Reader question: How can a bank turn thousands of assets, deposits, loans, derivatives and funding commitments into one number intended to answer a precise question: could it meet stressed cash needs for the next 30 calendar days?
The Liquidity Coverage Ratio (LCR) is best understood as a constrained accounting-and-stress algorithm. It does not ask whether a bank is profitable, whether every asset is safe, or whether a bank can survive every imaginable crisis. It asks whether the bank holds enough qualifying high-quality liquid assets (HQLA) to cover a prescribed measure of net cash outflows over a 30-calendar-day stress period.
At its core:
LCR = stock of HQLA / total net cash outflows over the next 30 calendar days.
Under the Basel standard, the ratio is intended to be at least 100% in normal times, while supervisors may allow or expect the buffer to be used during genuine stress. The difficult part is not the division. The difficult part is constructing the numerator and denominator correctly.
Page role and boundary
This article explains the mathematics, data logic and verification of the Basel LCR as an educational model. It is not financial advice, not a judgement on any bank, and not a substitute for the Basel Framework or a national supervisor’s implementation rules. Local rules can be more stringent or differ in scope and detail.
The page also has a deliberately different role from Bukit Timah Tutor’s article on optimising HQLA liquidity buffers. That article asks how a bank might choose among eligible liquid assets. This page asks how the regulatory ratio itself is assembled, constrained, checked and falsified.
The LCR is a pipeline, not one formula
A useful computational decomposition is:
- classify assets and liabilities;
- apply eligibility rules and haircuts to liquid assets;
- apply runoff or drawdown factors to expected cash outflows;
- apply recognition factors to expected cash inflows;
- cap recognised inflows;
- apply HQLA composition caps;
- aggregate by the required 30-day horizon;
- divide adjusted HQLA by adjusted net cash outflows;
- run reconciliation, sensitivity and exception checks.
Each step can fail independently. A bank can therefore calculate the final division perfectly while still producing a wrong LCR because an account was placed in the wrong category, a haircut was omitted, an inflow was over-recognised, an encumbered security was treated as freely available, or a maturity date was mapped to the wrong time bucket.
Step 1: construct the HQLA numerator
HQLA are not simply “assets that look safe.” Basel requires qualifying assets to meet eligibility and operational conditions intended to make them usable as liquidity in stress. The framework divides eligible assets into Level 1 and Level 2, with Level 2 split into Level 2A and Level 2B.
For a simple educational representation, suppose the bank holds eligible market values:
- Level 1: 80 units;
- Level 2A: 20 units;
- Level 2B: 10 units.
Level 1 assets are generally included without a Basel LCR haircut, although supervisors may impose additional treatment. Level 2A assets are subject to a 15% haircut. Level 2B assets have higher prescribed haircuts depending on the asset class. If, purely for illustration, the Level 2B assets in this example receive a 50% haircut, the preliminary adjusted amounts are:
Level 1 = 80
Level 2A = 20 × 0.85 = 17
Level 2B = 10 × 0.50 = 5
The naive sum is 102. But the algorithm is not finished. Basel also limits Level 2 assets as a share of the HQLA stock, with a 40% cap on total Level 2 and a 15% cap on Level 2B after the prescribed calculations. Short-term securities-financing and collateral-swap transactions can also affect the cap calculation. The numerator is therefore a haircut-and-cap problem, not a simple market-value total.
Operational availability matters as much as asset labels
An asset can satisfy a market-quality test yet still fail the operational test. Basel requires HQLA to be unencumbered and controlled in a way that allows the bank to monetise them when needed. Legal, regulatory, tax, accounting or transfer restrictions can prevent assets in one entity from being freely available to another.
This creates a useful diagnostic distinction:
economic liquidity ≠ regulatory eligibility ≠ operational availability.
A liquid government bond pledged elsewhere is not the same thing as an unencumbered government bond under the LCR. A calculation that only reads security type and ignores encumbrance can overstate the buffer.
Step 2: turn contractual obligations into stressed cash outflows
The denominator begins with expected cash outflows within 30 calendar days, but those outflows are not always taken at 100% of contractual balances. Basel applies category-specific runoff, drawdown or requirement factors intended to represent a prescribed stress scenario.
For a simplified balance B with runoff factor r:
stressed outflow = B × r.
