Quick answer: funds transfer pricing (FTP) is the internal algorithm a bank uses to assign a cost of funds to assets and a funding benefit to liabilities. A five-year fixed-rate loan should not look profitable merely because the branch compares its customer rate with today’s overnight deposit cost. FTP instead asks what it costs the bank to fund the loan for the relevant economic tenor, adds liquidity and option effects where appropriate, and transfers the resulting interest-rate and funding risk to a central treasury or asset-liability-management function. The output changes both measured profitability and behaviour.

An internal price is not just an accounting number. It is an instruction about which businesses the bank will encourage itself to grow.

Why this belongs in mathematics

FTP combines yield curves, discounting, weighted averages, behavioural maturity, liquidity spreads, optionality and performance measurement. But its most interesting mathematical feature is indirect: the chosen measurement rule changes incentives. If a long-dated loan is charged a short-term funding rate, the lending unit can appear unusually profitable while the bank silently accumulates maturity and liquidity risk elsewhere.

The Federal Reserve’s interagency FTP guidance makes exactly this point in risk-management language: FTP should allocate funding and contingent-liquidity costs to the businesses creating those risks rather than leaving them invisible at firm level. See the Interagency Guidance on Funds Transfer Pricing.

1. The basic internal transaction

Imagine a lending unit originates a S$1 million five-year fixed-rate loan at 6%. The branch has created an asset that uses funding for five years. A simple FTP system creates a synthetic internal transaction:

The lending unit is treated as if it borrows S$1 million internally from Treasury.
Treasury charges an internal funding rate appropriate to the loan’s maturity, repricing structure, liquidity use and optionality.
The lending unit keeps the residual spread attributable to customer pricing, credit risk, operating costs and commercial margin.
Treasury receives the interest-rate and liquidity-risk position that it is designed to manage centrally.

The deposit side works in reverse. A deposit-gathering business can receive an FTP credit because it provides funding to the bank.

2. Matched maturity: compare like with like

A common non-trading FTP approach is matched-maturity marginal cost of funding. Instead of assigning one bank-wide funding rate to every product, the system maps the economic maturity or repricing profile of each product to the bank’s internal funding curve.

The Federal Reserve describes matched-maturity marginal cost of funding as a commonly used methodology in which longer-dated assets are charged the cost of funding for their life, while stable funding such as core deposits can receive an appropriate benefit. This prevents a five-year loan from being judged against a one-day funding cost.

3. Building the internal FTP curve

An FTP curve is not necessarily a government yield curve. The bank is not the government, and its own marginal funding cost can contain bank-specific credit and liquidity components.

A simplified internal curve can be thought of as:

FTP rate(t) = reference term rate(t) + bank funding spread(t) + liquidity adjustment(t) + product-specific adjustments.

Different institutions construct the curve differently. The crucial governance requirement is not that every bank use one universal formula; it is that the methodology reflects the bank’s actual funding economics, is documented, understood by users and updated when those economics change.

The OCC’s current Interest Rate Risk Comptroller’s Handbook asks examiners to assess the source of the FTP funding curve, whether it reflects true borrowing costs, and whether charges are adjusted for liquidity, basis and embedded options.

4. A worked profitability example

Suppose the S$1 million loan earns 6%. The matched FTP charge is 4.2%. Assume an annualised expected credit cost of 0.6% and operating cost allocation of 0.4%.

Component	Rate	Annual amount on S$1m
Customer loan yield	6.0%	S$60,000
FTP funding charge	−4.2%	−S$42,000
Expected credit cost	−0.6%	−S$6,000
Operating cost	−0.4%	−S$4,000
Residual commercial margin	0.8%	S$8,000

Now imagine the branch had been charged an artificial 2% short-term FTP rate instead. Its apparent residual margin would jump by S$22,000 even though the customer loan and the bank’s real five-year funding exposure had not changed. The measurement system would be rewarding the branch for a risk transfer it did not actually solve.

5. Liquidity premiums: term matters even when the rate resets

A floating-rate five-year loan may reprice every three months, but that does not mean it requires only three months of liquidity. The borrower can keep the bank’s money for five years. An FTP framework therefore needs to distinguish interest-rate repricing from funding tenor.

This is why liquidity transfer pricing became a major supervisory focus after the global financial crisis. Short-term wholesale funding looked cheap until markets stopped rolling it. A robust framework should make long-term or contingent use of liquidity visible to the business line creating it.

6. Contingent liquidity is still a cost

Suppose a corporate client has a S$100 million committed credit line but normally draws only S$20 million. The undrawn S$80 million does not consume funded cash today, yet the bank has promised to provide it if contractual conditions are met. In stress, customers may draw exactly when external funding becomes more expensive.

An FTP system can therefore assign a contingent-liquidity charge to the commitment. The point is not to predict the exact draw. It is to prevent an apparently fee-rich product from being treated as costless simply because the funding need has not happened yet.

7. Non-maturity deposits create a behavioural-maturity problem

A current account may be withdrawable tomorrow, yet a large portion of a bank’s aggregate current-account balances can persist for years. Contractual maturity is therefore near zero while behavioural maturity may be much longer.

FTP cannot simply give every demand deposit an overnight funding benefit or assume it funds the bank for ten years. The bank needs evidence about deposit stability, pricing sensitivity and decay. The Federal Reserve guidance explicitly recognises that non-maturity deposits require special treatment because their horizons are less predictable.

This connects directly to the companion article in this batch on deposit betas, decay curves and digital run risk: behavioural maturity is not a fixed property of the account contract. It is an empirical property of a customer population under changing conditions.

