The Hidden Term Structure of Your Liquidity Buffer

Why a flat contingent liquidity premium is a cross-subsidy hiding in plain sight — and how the buffer's carry cost rises with the gap's horizon

Most banks charge a flat contingent liquidity premium — the same handful of basis points on a three-month working line and a ten-year mortgage. The number is round, the governance is simple, and everyone moves on. But ask a sharper question: how long must the bank carry the buffer to cover each gap? A funding gap that opens next quarter demands buffer carry for three months. A gap that opens in year four demands carry for four years. The buffer cost has a term structure, and ignoring it does something quiet but corrosive: it systematically undercharges long-tenor products while overcharging short ones. The cross-subsidy is invisible in the blended P&L, but it is real, it accumulates, and it distorts every origination decision the bank makes.

This article traces that term structure from first principles. The numbers come from a single illustrative institution — call it Avelmont — with a diversified balance sheet and a buffer pool that must cover three distinct types of contingent drain. The goal is not a formula. It is a way of thinking that turns a flat charge into a term-structured one and lets each product pay for the buffer it actually consumes.

What is the buffer for?

Before pricing the buffer, it helps to ask what the buffer is actually defending against. The answer is not a single threat — it is three, and they differ in timing, magnitude, and the degree to which the bank can see them coming.

Funding runoff

Wholesale tranches that won't roll. Non-core deposits that walk. These are funding-side gaps — the money the bank was counting on simply doesn't show up. A certificate of deposit matures and the counterparty declines to renew. A large corporate depositor moves its operating account to another institution. The gap is the difference between what was expected and what arrived. In stress, these gaps cluster: the same macro event that causes one depositor to pull also causes three others to hesitate, and the wholesale desk finds fewer bids for its commercial paper. The runoff is not random. It is correlated, and the correlation tightens precisely when funding markets are thinnest.

Collateral calls

Derivatives margin calls when rates move against the bank. These are instant, lumpy, and — here is the dangerous part — correlated with the very market stress that makes other funding harder to secure. A 100-basis-point rate shock on a moderately sized swap book can produce a margin call of several hundred million within hours. The call is bilateral, non-negotiable, and must be met in cash or high-quality collateral. The bank cannot defer it, negotiate it down, or explain it away. It either has the buffer or it doesn't.

Committed-line draws

Corporate borrowers pulling their credit lines. The pattern is almost perverse: firms draw on their committed facilities precisely in stress, precisely when the bank's own liquidity is tightest. A committed revolving facility that sits 20% drawn in normal times can jump to 80% drawn in a crisis — and the bank is contractually obligated to fund every dollar. The draw is the borrower's right. The funding is the bank's problem.

Why flat pricing fails

The cost of holding buffer assets — government bonds, central bank reserves, high-quality liquid assets — is the spread between what those assets earn and what the bank's own funding costs. That spread is the carry cost. It is almost always negative: the buffer earns less than it costs to fund. The difference is the price of keeping the insurance policy in force.

Now consider two gaps. The first opens at three months — a wholesale tranche maturing next quarter that may not roll. The second opens at four years — a committed facility that a borrower could draw in year four of a five-year credit line. How long must the bank carry the buffer for each?

For the three-month gap: carry cost = funding spread multiplied by 0.25 years. If the spread is 40 basis points, the carry is 10 basis points. Not trivial, but manageable.

For the four-year gap: carry cost = funding spread multiplied by 4 years. At the same 40-basis-point spread, the carry is 160 basis points. Sixteen times more.

Present-valuing the carry costs back to today creates a term structure: the contingent liquidity premium rises with the gap's horizon. Not linearly — because discounting compresses the far end — but relentlessly. A gap at five years carries more cost than a gap at two years, which carries more than a gap at six months. The CLP has a curve, and that curve belongs in the transfer price.

