Guide

Liquidity coverage ratio (LCR) explained

Harbor Credit Union’s treasury desk watched SOFR spike 22 basis points above IORB on the last business day of March 2024. Aggregate bank reserves still looked ample on H.4.1. The desk’s old model treated all cash and Treasuries as interchangeable liquidity. Replay of the week showed the miss: Harbor held $840 million in Level 2B corporate bonds counted toward regulatory liquidity, but the LCR runoff model assumed 50% of wholesale funding would not roll — a binding constraint the desk had not modeled. When engineers rebuilt the funding forecast around LCR net cash outflows and HQLA haircuts instead of raw balance-sheet cash, wholesale borrowing forecast error fell from 18 basis points to 5.

The liquidity coverage ratio (LCR) is a Basel III rule requiring large banks to hold enough high-quality liquid assets (HQLA) to survive a 30-day acute stress scenario without relying on central bank support. It answers a narrow question: if depositors flee and markets freeze for one month, can you pay your bills? This guide covers HQLA Level 1/2A/2B tiers, net cash outflow assumptions, the 100% minimum and supervisory buffers, how LCR differs from Fed reserve balances and corporate liquidity ratios, the Harbor Credit Union refactor, a technique decision table versus NSFR and CCAR stress tests, pitfalls, and a production checklist.

What the LCR measures

The LCR is defined as:

LCR = Stock of HQLA ÷ Total net cash outflows over 30 days ≥ 100%

HQLA is the numerator: unencumbered, liquid assets a bank can sell or repo quickly in stress. Net cash outflows is the denominator: expected cash leaving the bank minus contractual inflows, under standardized runoff rates prescribed by regulators. The ratio is a stock-to-flow test over a fixed 30-day horizon — not a profitability metric, not a capital ratio, and not the same as having large reserve balances at the Fed.

U.S. implementation (via the Federal Reserve’s enhanced prudential standards) applies to large bank holding companies and certain subsidiaries. Smaller institutions may face simplified liquidity rules, but the LCR framework shapes money-market behavior for the whole system because large banks are the marginal buyers of T-bills, agency debt, and repo collateral.

What LCR is not

Not a reserve requirement — reserves are Fed account balances; HQLA is a broader asset bucket with haircuts.
Not the NSFR — the Net Stable Funding Ratio covers one-year structural funding mismatch; LCR is a 30-day survival test.
Not corporate working capital — current and quick ratios use accounting book values without stress runoff assumptions.
Not a market liquidity guarantee — HQLA assumes markets exist; in extreme crises, even Treasuries can gap.

High-quality liquid assets (HQLA)

HQLA are split into three tiers with caps and haircuts. Only unencumbered assets count — securities pledged as collateral for repos, derivatives margin, or FHLB advances are excluded unless explicitly releasable within the stress window.

Level 1 (no haircut, up to 100% of HQLA stock)

Central bank reserves — Fed master account balances; the most liquid USD asset for U.S. banks.
Marketable securities unconditionally guaranteed by sovereigns — U.S. Treasuries, certain agency MBS pass-throughs per rule text.
Foreign sovereign debt in the bank’s operating currency, subject to 0% or 20% haircuts by rating and jurisdiction.

Level 2A (15% haircut, max 40% of HQLA after caps)

High-grade sovereign and agency securities not in Level 1.
Certain covered bonds and agency debentures meeting rating thresholds.

Level 2B (25–50% haircuts, max 15% of HQLA after caps)

Investment-grade corporate bonds and covered bonds below 2A thresholds.
Exchange-traded equities meeting size and listing criteria.
Lower-rated but still eligible RMBS and ABS tranches.

The tier structure explains why large banks hoard T-bills and reserves while limiting Level 2B corporates: a $100 million corporate bond might contribute only $50 million to HQLA after a 50% haircut and the 15% bucket cap. That is why Harbor’s March miss hurt — the desk saw $840 million face value but only ~$290 million counted toward the LCR numerator after haircuts and caps.

HQLA must be in the bank’s control and convertible to cash within the stress period. Operational readiness matters: settlement accounts, tri-party repo arrangements, and encumbrance tracking systems are part of LCR compliance, not just portfolio allocation.

Net cash outflows: the 30-day stress denominator

The denominator sums cash outflows minus cash inflows, each multiplied by regulatory runoff or inflow rates. The scenario assumes no access to wholesale markets for unsecured funding and limited rollover of secured funding — a coordinated run, not a normal month.

Major outflow categories

Retail deposits — stable vs less-stable buckets; runoff rates typically 3–10% for insured retail, higher for uninsured portions above thresholds.
Wholesale funding — commercial paper, unsecured interbank, and non-operational deposits from financial counterparties; runoff often 25–100% depending on counterparty type.
Secured funding — repos and securities lending; runoff depends on collateral type and counterparty (e.g., 0% on Treasuries vs higher on corporates).
Derivatives and commitments — net collateral outflows, liquidity puts, and undrawn credit lines to financials at prescribed draw rates.
Debt maturities — contractual principal and coupon payments due within 30 days.

Inflow caps

Contractual inflows (loan repayments, interest) are recognized only up to 75% of gross outflows in the standard rule — you cannot assume every borrower pays you while every depositor runs. That cap is why asset-heavy banks with large loan books do not get full credit for scheduled principal receipts in the LCR denominator.

The result is a single total net cash outflow number, usually billions for a G-SIB. A bank with $50 billion net outflows needs $50 billion of risk-weighted HQLA to hit 100% LCR before any internal buffer.