8. Option costs: customers can change the expected cash-flow path

A mortgage borrower may prepay early. A term depositor may break a deposit if the contract permits it. A capped or floored floating-rate product changes behaviour when market rates cross its option boundaries.

If FTP ignores those options, one business can originate products whose option cost appears later inside Treasury. A strong FTP system therefore asks whether embedded prepayment, withdrawal, cap or floor behaviour needs to be priced explicitly.

For mortgage optionality, see How Banks Model Mortgage Prepayment.

9. FTP is an incentive system

Suppose one deposit product receives too generous an FTP credit. Sales teams can be rewarded for collecting balances that are actually highly rate-sensitive and unstable. Suppose long loans are undercharged for term liquidity. Lending teams can grow duration risk while reporting strong product margins.

The Federal Reserve guidance therefore connects FTP to business-line incentives, product pricing and new-product approval. A risk that is not charged to the activity creating it can become somebody else’s problem until it becomes the whole bank’s problem.

10. Creative-work lens: Moneyball and the danger of the wrong metric

Moneyball is not a banking source, but it is a useful creative lens because its central tension is measurement: if an organisation rewards the wrong statistic, people rationally optimise toward the wrong statistic. FTP creates the same structural question inside a bank. A branch manager does not need to be reckless to create hidden maturity risk; the manager may simply be following the profitability measure the institution supplied.

The creative work helps us notice incentives. The banking conclusion must still be grounded in funding curves, liquidity behaviour and supervisory evidence.

11. The algorithmic pipeline

Classify the product. Fixed, floating, amortising, bullet, non-maturity, optional or contingent.
Map contractual cash flows. Principal, interest, reset dates and commitments.
Estimate behavioural cash flows where needed. Prepayment, deposit decay and drawdown assumptions modify contractual timing.
Select the appropriate FTP curve. Currency and business funding structure matter.
Match maturity or repricing buckets. Avoid using one rate for economically different tenors.
Add liquidity and basis adjustments. The bank’s true funding economics may differ from a reference curve.
Add option or contingent-liquidity costs where justified.
Lock or reset the transfer rate according to product design. Fixed-rate and floating-rate products require different treatment.
Allocate the resulting margin. Separate customer spread from centrally managed interest-rate and liquidity risk.
Report profitability by product and business line.
Backtest behavioural assumptions. Compare actual prepayments, deposit retention and funding costs with assumptions.
Update governance when markets or business models change.

12. Failure modes

Single-rate FTP. Overnight and ten-year products receive the same internal funding rate.
Reference-curve illusion. A government or swap curve is used without the bank’s actual marginal funding spread.
Liquidity blindness. Floating-rate assets are treated as short funding simply because their coupon resets quickly.
Behavioural overconfidence. Demand deposits are assigned long maturities using stale historical stability.
Option blindness. Mortgage prepayment or deposit early-withdrawal options are left in Treasury without an internal charge.
Contingent-liquidity blindness. Undrawn commitments appear profitable because only funded balances are charged.
Cross-subsidy. One product looks cheap because another business silently carries its funding burden.
Metric gaming. Business lines optimise product mix around transfer-price weaknesses rather than economic value.

13. Diagnostics and falsifiers

Does a five-year asset receive a materially different funding charge from a one-month asset?
Which part of the FTP rate represents term liquidity rather than pure interest-rate risk?
How is a non-maturity deposit’s behavioural life supported by data?
Does product profitability reverse if the bank’s real marginal funding spread is used?
Who receives the gain or loss when rates change after a fixed-rate loan is originated?
Are prepayment and early-withdrawal options priced consistently with observed behaviour?
Do undrawn commitments receive a liquidity charge?
What business decision would change if the FTP model were corrected?

Suppose someone claims, “This lending business earns a 4% margin because it lends at 6% and deposits cost 2%.” A falsifier is an FTP calculation showing that the relevant five-year marginal funding cost is 4.2%, not 2%. If the original profitability disappears when like is compared with like, the 4% margin was an artefact of the measurement rule.

14. Verification and update triggers

reconcile the FTP curve with observable marginal funding costs;
compare predicted and realised deposit retention;
backtest loan prepayment and commitment drawdown behaviour;
review whether Treasury’s residual risk matches the risks supposedly transferred to it;
check for systematic profitability shifts caused only by FTP methodology changes;
update liquidity premiums when funding markets reprice;
revisit behavioural maturity after digital-channel or customer-base changes;
independently validate material model assumptions and governance.

Connections across the finance-and-banking algorithms lane

Yield-curve algorithms — the term structure beneath matched-maturity pricing.
Liquidity stress testing — the firm-wide risk FTP is supposed to allocate back to its source.
Interest-rate risk — the Treasury risk that customer-facing units should not accidentally retain or hide.
Loan amortisation — contractual cash-flow timing used to match internal funding tenors.

Research anchors

The deeper lesson

FTP is a reminder that accounting inside an organisation is also control engineering. If internal prices do not reflect the real term, liquidity and option costs of a product, the bank can train itself to create the wrong balance sheet while every business line appears individually successful. A strong FTP algorithm therefore makes hidden funding economics visible at the point where decisions are made.

Educational note: This article explains banking mathematics and public risk-management concepts. It is not financial advice, product-pricing advice or institution-specific regulatory guidance.

Small Group Tutorials

How Banks Use Funds Transfer Pricing Algorithms: Internal Yield Curves, Liquidity Premiums, Behavioural Maturity and Risk-Adjusted Profitability