The Liquidity Coverage Ratio (LCR)

The Liquidity Coverage Ratio (LCR) Explained

The Liquidity Coverage Ratio is the regulatory floor for 30-day liquidity stress. It answers a simple question: "If everything goes wrong in the next month, can you pay your obligations?" The answer must be yes, with a safety margin.

The Basic Formula

``
LCR = High-Quality Liquid Assets / Net Cash Outflows Over 30 Days
``

Regulators require LCR ≥ 100%. This means your HQLA must equal or exceed your net cash outflows under stress. At 100%, you're at the minimum. Most large banks operate at 120%–140% to maintain a comfortable buffer and signal strength to market participants.

What Is HQLA?

HQLA is defined very precisely by regulators (Fed, OCC, FDIC). Not all assets are equally liquid, so regulators assign haircuts:

Level 1 Assets (0% haircut):

• Cash and central bank reserves


• US Treasury securities


• Securities issued or guaranteed by US government agencies (Fannie Mae, Freddie Mac)

Level 2A Assets (15% haircut):

• Debt issued by US government-sponsored enterprises (GSEs)


• Certain highly-rated corporate bonds


• Certain municipal securities

Level 2B Assets (25% haircut):

• Residential mortgage-backed securities (agency MBS)


• Certain lower-rated corporate securities


• Certain equity securities (with restrictions)

Level 1 assets dominate for most banks. A $100 billion Treasury position counts as $100 billion HQLA. The same $100 billion in corporate bonds (Level 2A) counts as $85 billion HQLA.

The Stress Scenario

LCR assumes a "severe combined stress scenario" over 30 days. Key assumptions:

• Insured deposits: 5% outflow (retail deposits are stable)
• Uninsured deposits: 25% outflow (wholesale-like behavior when stressed)
• Brokered deposits: Often 25%–40% (very unstable)
• Operational deposits: 25%–40% (corporate customer deposits used for operations; partially used for payments)
• Wholesale funding: Varies by tenor and counterparty. Short-term wholesale (overnight repo, commercial paper) can see 75%+ outflow
• Asset outflows: Contingent liabilities trigger (letters of credit, revolving credit commitments)

For SVB, the stress scenario proved optimistic. In reality, uninsured depositors withdrew 100%+ of their balances in 2 days.

Practical Calculation: A $50 Billion Bank Example

Assume a regional bank with:

• Deposits: $40 billion (30% uninsured, 70% insured)


• Wholesale funding: $5 billion (1-year term debt, 2-year term debt)


• HQLA: $12 billion (Treasury bonds, agency MBS)

Under LCR stress:

• Insured deposits ($28 billion): 5% outflow = $1.4 billion


• Uninsured deposits ($12 billion): 25% outflow = $3 billion


• Wholesale funding maturity or early termination: $5 billion assumed to refinance (worst case: all 30-day term debt rolls off)


• Contingent liabilities (committed lines): $2 billion

Total outflows: $1.4 + $3 + $5 + $2 = $11.4 billion

Net inflows: Loan repayments, deposit interest, etc. Assume $1 billion net inflows.

Net cash outflows: $11.4 billion - $1 billion = $10.4 billion

LCR: $12 billion / $10.4 billion = 115%

This bank passes the test but doesn't have a huge buffer. A slightly worse scenario (more wholesale funding, more contingent liability draws) could breach 100%.

Why It Matters

LCR is a regulatory floor, but it's a binding floor. Banks that drop below 100% face mandatory action: capital charges, increased supervisory attention, and pressure to raise funding or reduce illiquid assets. In 2023, several regional banks operated close to 100%, which telegraphed liquidity stress to market participants and accelerated deposit runs.

A strong LCR (120%+) signals: "We can absorb a 30-day stress without asset sales or new funding." A weak LCR (100%–110%) signals: "We're relying on the stress scenario being accurate. If it's worse, we're in trouble."

Integration with Balance Sheet Strategy

LCR directly influences ALM strategy:
1. HQLA Portfolio: You must maintain a substantial buffer of Treasuries and agency MBS. This creates a drag on earnings (Treasuries yield less than corporate bonds), but it's non-negotiable.
2. Deposit Mix: Uninsured deposits are cheaper (no FDIC insurance costs) but they increase your LCR requirement. Each $1 of uninsured deposits increases net outflows by $0.25.
3. Wholesale Funding Tenor: Shorter-tenor funding (30-day commercial paper) counts heavily against you. 2-year funding is stickier and creates more runway.
4. Asset Liquidity: Illiquid assets (commercial real estate loans, illiquid MBS) don't help LCR, so you can't "sell" liquidity from the asset side.

