rates back-tested against realised events? SVB and Credit Suisse are the reference cases for the current cycle.
- HQLA composition: is the HQLA book concentrated in a single asset class, a single counterparty, or a single collateral chain?
- Scenario severity: are stress calibrations updated when new episodes reveal thicker tails, or are they anchored to a pre-2023 world?
- Reverse stress usefulness: does reverse stress name specific concentrations and produce action, or is it a paper exercise?
See The Model Validation Lifecycle for the broader validation framework.
Limitations
Liquidity models are stress-scenario models. The scenarios that matter most are the ones no committee has yet imagined — social-media-driven deposit runs, cyber events that immobilise payment infrastructure, sovereign events that reprice HQLA overnight. A modest amount of humility in the calibration, and a well-rehearsed CFP, matter more than a decimal-place refinement of the runoff rates.
Conclusion
The banks that survive liquidity crises share three habits: they measure behavioural cash flows honestly, they stress harder than the regulatory calibrations, and they treat the contingency plan as a live document. Everything else — the LCR ratio, the NSFR ratio, the ILAAP write-up — follows from those three.
Related reading
References and further reading
- Basel Committee on Banking Supervision, Basel III: The Liquidity Coverage Ratio and liquidity risk monitoring tools, BCBS 238.
- Basel Committee on Banking Supervision, Basel III: The Net Stable Funding Ratio, BCBS 295.
- Basel Committee on Banking Supervision, Monitoring tools for intraday liquidity management, BCBS 248.
- European Banking Authority, Guidelines on ICAAP and ILAAP information collected for SREP purposes (EBA/GL/2016/10).
About the author
Part of an ongoing series on risk, capital and modelling — more about the author.