How Banks Model Wholesale-Funding Rollover Risk: Maturity Ladders, Refinancing Probabilities, Market Access, Concentration and Spread Stress

Quick answer: wholesale-funding rollover risk is the risk that a bank reaches a funding maturity and cannot replace the liability at an acceptable price, in the required amount, or at all. The mathematics starts with a maturity ladder, but a serious model does more than sum debt coming due. It assigns each funding source a renewal probability, stressed refinancing spread, investor concentration, collateral dependence, currency and legal-entity constraint, and alternative funding route. The model then asks whether the bank can survive a sequence in which some liabilities mature normally, some refinance only at punitive spreads, and some markets close altogether.

A liability can look stable every day until the one day it must be replaced.

Page role: this article owns the refinancing-and-maturity mechanism. How Banks Stress-Test Liquidity owns the wider cash-flow survival problem. How Banks Calculate and Optimise the NSFR owns the regulatory structural-funding ratio. This page asks a narrower operational question: what happens when specific wholesale liabilities actually reach their maturity dates?

1. Wholesale funding is useful precisely because it can be replaced quickly

Wholesale funding can include bank debt, certificates of deposit, commercial paper, repos, brokered or other large deposits, interbank borrowing and institutional funding. Different institutions use different mixes.

Compared with many retail deposits, wholesale sources can often be raised in large amounts and targeted to specific maturities. That flexibility is valuable. It is also the source of rollover risk: the investor providing today’s six-month funding is under no obligation to provide another six months when the instrument matures.

The Federal Reserve’s interagency funding-and-liquidity guidance explicitly emphasises diversified funding sources and tenor, ongoing market access, and limits on concentrations in secured and unsecured wholesale funding. See Interagency Policy Statement on Funding and Liquidity Risk Management.

2. Start with the maturity ladder

Suppose a bank has the following wholesale liabilities:

Funding source	Amount	Residual maturity
Commercial paper	S$500m	30 days
Institutional term deposit	S$700m	90 days
Senior unsecured bond	S$1.2bn	9 months
Secured repo	S$400m	7 days

A simple ladder places each maturity into time buckets:

MaturityNeed(h) = Σ liabilities maturing within horizon h.

This immediately reveals “maturity walls”—periods in which an unusually large amount must be refinanced at once.

3. Contractual maturity is not expected rollover

If S$1 billion matures next month, the bank does not necessarily need S$1 billion of cash from assets. It may expect to issue new debt. The risk model therefore introduces a rollover probability:

Expected refinancing amount = maturity × P(successful rollover).

and:

Expected funding gap = maturity × [1 − P(successful rollover)].

This is only a first approximation. A bank can refinance partially, refinance at a shorter tenor, post more collateral, or pay a much wider spread. “Success” is not binary in the real market.

4. Refinance probability should depend on the state of the world

A useful model conditions rollover on variables such as:

• the bank’s credit spread and rating;
• market volatility;
• system-wide bank spreads;
• investor fund flows;
• current utilisation of credit lines;
• collateral availability for secured funding;
• currency basis and swap-market conditions;
• recent issuance success or failure;
• maturity size relative to normal market capacity.

One generic formulation is logistic:

P(rollover) = 1 / [1 + exp(−β·x)].

The important question is not whether logistic regression is the “right” model. It is whether the chosen features and calibration remain informative when the market is stressed rather than merely during normal refinancing cycles.

5. Spread stress turns refinancing from a yes/no question into a price question

A bank may still access the market but only at a much wider spread. If a S$1 billion bond can be refinanced at 80 basis points over the benchmark in normal conditions and 280 basis points in stress, the extra annualised funding cost is approximately:

S$1bn × 2.00% = S$20m per year

before considering tenor, fees, hedging and other costs.

The model therefore needs at least two outputs:

• refinancing capacity;
• refinancing price.

A funding plan that survives only by issuing very short or prohibitively expensive debt may preserve immediate liquidity while damaging future profitability and creating an even larger next maturity wall.

6. Commercial paper makes rollover risk visible because the clock is short

Commercial paper often has very short maturity. The Federal Reserve noted in a February 2025 research note that commercial paper is typically very short-dated and subject to considerable rollover risk because primary markets can become fragile when major investors such as money market funds face outflows.

See Does Mutual Fund Fragility Impact Primary Market Pricing? Evidence from Commercial Paper.

This produces a feedback loop:

investor redemptions → lower demand for short bank paper → wider funding spread or failed issuance → issuer funding stress.

The bank’s own credit quality can be unchanged while its investor base becomes less able to roll the paper.

7. Funding concentration makes probability estimates correlated

Imagine ten S$100 million funding lines. They look diversified by lender count, but all ten investors are money market funds exposed to the same redemption shock. Treating the ten rollover decisions as independent dramatically overstates diversification.