For most banks, LCR is not the binding constraint anymore—NSFR is. But LCR is still critical in acute crises.

The Liquidity Coverage Ratio: Mechanics, Calculation, and Hidden Gaps

The Liquidity Coverage Ratio is the most widely cited regulatory liquidity metric, but most practitioners misunderstand what it actually measures—and critically, what it misses. The LCR is fundamentally a 30-day acute stress test: it asks, "Can the bank survive 30 days of severe funding stress using only its liquid assets and projected inflows?" It's a point-in-time metric, highly sensitive to balance sheet composition, and prone to both false confidence and unnecessary conservatism depending on how assets and liabilities are classified.

The Numerator: High-Quality Liquid Assets and Hidden Execution Costs

The asset side of the LCR looks simple on paper: count Treasuries, agencies, investment-grade corporates, and other securities, apply haircuts, and sum them. The regulatory haircuts are low (0% for Treasuries, 15% for corporate bonds). But the real world is messier.

Level 1 Assets: Treasuries and Central Bank Reserves

These carry a 0% regulatory haircut, which implies they can be sold or monetized instantly with zero execution cost. In reality, there are market execution costs that aren't captured. When the Fed tightens and Treasury yields rise, bid-ask spreads on Treasuries widen. A bank attempting to sell $5 billion of 10-year Treasuries into a stressed market faces execution costs of 5–10 basis points, or $2.5–5 million per maturity. Across a $15 billion portfolio of Treasuries, execution costs could total $7.5–15 million.

During the acute stress of March 2023, Treasury bid-ask spreads on 2-year and 10-year notes widened from 1–2 basis points to 5‱0 basis points. Over a 30-day period, if a bank needed to liquidate a significant portion of its Treasury position, the execution drag would be material. LCR doesn't account for this because the regulatory assumption is that Treasuries are "cash-equivalent," but operationally, fire-selling them costs money.

Level 2B Assets: Agency Residential and Commercial Mortgage-Backed Securities

These securities are classified as Level 2B HQLA and carry a 25% regulatory haircut. The reasoning: they're backed by government-sponsored enterprises and therefore have low default risk, but they're less liquid than Treasuries because the secondary market is smaller. The 25% haircut seems to account for this.

However, in true stress scenarios, this haircut understates the execution challenge. In March 2023, when credit spreads widened 150–200 basis points system-wide, bid-ask spreads on agency MBS widened dramatically—from 1–3 basis points to 10–20 basis points or wider. This means selling $2 billion of MBS could cost $20–40 million in execution slippage. Furthermore, MBS face prepayment risk: if rates fall, homeowners refinance and the bank gets cash back earlier than expected (losing the higher yield it was counting on). In a stress scenario where rates are volatile, MBS prepayment becomes unpredictable.

The LCR regulatory framework also caps Level 2B assets at 50% of total HQLA. This means a bank cannot fund its entire liquidity buffer from MBS, even if it wanted to. It must maintain at least 50% of HQLA in Level 1 assets or lower-haircut Level 2A assets (corporate bonds, municipal bonds). This is a prudent design feature that prevents banks from gaming LCR by loading up on illiquid securities.

Level 2A Assets: Corporate Bonds and Investment-Grade Securities

These carry a 15% regulatory haircut and include investment-grade corporate debt, agency debt, and certain municipal securities. The execution mechanics are better than MBS (more dispersed buyer base) but still real. In a stress scenario with widening credit spreads, execution costs on corporate bonds can exceed the 15% haircut, especially for non-financial corporations where buyer interest evaporates quickly.

The Haircut Assumption Tightness

A critical oversight in LCR calculation is that regulatory haircuts are fixed regardless of market conditions. In calm markets, a 0% haircut on Treasuries is conservative because execution costs are near zero. In stress, when markets are volatile and volumes are high, the 0% haircut becomes unrealistic. This is a structural gap in the framework: it doesn't adjust for realized volatility or market conditions.