Minimum requirements and buffers

The Basel minimum is 100% LCR. U.S. globally systemically important banks (G-SIBs) typically maintain internal targets of 110–120% to absorb modeling noise and supervisory expectations. The LCR is reported publicly in quarterly regulatory filings (FR Y-9C, call reports for applicable entities) and monitored in CCAR/DFAST liquidity-risk scenarios for the largest firms.

LCR interacts with other plumbing:

QT and reserves — as the Fed shrinks its balance sheet, large banks may hold fewer reserves but must still meet LCR; T-bill demand rises.
Money market funds — MMF inflows during stress can drain bank deposits while adding MMF claims on bank paper — a feedback loop LCR models try to capture.
ON RRP — reserves parked at the Fed via reverse repo count as Level 1 HQLA; the ON RRP facility competes with bank deposit funding.

Harbor Credit Union refactor

Harbor is not a G-SIB, but its ALM desk models large-bank LCR behavior because wholesale funding costs correlate with system-wide HQLA demand. The refactor had three layers:

HQLA shadow book — tag every liquid asset with Level 1/2A/2B, haircut, encumbrance flag, and contribution after the 40%/15% caps.
Outflow taxonomy — map deposit and funding products to regulatory runoff buckets instead of treating all wholesale as 7-day paper.
Quarter-end calendar — overlay balance-sheet window dressing: banks shrink HQLA-eligible books into quarter-end, then reload T-bills in the first week of the new quarter, moving SOFR and repo rates.

Before the refactor, Harbor’s model used total cash + Treasuries as “liquidity” and 5% deposit runoff. After mapping to LCR logic, effective stress liquidity was 34% lower and wholesale runoff sensitivity was 3.2× higher on financial-counterparty funding. Forecast error on FHLB advance and repo spreads dropped from 18 bp to 5 bp on six quarter-ends.

What did not help: Using corporate quick ratio for bank counterparties — book liquidity overstated HQLA by 40%.

Technique decision table

Question	Use LCR framework	Use instead
Can a large bank survive 30 days of funding stress?	LCR — HQLA vs net outflows	Raw cash balances miss haircuts and runoff
Is long-term funding stable vs long-term assets?	NSFR (one-year horizon)	LCR ignores maturity beyond 30 days
Will CET1 absorb credit losses in recession?	CCAR/DFAST capital stress	LCR is liquidity, not solvency
Is the banking system reserve-ample?	Reserve aggregates + SOFR vs IORB	LCR is institution-level; reserves are system-level
Can an industrial company pay bills next quarter?	Corporate liquidity ratios + FCF	LCR runoff rates apply to banks, not corporates
Will MMF reform shift deposit beta?	LCR + MMF flow models	Fed funds target alone

Implementation patterns

Data and systems

Production LCR engines ingest core banking deposits by legal entity, securities inventory with encumbrance flags, derivatives margin schedules, and wholesale funding maturities. Daily vs monthly reporting granularity matters: trading desks move HQLA intraday; regulatory LCR is often a month-end snapshot with averages.

ALM integration

Treasury should tie LCR to the funds-transfer-pricing curve: HQLA carries an opportunity cost versus lending. When LCR binds, banks issue more commercial paper and CDs, bid aggressively for retail deposits, and buy T-bills — all observable in market spreads.

Investor and counterparty monitoring

Even if you are not LCR-regulated, watch large-bank LCR disclosures in 10-Q footnotes. Falling LCR toward 100% with rising Level 2B share is an early signal of wholesale funding pressure before TED-style stress indicators move.

Common pitfalls

Counting encumbered securities as HQLA — pledged collateral does not count unless releasable within the stress window.
Ignoring Level 2 caps — corporates may be 30% of the portfolio but only 15% of HQLA after haircuts.
Using book cash without runoff — uninsured wholesale deposits can run at 100% in the rule; book cash overstates survival liquidity.
Assuming loan inflows offset outflows — the 75% inflow cap limits credit to scheduled repayments.
Confusing LCR with reserves — ample system reserves do not guarantee each bank’s LCR is comfortable.
Missing quarter-end dynamics — banks optimize reporting dates; mid-quarter market liquidity can differ sharply.
Static runoff rates in live stress — regulatory rates are floors; actual runs can exceed modeled outflows.

Production checklist

Inventory all assets with Level 1/2A/2B tags, haircuts, and encumbrance status.
Apply 40% Level 2A and 15% Level 2B caps to compute eligible HQLA stock.
Map deposits and funding to regulatory runoff categories (retail stable/unstable, financial wholesale, secured).
Calculate gross outflows, capped inflows (75% limit), and total net cash outflows.
Compute LCR = HQLA / net outflows; set internal target above 100% if applicable.
Stress-test quarter-end vs mid-quarter snapshots for window-dressing effects.
Link LCR binding constraints to T-bill demand, repo rates, and CD pricing forecasts.
Reconcile LCR liquidity view with CCAR liquidity narratives for large counterparties.
Monitor public LCR disclosures of G-SIBs you fund or trade with.
Document assumptions; refresh when products, counterparties, or rules change.

Key takeaways

LCR asks whether HQLA covers 30 days of stressed net cash outflows at a 100% minimum.
HQLA tiers and haircuts mean not all cash and bonds count equally toward survival liquidity.
Wholesale funding runoff assumptions often bind before retail deposit runs do.
Harbor Credit Union cut wholesale forecast error from 18 bp to 5 bp by modeling LCR instead of raw cash.
LCR complements but does not replace reserve monitoring, NSFR, or CCAR capital stress.