For example, if a qualifying deposit category has a balance of 1,000 and an applicable runoff factor of 5%, its modelled outflow is 50. A different category with the same 1,000 balance but a 40% factor contributes 400. The labels and evidence behind classification therefore dominate the arithmetic.
Basel’s application guidance illustrates why classification matters: stable retail deposits, less stable retail deposits, operational wholesale deposits, unsecured corporate funding, secured funding and other legal-entity funding can carry very different prescribed outflow rates. The model is intentionally not saying that exactly that percentage will leave. It is specifying a standardised stress treatment.
A classification error can be larger than a forecasting error
Suppose 500 units of funding are accidentally classified into a 5% runoff bucket instead of a 40% bucket. The reported stressed outflow would be 25 rather than 200, a difference of 175. No sophisticated forecasting model can repair that if the source data are wrong.
This is a general lesson in applied mathematics: the type system is part of the model. A formula receives meaning from how its inputs are defined. Data lineage, customer classification, deposit-insurance attributes, product terms, maturity and collateral status are therefore computational inputs, not administrative decoration.
Step 3: recognise cash inflows — but do not let inflows erase the need for a buffer
Expected inflows over the same horizon reduce net cash outflows, subject to Basel recognition factors and conditions. The general structure is:
net cash outflows = total outflows − min(total recognised inflows, inflow cap).
The standard generally caps cash inflows at 75% of total expected cash outflows. This means the bank must maintain a minimum stock of HQLA rather than assuming that incoming cash will perfectly offset outgoing cash.
Suppose stressed outflows are 120 and otherwise eligible inflows are 110. Without a cap, net outflows would be only 10. With a 75% inflow cap, recognised inflows are at most:
0.75 × 120 = 90.
So net cash outflows become:
120 − 90 = 30.
This is not merely a technicality. It is a structural design choice: the ratio should not be reduced to near zero by optimistic dependence on counterparties paying exactly as expected during a stress episode.
A complete worked example
Consider a simplified bank after all asset haircuts and composition caps have been applied. Its eligible HQLA stock is 105. Its prescribed 30-day stressed cash flows are:
- retail and wholesale funding outflows: 70;
- secured funding and collateral-related outflows: 20;
- committed-facility and other contractual outflows: 30;
- total outflows: 120;
- otherwise eligible inflows: 80.
The 75% inflow cap is 90, so all 80 of inflows can be recognised in this example. Net cash outflows are:
120 − 80 = 40.
The LCR is therefore:
105 / 40 = 262.5%.
That percentage is an output of the prescribed framework. It does not mean the bank has “262.5% safety” in a universal sense. A different stress, different currency, different legal entity, different behavioural pattern or different asset monetisation environment can produce a very different liquidity need.
Why the 30-day horizon is not a forecast of day 30
The LCR horizon is a regulatory stress horizon, not a claim that all liquidity risk can be compressed into one monthly endpoint. Intraday settlement pressure can occur in minutes. Some funding vulnerabilities emerge over months. The LCR deliberately focuses on short-term resilience.
This is why it belongs beside, rather than replaces, bank liquidity stress testing, the Net Stable Funding Ratio and intraday-liquidity forecasting. Each answers a different time-scale question.
Inputs and outputs
A production LCR engine can require thousands or millions of records, but its logic can be summarised by a typed input set:
- asset inputs: security identifier, issuer, market value, currency, rating or risk-weight treatment where relevant, liquidity characteristics, encumbrance, legal entity, maturity and transaction status;
- liability inputs: counterparty type, product type, insurance status, relationship or operational attributes where permitted, maturity and contractual terms;
- off-balance-sheet inputs: committed facilities, contingent obligations and drawdown categories;
- derivatives and collateral inputs: contractual cash flows, collateral requirements, downgrade triggers and historical collateral-flow evidence where required;
- rule inputs: Basel or local haircuts, caps, runoff factors, inflow factors and eligibility rules;
- output: adjusted HQLA, total stressed outflows, recognised inflows, net cash outflows, LCR and a set of explainable exceptions.
Failure modes
Wrong customer type: a corporate balance is mapped as retail or small business, changing the runoff treatment.
Encumbrance blindness: a security is counted as HQLA even though it is pledged and unavailable.