Useful concentration dimensions include:

• single investor;
• investor type;
• funding instrument;
• secured versus unsecured;
• currency;
• geographic market;
• maturity date;
• distribution channel.

The interagency policy statement specifically warns against undue reliance on one funding source and calls for diversification by provider, secured/unsecured source, instrument type and market.

8. Maturity clustering is a concentration even when investor names differ

If S$4 billion of debt matures in the same week, the bank is making a large bet on market access during that week. A diversified investor base cannot fully eliminate this timing concentration.

A simple concentration statistic is the share of annual maturities occurring in the largest month or quarter:

MaturityConcentration = largest bucket maturity / total maturity over horizon.

More advanced models can penalise squared bucket shares, similar to an HHI, so multiple moderate concentrations still matter.

9. Secured rollover depends on collateral, not just lender appetite

A repo can be refinanced only if acceptable collateral remains available and the market haircut remains manageable. Under stress, the same security can require more collateral for the same cash amount.

For a secured borrowing amount F and haircut h:

Required collateral market value ≈ F / (1 − h).

If h rises from 2% to 10%, a S$100 million borrowing requires approximately S$102.0m of collateral at 2% but S$111.1m at 10%.

The funding problem therefore becomes a collateral problem. See How Banks Optimise Collateral for Repo and Margin.

10. Foreign-currency funding adds a basis and transfer problem

A bank can refinance in one currency and swap the proceeds into another, but cross-currency basis can widen sharply under stress. The effective refinancing cost is therefore:

Domestic funding cost + FX swap / cross-currency basis + collateral and liquidity effects.

This connects directly to How Cross-Currency Swap and FX-Swap Algorithms Price Funding.

A maturity ladder should therefore be built by currency, not only on a consolidated total.

11. NSFR can look healthy while a near-term maturity wall remains

The Net Stable Funding Ratio encourages structural funding over a one-year horizon. It is valuable, but it compresses many positions into ASF and RSF factors. Two banks can report the same NSFR while one has much more debt maturing next Tuesday.

That is why regulatory ratio and cash-flow ladder are complementary:

• NSFR — structural funding sufficiency;
• maturity ladder — exact timing of contractual and expected cash needs;
• rollover model — probability and price of replacing those liabilities.

12. Market access needs evidence, not a line in a contingency plan

Supervisory guidance says banks should maintain an ongoing presence in chosen funding markets and regularly gauge their capacity to raise funds quickly. That means “we can issue debt” should be tested against evidence such as:

• recent issuance sizes;
• order-book quality;
• secondary spreads;
• dealer/investor feedback;
• unused committed lines;
• repo capacity;
• settlement and documentation readiness;
• ability to issue at different tenors and currencies.

A funding source that has never been used or tested is an option with uncertain exercise value.

13. Current evidence: market access can reprice abruptly even when issuance resumes

The European Banking Authority’s June 2026 Risk Assessment Report described solid investor demand at the start of 2026, followed by wider spreads and a temporary slowdown in bank issuance after geopolitical stress at the end of February; issuance volumes later recovered. The important modelling lesson is not the particular event. It is that market access can move through states: open and cheap, open but expensive, thin, delayed, or closed.

See EBA Risk Assessment Report — June 2026.

Similarly, the Federal Reserve’s December 2025 supervision report noted that wholesale funding remained broadly stable in aggregate while short-term wholesale funding at large banks can be more costly and less stable than insured deposits. See Banking System Conditions.

14. A stress model should separate four failure states

• Quantity failure. Only part of the maturing amount can be replaced.
• Price failure. Full refinancing is available, but only at a sharply wider spread.
• Tenor failure. Funding is available only at a shorter maturity, increasing future rollover risk.
• Collateral failure. Secured funding capacity collapses because haircuts rise or eligible collateral is unavailable.

Putting all four into one “runoff percentage” loses mechanism. A bank needs to know which repair action would work.

15. A stochastic rollover simulation

For each liability j maturing at time t, simulate:

• renewal fraction R_j,t between 0 and 1;
• new spread S_j,t;
• new tenor T_j,t;
• collateral haircut H_j,t where secured;
• market-access state M_t.

Then the cash gap at time t is approximately:

Gap_t = Σ maturity_j,t × (1 − R_j,t) − alternative funding raised_t.

Running many paths produces a distribution of funding gaps rather than one deterministic answer. The tails reveal combinations that ordinary business plans often ignore.

16. Contingency funding is the alternative-route graph

If one market closes, the bank needs routes such as:

• drawing committed facilities;
• repoing or selling liquid assets;
• issuing longer-tenor debt;
• raising deposits;
• slowing asset growth;
• securitising or selling assets where feasible;
• using central-bank facilities where eligible and appropriate.