Sophisticated ALM teams often apply their own haircuts that are tighter than regulatory haircuts. For example, they might assume a 2% execution haircut on Treasuries even though the regulatory haircut is 0%, or they might assume a 40% haircut on corporate bonds instead of the regulatory 15%. This provides a more realistic liquidity buffer.

The Denominator: Funding Outflows and Their Hidden Assumptions

The liability side of the LCR is where most of the controversy sits, because it requires assumptions about which funding sources will leave under stress.

Insured Deposits: The 5% Assumption

Deposits up to the FDIC insurance limit ($250,000 for single accounts, $500,000 for joint accounts) are assumed to have only a 5% outflow rate under stress. The assumption is sound in principle: if your deposit is government-insured, why would you move it just because the bank is under stress? The bank's credit quality is no longer relevant to your principal recovery.

However, there are execution challenges practitioners often overlook:

1. Deposit Complexity: A customer might have multiple accounts across different deposit "categories" (individual, joint, business, trust), each insured separately up to $250k. If the bank structures deposits across categories, it increases perceived insurance coverage. But under stress, when customers panic, they may not understand the nuances of coverage and move their entire balance anyway.

2. Non-Traditional Insured Deposits: Some banks use products like "sweep accounts" that automatically move money between deposit and investment accounts at the end of each day. If the investment portion is in non-FDIC-insured securities and the customer panics, they may withdraw the entire balance despite the technical FDIC coverage on the deposit portion.

3. Network Effects: During systemic stress (when multiple banks are under pressure simultaneously), even insured depositors may move money simply to consolidate at a stronger bank. The 5% assumption presumes rational behavior; during crises, behavior is often driven by confidence and narrative, not pure economic calculation.

Uninsured Deposits: The 25% Assumption and Its Insufficiency

Deposits above the FDIC insurance limit are assumed to have a 25% outflow rate under stress. This is the most controversial assumption in the LCR framework, and 2023 proved why.

SVB, Signature Bank, and First Republic all had LCR ratios above 100% (often 120%+) weeks before they failed. Why? Because the 25% uninsured deposit outflow assumption was grievously understated for banks with concentrated, sophisticated depositor bases.

• SVB: Over 70% of deposits were uninsured, and those deposits came from venture capital-backed tech firms. When that sector faced a funding crisis in early 2023, uninsured deposits at SVB fell from $91.3 billion to approximately $15 billion in 48 hours. That's an 85% outflow rate, not 25%.

• Signature Bank: Similarly concentrated in digital asset and crypto firms, with uninsured deposits making up roughly 85% of deposits. The outflow rate exceeded 80% within a week.

• First Republic: A wealth management bank with large uninsured deposits from high-net-worth individuals. When contagion from the SVB/Signature failures spread and regulators scrutinized regional bank stability, uninsured deposits declined approximately 60% in 10 days.

What these cases revealed is that the LCR's 25% assumption works for diversified, idiosyncratic deposit stress (where 25% of random uninsured customers leave). It does not work for systematic, correlated deposit stress (where 80%+ of a concentrated customer segment leaves because they all face the same shock).

Post-March 2023, regulators are quietly revising uninsured deposit assumptions downward. Some bank supervisors have begun asking, "What's your concentration of uninsured deposits in your top 10 depositors? Top 100 depositors?" and stress-testing assuming much higher outflow rates (50%+) for concentrated segments.

Brokered Deposits: The 25%-40% Assumption

Brokered deposits are deposits intermediated by third-party brokers (InvestorPlace, StockNews, etc.). They're "shopped" to banks by brokers who select banks based on rate competitiveness. When stress hits, brokers immediately redirect flows to safer banks offering the same or higher rates. These deposits are among the most volatile funding sources a bank can have.

The LCR assumes 25%–40% outflow depending on tenor (shorter tenors have higher outflows). In practice, brokered deposits often see 100% non-refinancing in stress; brokers simply stop placing new money, and existing balances mature and walk away. Some banks had 10%–15% of their total funding as brokered deposits in 2020–2023, a material concentration.

The advantage of brokered deposits is they're tracked separately on the balance sheet, so banks and regulators can monitor concentration. The disadvantage is they're vulnerable to rate-shopping in a competitive funding market.