Double counting: the same security or cash flow appears in more than one source system.
Maturity drift: a contractual payment close to the 30-day boundary is placed in the wrong bucket because of business-day or timestamp handling.
Cap failure: the engine applies haircuts but forgets the Level 2 or Level 2B composition cap, or recognises inflows beyond the permitted cap.
Legal-entity leakage: liquidity located in one entity is assumed freely transferable to another despite restrictions.
Static rules: the calculation uses an obsolete parameter table after regulatory treatment changes.
Precision illusion: a ratio reported to two decimal places is treated as more certain than the classifications and assumptions that produced it.
Diagnostics and verification
- Source-to-report reconciliation: every material balance should trace back to authoritative books and records.
- Rule replay: freeze a data snapshot and rerun the calculation with the exact version of the rule table used for reporting.
- Boundary tests: place synthetic items immediately before and after 30 days, exactly at category thresholds, and around HQLA caps.
- Cap invariants: after the full calculation, confirm Level 2 and Level 2B do not exceed their permitted shares and recognised inflows obey the cap.
- Negative and impossible-value tests: detect duplicated balances, negative runoff factors, missing currencies, impossible maturity dates and HQLA values exceeding eligible holdings.
- Shock sensitivity: alter one runoff factor or a large deposit classification and check that the output moves in the expected direction.
- Independent recomputation: reproduce a sample of categories outside the production engine using a transparent spreadsheet or script.
- Disclosure reconciliation: where public disclosure applies, reconcile internal aggregates to the reported template and investigate unexplained differences.
Counterexamples and limits
A high LCR does not prove a bank cannot face a run. Depositors can leave faster than the prescribed scenario; markets can become less liquid; operational failures can block asset monetisation; intraday payment needs can arrive before asset sales settle; currencies may be mismatched; legal entities may be ring-fenced; and a bank can have strong liquidity while suffering a solvency problem.
The reverse is also important. A temporary LCR decline during genuine stress does not automatically prove a bank has failed its liquidity function. Basel explicitly designed the stock of HQLA as a buffer that can be drawn down in stress, subject to supervisory response. The ratio must therefore be interpreted with the state of the system, not as a number detached from context.
What would falsify confidence in the calculation?
Confidence should fall if independent recomputation cannot reproduce the numerator or denominator; if material balances have no source lineage; if known encumbered assets remain in HQLA; if a category change fails to alter runoff as expected; if composition or inflow caps can be breached without an exception; if rule versions are undocumented; or if disclosed values cannot be reconciled to the underlying calculation.
A useful control is falsifiable: it should be possible to construct a test case that the engine must reject.
Alternatives answer different questions
The LCR is not a universal liquidity model. A bank can also use contractual maturity ladders, behavioural deposit models, survival-horizon stress tests, intraday-liquidity simulations and the NSFR. These are not competing formulas where one must win. They illuminate different weak links.
For example, balance-sheet optimisation asks how multiple capital, liquidity, funding and profitability constraints interact. The LCR provides one of those constraints; it does not solve the whole optimisation problem.
Current status and update triggers
As of August 2026, the consolidated Basel Framework continues to express the LCR as HQLA divided by 30-day net cash outflows, with the core HQLA categories, haircuts, composition caps and cash-inflow cap described above. National implementation can differ in scope and calibration, and supervisors may impose stricter requirements.
Re-check this article when the Basel LCR chapters change, when national rules alter eligibility or runoff treatment, when disclosure templates are revised, or when a bank changes source-system mappings that determine customer, collateral, maturity or encumbrance classifications.
Primary and high-quality references
- Bank for International Settlements, Liquidity Coverage Ratio — Executive Summary.
- Basel Committee on Banking Supervision, LCR20 — Introduction and objective.
- Basel Committee on Banking Supervision, LCR30 — High-quality liquid assets.
- Basel Committee on Banking Supervision, LCR40 — Cash inflows and outflows.
- Basel Committee on Banking Supervision, DIS85 — Liquidity disclosure requirements.
- Basel Committee on Banking Supervision, Consolidated Basel Framework.
Educational boundary: This article teaches a regulatory-liquidity calculation as mathematics and computation. It does not assess any bank or recommend any financial action.