Matched-maturity vs pooled FTP: choosing a transfer pricing method

The two methods, side by side#

Pooled funds transfer pricing applies one rate, or a small number of rates, to every transaction regardless of its maturity: the bank calculates an average cost of funds and charges lending businesses that rate while crediting deposit businesses with it.

Matched-maturity funds transfer pricing assigns each transaction the rate corresponding to its own maturity on an FTP curve, and holds that rate fixed for the life of the transaction.

The two methods answer the same question — what did this money cost — with opposite levels of resolution, and the difference is not cosmetic.

Small, short and homogeneous#

A single pooled rate is defensible in a narrow set of circumstances:

• a small institution
• a short and homogeneous balance sheet
• funding concentrated in one tenor
• no material term transformation

Where those conditions genuinely hold, the additional precision of a curve buys very little, and the operational cost of maintaining one is real. The trouble is that most banks describe themselves as fitting those conditions long after they have stopped fitting them, usually because the mechanism was set up when they did and nobody revisited it as the balance sheet lengthened.

The ten-year mortgage and the thirty-day facility#

What a single rate hides is term transformation, which is where a large part of a bank’s interest-rate risk and a large part of its apparent profit both live. Charging a ten-year fixed-rate mortgage the same funding rate as a thirty-day working-capital facility means the mortgage is being funded, in the internal accounts, at short-term cost.

It will therefore appear to carry a wide margin, and the business writing it will be rewarded for volume, while the actual cost of funding it to term — and the risk of refinancing it repeatedly through a rate cycle — sits unpriced in the treasury book.

The subsidy runs consistently in one direction: long lending is flattered, short lending is penalised, and the strategy quietly tilts toward whichever product the mechanism happens to favour.

Multiple maturity pools#

Multiple pools are the usual intermediate step: instead of one rate, the bank runs a handful of maturity buckets — short, medium, long — and prices each transaction into the appropriate bucket. This removes the worst of the distortion at a fraction of the implementation cost of a full curve, and for many mid-sized institutions it is the right stopping point for several years.

Its weakness is at the boundaries, where a transaction just inside one bucket is priced identically to one at the far end of it, which creates an incentive to structure maturities to land favourably. Where that gaming appears, it is a reliable signal that the buckets have become too coarse for the book.

The rate that is set at origination#

Matched maturity works differently in one respect that matters more than its precision: the rate is set at origination and does not change. A five-year loan written today carries today’s five-year transfer rate for its entire life, whatever the curve does afterwards.

That single design choice is what makes the resulting margin a fair measure of the pricing decision the business actually made, and it is what allows a lending book’s performance to be assessed years later without re-litigating market moves nobody controlled.

Frameworks that refresh transfer rates on the back book each period lose this property immediately, and reintroduce exactly the noise the mechanism exists to remove.

What matched maturity demands#

The demands are correspondingly higher.

Matched maturity needs:

• transaction-level data with a reliable maturity or repricing date
• a curve maintained and published on a stated cycle
• behavioural models for every product without a contractual maturity
• a system able to store the rate assigned at origination and carry it forward for years

The last of these defeats more implementations than the modelling does: banks that can build the curve often cannot retain the rate against the transaction, and end up recalculating history each month, which is the pooled-rate problem wearing better clothes.

What to fix first#

For an institution that cannot yet reach transaction level, the honest sequence is to fix the distortions in order of size rather than to wait for the target system.

1. Separate the book into maturity buckets first, since that removes most of the term subsidy.
2. Add an explicit liquidity component next, because a zero liquidity charge misprices undrawn commitments and volatile deposits more severely than an imperfect term rate ever will.
3. Build behavioural models for the non-maturity deposit base third, since that is usually the largest single balance in the bank.
4. Transaction-level matched maturity comes last, and often only for new business, with the legacy book run on buckets until it amortises.

Three questions that decide the method#

Choosing between the methods is best done with three questions rather than a maturity model.

1. First: how much term transformation does the balance sheet actually carry — if the weighted average life of assets and liabilities differ materially, a single rate is misallocating a large number.
2. Second: does the product range differ enough in maturity and optionality that one rate would price a mortgage and an overdraft identically — if so, the pricing signal is already wrong.
3. Third: is anyone making decisions on the numbers — because a bank that does not use FTP output in product pricing, incentive setting or portfolio decisions will get no return on a more precise mechanism, and should fix the use before fixing the method.

The real argument for moving up the ladder#

The comparison is often framed as accuracy against simplicity, which understates what is at stake.

A pooled rate is not merely less accurate; it is systematically biased in a known direction, and the bias compounds because the businesses it flatters grow.

Over a few years a mechanism nobody examines can reshape a balance sheet toward exactly the maturity and optionality profile the institution would not have chosen deliberately. That is the real argument for moving up the ladder — not precision for its own sake, but removing a standing incentive that points the wrong way.