The Castagna & Fede framework

The intuition above can be made precise. The framework developed by Castagna and Fede gives the contingent liquidity premium for a loan as the present value of all the buffer carry costs attributable to the gaps that loan creates. No single formula is needed here — the logic is what matters.

Each gap is weighted by four quantities:

1. The probability that the drain occurs. A wholesale tranche failing to roll in normal conditions might have a 5% probability. In stress, 30%. A committed-line draw might have a 60% probability in a severe downturn. The probability scales the gap.
2. The funding spread the bank pays on the buffer. This is the carry cost per unit of time — the difference between what the buffer earns and what the bank pays to fund it. If the bank funds at SOFR + 40 bps and the buffer earns SOFR flat, the carry spread is 40 basis points per year.
3. The time until the gap bites. A gap at three months accumulates carry for three months. A gap at four years accumulates carry for four years. This is the term-structure mechanism: time multiplies the spread.
4. A discount factor. Future carry costs are present-valued back to today. This compresses the far end of the curve: a gap at year five costs more than a gap at year two, but not two and a half times more, because discounting shrinks the distant cash flows.

The result is a term-structured charge. Near gaps contribute little carry. Far gaps contribute more. The CLP rises with tenor — but not linearly, because discounting bends the curve back. The shape is concave: steep at the short end, flattening at the long end. And the level depends on the bank's own funding spread, which means two banks with different credit quality will produce different CLP curves even if their gap profiles are identical.

The CLP is not a fee. It is a present-valued cost of insurance. The insurance premium is the carry spread. The policy duration is the gap's horizon. And the claim probability is the likelihood that the drain actually materializes. Every product that creates a contingent gap must pay for its own insurance, at its own duration, at its own probability.

Why products differ — same bank, same day, same pool

If the CLP is term-structured and probability-weighted, then different products must carry different charges — even when they sit on the same balance sheet, funded from the same buffer pool, on the same afternoon. This is not a modeling curiosity. It is the central pricing implication.

Consider three products on Avelmont's book:

A fully drawn five-year bullet loan. The money is out. The gap is known. There is no contingent exposure — the bank has already funded the loan and the cash has left. The only buffer needed covers the funding-side risk that existing sources might not roll, but the loan itself creates almost no incremental contingent demand. CLP: roughly 4 basis points.

A revolving credit facility with a large undrawn commitment. The borrower can draw at any time, in any amount up to the limit, and will do so precisely when funding markets are stressed. The contingent exposure is massive: the gap could open at any horizon, the probability of draw rises in stress, and the magnitude is the full undrawn amount. CLP: 30 to 40 basis points.

A mortgage with no prepayment option. The money is out, like the bullet, but the bank faces funding-side runoff risk over a long horizon. The contingent exposure is moderate — it comes from the liability side, not the asset side. CLP: 5 to 8 basis points.

The range is nearly tenfold — from 4 to 35 basis points — and every number comes from the same buffer pool. The difference is entirely in the product's contingent footprint: how large a gap it can create, at what horizon, with what probability. A flat CLP would force the bullet to subsidize the revolver. The borrower drawing the bullet would overpay. The borrower holding the undrawn revolver would underpay. And the bank's relationship managers, armed with the wrong price signal, would chase the wrong business.

The LCR floor versus the economic buffer

Regulatory LCR requires banks to hold enough high-quality liquid assets to cover thirty-day net outflows under a prescribed stress scenario. This is a floor — a minimum buffer size. It is not a ceiling, and it is not an economic answer.

The economic buffer may be larger. It covers gaps beyond thirty days — the three-month wholesale maturity, the six-month deposit erosion, the eighteen-month committed-line draw in a protracted downturn. It accounts for correlation between drain types: the fact that funding runoff, collateral calls, and line draws tend to cluster in the same bad quarter. And it reflects the bank's own risk appetite, which may be more conservative than the regulatory minimum.

Some gaps breach the LCR first. A short, intense funding run — a wholesale counterparty refusing to roll overnight paper — hits the thirty-day window squarely. The LCR binds. The economic buffer, sized for longer horizons, may have room to spare.