Operational Deposits: The Stability Assumption Gap

Operational deposits are balances held by business customers who use them for day-to-day operations: payroll processing, accounts payable, tax deposits, etc. LCR assumes these are somewhat stable (outflow rates of 10%–25%) because customers "need" the bank for payment processing.

However, this assumption breaks down in idiosyncratic stress. If a bank's largest depositor is a commercial real estate developer, and that sector faces a downturn, the customer may withdraw 100% of its operating balance despite relying on the bank for payment processing. Payment processing isn't hard to switch; it's a commodity service offered by every bank and fintech.

Alternatively, if a bank is a lender to a specific industry (e.g., aircraft financing, auto lending) and that industry faces stress, its business depositors are simultaneously under stress, and they all withdraw at once. This correlated outflow isn't captured in LCR's aggregate outflow assumptions.

A Realistic 30-Day Stress Scenario

Let's build a detailed LCR model for a realistic mid-sized bank to illustrate the mechanics and assumptions:

Balance Sheet: $80 Billion Bank

Assets:

• Treasury securities: $15 billion


• Agency MBS: $8 billion


• Agency debt: $3 billion


• Corporate bonds (investment grade): $4 billion


• Cash and Fed reserves: $2 billion


• Total HQLA: $32 billion

- After regulatory haircuts: Level 1 assets = $15B + $2B + part of agency debt = $17B (0% haircut)
- Level 2A assets = $4B corporate bonds at 15% haircut = $3.4B
- Level 2B assets = $8B MBS at 25% haircut + $3B agency debt at 15% haircut = $6B + $2.55B = $8.55B
- Total HQLA (post-haircut): ~$29 billion

• Residential mortgages: $35 billion


• Commercial real estate loans: $8 billion


• Consumer auto/personal loans: $3 billion


• Commercial and industrial loans: $2 billion


• Other assets (illiquid): $2 billion


• Total Assets: $80 billion

Liabilities and Equity:

• Insured deposits: $35 billion


• Uninsured deposits: $20 billion


• Brokered deposits: $5 billion


• 1-month to 3-month wholesale funding (commercial paper, short-term debt): $10 billion


• 3-month to 1-year wholesale funding: $10 billion


• Equity: $5 billion


• Total Liabilities and Equity: $80 billion

LCR Numerator: HQLA

As calculated above: $29 billion (post-haircut HQLA)

LCR Denominator: 30-Day Outflows

Deposit Outflows:

• Insured deposits: $35B × 5% = $1.75B


• Uninsured deposits: $20B × 25% = $5B


• Brokered deposits: $5B × 40% = $2B

Wholesale Funding Outflows:

• 1-month to 3-month wholesale: Assumes full maturity and non-refinancing in stress = $10B


• 3-month to 1-year wholesale: Partial rollover assumption = $10B × 10% = $1B (some expected to roll despite stress)

Contingent Outflows:

• Undrawn credit commitments: Bank has $5B of revolving credit lines, 60% drawn, 40% undrawn. Assume 75% of undrawn gets drawn = $5B × 40% × 75% = $1.5B


• Other contingent items: $0.5B

Total Outflows: $1.75B + $5B + $2B + $10B + $1B + $1.5B + $0.5B = $21.75B

LCR Inflows:

The LCR allows some inflows to offset outflows (subject to caps). Typical inflows:

• Loan repayments (principal): Expected ~$1.5B (not all loans pay off in any 30-day period, but some do)


• Interest/dividend inflows: ~$0.5B


• Inflows from other customer activities: Minimal in a stress scenario, assume $0

Total Inflows: $2B (capped at 75% of outflows under regulation, so $21.75B × 75% = $16.3B, but we calculate $2B, so use $2B)

Net Outflows: $21.75B - $2B = $19.75B

LCR Calculation: $29B HQLA / $19.75B Net Outflows = 147%

This bank passes LCR comfortably (>100%). However, note the sensitivity:

• If uninsured deposit outflows were 50% instead of 25%, outflows would be $26.75B, bringing LCR to 108% (barely passing)


• If brokered deposits had 60% outflow, total outflows would be $22.75B, bringing LCR to 127%

This illustrates why concentration matters: a small change in deposit assumptions materially shifts the LCR ratio.