But the same stress can impair several routes simultaneously. A contingency funding plan should therefore model route correlation, not simply list alternatives.

17. Creative-work lens: a bridge whose replacement span arrives tomorrow

Imagine a bridge built from temporary sections that must be replaced on scheduled dates. It can carry traffic safely today while still containing a future failure if a replacement section does not arrive on time. Wholesale funding behaves similarly. The bank can be perfectly liquid now while accumulating a future cluster of maturities that all depend on market access later.

The analogy helps separate current cash from future refinancing capacity. The evidence must still come from contractual maturities, investor behaviour, spreads, collateral and tested market access.

18. The rollover-risk algorithmic pipeline

1. Inventory wholesale liabilities by amount, maturity, currency and legal entity.
2. Build contractual maturity ladders.
3. Identify maturity walls and provider concentrations.
4. Classify secured versus unsecured funding.
5. Estimate normal and stressed rollover fractions.
6. Estimate stressed refinancing spreads and tenor shortening.
7. Model collateral haircuts for secured sources.
8. Map alternative funding routes.
9. Run bank-specific and market-wide closure scenarios.
10. Calculate cash gaps and survival horizon.
11. Translate spread stress into PPNR/profitability effects.
12. Set maturity, investor and market limits.
13. Test actual market access periodically.
14. Update after every significant issuance, market disruption or funding-policy change.

19. Failure modes

• 100% rollover assumption. Every maturing liability is replaced because that usually happened historically.
• Binary market access. The model ignores the state “available but prohibitively expensive.”
• Investor-count diversification. Ten investors are treated as independent despite sharing the same funding shock.
• Maturity-wall blindness. Annual totals hide a one-week concentration.
• Collateral omission. Secured rollover is assumed without stressed haircut capacity.
• Currency aggregation. Strong domestic funding hides a foreign-currency maturity gap.
• Tenor shortening. Refinancing succeeds today by creating a worse maturity wall next month.
• Untested contingency sources. Funding options exist on paper but have not been operationally exercised.

20. Diagnostics and falsifiers

• What is the largest seven-day and 30-day wholesale maturity bucket?
• Which investor type supplies the greatest fraction of short-term funding?
• How much funding remains if unsecured markets close for 30 days?
• What is the extra annual cost if refinancing spreads widen 200 basis points?
• Which secured sources disappear first under stressed haircuts?
• Can the bank issue in the alternative currency without creating an FX basis problem?
• How often has each contingency source actually been tested?
• What observation would show that historical rollover probabilities are no longer usable?

Suppose someone claims, “The bank has always rolled its commercial paper, so next month’s maturity is low risk.” A falsifier is a market state in which the investor base suffers redemptions or refuses new short-term paper even though the bank itself has not yet defaulted. Historical renewal frequency does not guarantee future market capacity.

21. Verification and update triggers

• reconcile liability maturities to treasury and general-ledger systems;
• compare predicted rollover fractions with actual issuance outcomes;
• track spread forecast error separately from quantity forecast error;
• stress correlation across funding providers;
• retest repo capacity after collateral or haircut changes;
• rebuild ladders after material issuance or liability management exercises;
• update market-access assumptions after failed, delayed or undersubscribed deals;
• keep NSFR, LCR and maturity-ladder diagnostics distinct so one ratio cannot mask another constraint.

Connections across the finance-and-banking algorithms lane

• Liquidity stress testing — converts rollover failure into cash-flow survival.
• NSFR optimisation — the structural one-year funding constraint.
• HQLA buffer optimisation — the reserve available when refinancing does not arrive.
• Funds transfer pricing — translates funding tenor and liquidity cost into internal pricing.

Research anchors

• Federal Reserve — Interagency Policy Statement on Funding and Liquidity Risk Management.
• Basel Committee — Net Stable Funding Ratio.
• Federal Reserve — Banking System Conditions, December 2025.
• EBA — Risk Assessment Report, June 2026.
• Federal Reserve — Commercial paper rollover fragility research.

The deeper lesson

Wholesale-funding rollover risk is the mathematics of a future dependency. Today’s liability supplied cash in the past; its maturity creates a claim on tomorrow’s market. A maturity ladder reveals the dates. Rollover probabilities reveal uncertainty. Spread stress reveals price. Concentration reveals correlation. Collateral reveals secured capacity. The strongest model therefore does not ask only “What matures?” It asks “Who must choose to fund us again, under what market conditions, at what price, and what will we do if that choice changes?”

Educational note: This article explains public banking and funding-risk mathematics. It is not treasury advice, investment advice or a contingency funding plan for any institution.