Other gaps breach the economic buffer but not the LCR. A slow, persistent deposit erosion over quarters — retail customers gradually moving balances to a fintech competitor — never triggers a thirty-day spike. The LCR is comfortable. But the economic buffer, which must cover the cumulated drain over the full horizon, is strained.

The pricing implication is direct: the CLP must reflect whichever constraint binds — LCR or economic — and charge the loan for the one that actually forces the bank to hold more buffer. For a product whose contingent gap falls squarely in the thirty-day window, the LCR is likely the binding constraint, and the CLP reflects the regulatory buffer cost. For a product whose contingent gap stretches over years, the economic buffer binds, and the CLP must reflect the longer carry horizon. Charging the same rate for both is, once again, a cross-subsidy.

The oversizing slip

Banks tend to carry more buffer than strictly needed. The reasons are familiar: regulatory uncertainty about future calibrations, rating agency expectations that reward excess liquidity, board conservatism in the wake of the last crisis, and the simple institutional tendency to round up rather than down. The result is an actual buffer that exceeds the economic buffer, sometimes by a significant margin.

The excess buffer earns below the bank's funding cost. It has negative carry. Every basis point of excess is a drag on returns that must be absorbed somewhere. And here is where a quiet governance failure often occurs: if the CLP is calibrated to the economic buffer only, but the bank actually holds a larger buffer, the carry cost of the excess is not recovered from any product. It falls through the cracks. It becomes a hidden tax on equity returns — present in the P&L, visible to the CFO, but untraceable to any specific origination decision.

There are two honest responses to this problem. The first is to charge the economic buffer only — the buffer that the risk framework says is needed — and report the excess carry separately as a strategic cost of conservatism. This approach is clean: the CLP reflects the true cost of the liquidity risk, and the oversizing cost is attributed to a governance choice rather than to any particular loan. The second is to charge the actual buffer held, which means the CLP recovers the full carry cost, including the excess. This approach is more conservative and penalizes products for buffer that exists not because of their risk, but because of the board's risk appetite.

Most institutions choose the first route: charge the economic buffer, report the excess separately. The discipline is in the reporting. If the excess carry is visible, the board can decide whether the conservatism is worth the cost. If it is buried in the blended cost of funds, no one ever asks the question.

Letting the term structure speak

The liquidity buffer is not a fixed cost. It is a term-structured commitment whose price varies with the gap's horizon, the product's contingent footprint, and the constraint that binds. A flat CLP collapses all of this into a single number — and in doing so, it creates a cross-subsidy that favors long-dated, high-contingency products at the expense of short-dated, low-contingency ones. The subsidy is invisible in the aggregate P&L. It is corrosive in the origination pipeline.

The discipline is to map each loan's contingent gaps — funding runoff, collateral calls, committed-line draws — at their specific horizons. Price the carry on each gap at its own horizon. Weight by the probability that each drain materializes. Discount back to today. And let the term structure speak.

The short products will get cheaper. The working capital line, the three-month trade finance facility, the overnight repo — all of them will shed the excess CLP that a flat charge imposed on them. The long products will get more expensive. The five-year revolver, the ten-year project finance facility, the mortgage with decades of funding-side exposure — all of them will absorb the carry cost that the flat charge was quietly exempting them from.

And the pricing will finally reflect what the buffer actually costs to hold: not a flat tax, but a term-structured premium that rises with the horizon, varies with the product, and charges each loan for the insurance it actually consumes.

A flat CLP is a cross-subsidy hiding in plain sight. The term structure is the correction. The discipline is to let it through.

The worked example uses Avelmont, a fictional institution, with illustrative parameters. The CLP values shown are directional — actual charges depend on the bank's funding spread, buffer composition, gap profile, and the constraint (LCR or economic) that binds. The Castagna & Fede framework is one approach; other formulations exist and may produce different term structures for the same gap profile.