The LCR Maturity Cliff Problem

One structural issue with LCR is that it's a 30-day metric. If a bank has $10 billion of wholesale funding maturing in 31 days, it doesn't count as a 30-day outflow. But on day 31, it's due, and the bank must refinance or liquidate assets. Many banks have "cliffs" in their wholesale funding maturity schedules where large amounts mature just beyond the 30-day window.

A best-practice monitoring approach includes:

• Actual LCR (30-day): Regulatory requirement


• Implied LCR (10-day, 14-day): Monitoring shorter stress windows


• Extended LCR (60-day, 90-day): Forward-looking assessment of liquidity needs

Real Bank Governance: LCR Monitoring in Practice

The most sophisticated ALM teams don't just calculate LCR monthly for regulatory filing. They monitor it continuously:

Daily Monitoring:

• Run intra-month LCR estimates using actual balance sheet data


• Alert if LCR is trending downward


• If the month is trending toward LCR < 120%, escalate to management

Weekly Reviews:

• Compare actual deposit flows to forecast


• Identify divergence (e.g., uninsured deposits declining faster than expected)


• Adjust wholesale funding issuance plans accordingly

Monthly (Regulatory Filing):

• Calculate audited LCR using month-end balance sheet


• Decompose by asset class and liability class to understand drivers


• Stress test: Model LCR under tighter assumptions (e.g., 50% uninsured outflow instead of 25%)

Quarterly (CCAR/Stress Test):

• Model LCR under Fed-published stress scenarios (severe recession, rating downgrade, etc.)


• Ensure LCR remains >100% under stress


• Identify which balance sheet actions improve LCR most efficiently (e.g., selling illiquid loans vs. reducing uninsured deposits)

What LCR Misses: The Critical Gaps Exposed by 2023

The 2023 crisis revealed several ways LCR is an imperfect liquidity measure:

1. Concentration Risk: LCR treats "25% of uninsured deposits" as an aggregate flow, regardless of whether deposits are diversified or concentrated in 50 large relationships. A bank with 20% uninsured deposits spread across 5,000 customers has lower idiosyncratic risk than one with 20% uninsured deposits from 30 large hedge funds. LCR rates them identically.

2. Correlated Shock Risk: LCR assumes deposits from different customer segments outflow independently. If 25% of uninsured deposits leave, that's 25% of the whole pool. But if a bank's top 100 depositors are all from a single sector (tech VC-backed firms, crypto firms, real estate developers), and that sector faces a shock, 80% of uninsured deposits leave in 48 hours. The correlation structure of the customer base is invisible in LCR.

3. Acute Confidence Collapses: LCR models 30-day outflows. But in acute confidence crises, 80% of uninsured deposits can leave in 2–3 days (as happened at SVB and Signature). After day 3, LCR might show ample liquidity, but the bank is already insolvent because it's run out of cash before day 3 ends.

4. Mark-to-Market Risk: LCR doesn't account for the possibility that as a bank comes under stress (news reports, regulatory scrutiny), the market value of its illiquid assets (loans) declines due to information asymmetry. Buyers assume distressed banks have worse assets than average; this widens bid-ask spreads on secondary markets and makes HQLA fire sales more expensive.

5. HQLA Concentration and Execution Risk: If 50% of a bank's HQLA is agency MBS from a single issuer, and that issuer faces stress (e.g., Fannie Mae/Freddie Mac in a severe recession), the bank can't rely on selling those securities quickly.

These gaps explain why regulators increasingly emphasize stress testing, internal liquidity scenario modeling, and deposit concentration analysis—frameworks that go beyond LCR to capture dynamics the ratio misses.

The 2023 Regulatory Tightening

Post-March 2023, the Fed and OCC are tightening LCR supervision:

• Requiring banks to model LCR under more severe deposit outflow assumptions (especially uninsured and brokered)


• Scrutinizing deposit concentration and building concentration limits into stress tests


• Extending the stress horizon beyond 30 days to 60–90 days


• Requiring banks to maintain higher HQLA buffers than the minimum 100% LCR (e.g., target 125%+ in internal governance)

Banks that fail to adapt to this tightening will face regulatory pressure, higher capital requirements, or dividend restrictions. Those that already built conservative liquidity buffers and monitored concentration closely will have smoother compliance.