Insights Feed

Automated FTP: Millions of Contracts in About 30 Minutes

Written by Álvaro Nieto | Nov 19, 2026, 8:00:06 AM

A real margin for one loan is a subtraction. A real margin for every loan and deposit in a bank's book, recalculated every night, is a problem of scale, and it decides whether funds transfer pricing (FTP) ever becomes more than a treasury report.

The rate cycle made that problem urgent. When central banks raised rates sharply in 2022 and 2023, net interest income (NII) climbed across most of the industry, and reported results could not say on their own how much of it came from commercial decisions and how much from the level of rates. The answer runs through FTP, the process that sets the real cost of the money behind every loan and deposit, and it has to hold for every contract in the book.

This article examines the numbers behind that question: what the rate cycle added to bank returns, what remained as the lift receded, and what it takes to calculate a real margin for millions of contracts overnight, on the same assumptions the bank uses for liquidity and regulatory ratios.

What the Rate Cycle Added to Bank Returns

McKinsey's Global Banking Annual Review 2024 puts the industry's return on tangible equity (ROTE) at 11.7% in 2023 and attributes most of the improvement since 2021 to wider net interest margins. The same review carries a warning: without the tailwind of higher rates, industry ROTE in many geographies would have been around 8%, near or below the cost of capital.

The 8% figure is an estimate rather than a measurement, and it describes the industry, not any single bank. Its value lies in the order of magnitude. A large part of the profitability of those years came from where rates stood rather than from how products were priced or where the commercial network sold them.

Rising rates reward the way a balance sheet is positioned as well as the way its products are priced, and both are legitimate sources of income. The difficulty lies in telling them apart. Many institutions could see NII climbing month after month without being able to say how much of the increase was contribution margin and how much was the rate effect, which is the part that fades as soon as rates come down.

What Remained When the Lift Receded

By mid-2025 the tailwind had begun to fade. The European Banking Authority (EBA) risk dashboard for the second quarter of 2025 put the net interest margin of the EU and EEA banks it covers at 1.58%, down from 1.68% a year earlier, and net interest income had fallen back to levels last observed in December 2023. The figures describe the banks in the EBA's sample, not every market.

While rates were rising, the lift reached most of the business at once. A product priced too close to its FTP rate still appeared to contribute, and a branch whose lending barely covered its funding cost still showed growing income. As margins tightened, that lift receded from all of them together, and what remained on each line was its contribution margin, positive or negative.

The next move may run the other way. In the reading of Álvaro Nieto, Head of Product at Mirai, the environment now points to a possible rise in rates, which would bring the lift back and cover the same lines again. Whichever way rates move, the question for management is which products, business areas and branches remain profitable once the rate effect is taken out. Nieto frames the role of FTP in those terms: "FTP is key for management: it is how the bank knows what each product, area and branch really earns."

Why the Real Margin Is a Problem of Scale

For a single contract, the arithmetic of a real margin is simple. As what funds transfer pricing really measures sets out, every loan and deposit receives an FTP rate, made up of a base rate taken from the yield curve plus FTP spreads that each institution defines for itself; a liquidity spread, an optionality spread and a regulatory spread are one example. The rate the bank applies minus the FTP rate is the contribution margin. A 30-year mortgage priced at 4% against an FTP rate of 3% leaves a contribution margin of 1%, in round numbers chosen for illustration rather than market rates.

The difficulty starts when that subtraction is repeated across the book. A real FTP takes every contract, projects every cash flow and prices the base rate and every FTP spread against it, each on the bank's real curves and behavioral assumptions. It then has to do so again the next day, because curves move, balances run off and new business is booked. At the largest institutions, the contract count runs into the millions.

This volume explains why so many margins are still calculated on averages, the pattern behind why spreadsheet FTP stays a report. An average funding rate charges the 30-year mortgage what it charges a one-year loan, so the liquidity spread for money tied up for three decades drops out of the calculation without anyone deciding to drop it.

What a Spreadsheet Can Carry

Where a bank has no dedicated FTP engine, the calculation runs by hand in spreadsheets, on simplified curves and assumptions, on averages rather than contracts, and slowly. Nieto is direct about where that leads: "With spreadsheets it is very hard to run a real FTP, with real assumptions, real yield curves and market-standard methodologies. What you get is a simplified model that doesn't hold up in a real use case." The survey evidence points the same way, although none of it measures spreadsheet use directly: the Bank for International Settlements (BIS) Financial Stability Institute, surveying liquidity transfer pricing at 38 large banks in nine countries, found that many relied on manual, offline processes to update their transfer prices.

A number that is approximate, averaged and late can describe last quarter. It cannot change next quarter's pricing.

Millions of Contracts in About 30 Minutes

The alternative is a process that runs by itself. In the implementations Mirai runs with client institutions, contract-level FTP is a fully automated overnight process with no manual step anywhere in it:

  • data loaded directly from the bank's source systems
  • the FTP calculation and its scenarios run on that data, with no one preparing or adjusting it in between
  • the FTP rate for every contract, broken down into its base rate and each FTP spread, written back into the operational systems where pricing, reporting and commercial teams already work

Each step removes a hand-off at which a manual process tends to go wrong: a figure retyped, a file filtered, a curve left stale. The user does not intervene at any point, and the contribution margin is simply there in the morning, contract by contract.

Speed is what makes that cycle possible. At one client, a global systemically important bank (G-SIB) that is not named, the calculation processes millions of contracts in about 30 minutes. In the institutions Nieto has worked with, automating the process takes it from hours, and in some cases days, to minutes.

A calculation that takes days can be run at month-end and reported. One that takes half an hour can run every night and still leave time to run again when a curve moves.

Under Stable Conditions and in Volatile Ones

Under stable conditions, a slow calculation is an inconvenience. Curves drift rather than jump, and last month's FTP rates are close enough to this month's for most pricing decisions.

In more volatile periods, curves can move more in a week than in a quiet quarter, and a margin that arrives days late describes a balance sheet that has already changed. The delay becomes the error.

Supervisors make a related point from the side of interest rate risk control. The Basel Committee on Banking Supervision's standard on interest rate risk in the banking book asks for management information that retrieves accurate data in a timely manner on all material exposures, and it is direct about how: "Data inputs should be automated as much as possible to reduce administrative errors."

One Set of Assumptions for Margin, Liquidity and Ratios

A fast contribution margin becomes more useful when it is calculated from the same data load and on the same assumptions as the bank's liquidity and regulatory metrics. Decisions do not travel through the balance sheet one metric at a time.

For example, a decision to grow long-dated fixed-rate mortgages changes the contribution margin through the base rate and the FTP spreads charged against each loan. It also changes the liquidity position through the funding the new book consumes, and the regulatory ratios through the capital it requires. Similarly, a campaign to gather term deposits improves the liquidity position and earns the deposit side a liquidity credit, while the rate paid to win those balances lands on its contribution margin.

When FTP, liquidity and regulatory metrics run in separate environments, each tends to carry its own assumptions. The interagency guidance on FTP issued by the Federal Reserve, the FDIC and the OCC (SR 16-3) asks firms whose FTP assumptions differ from those used in their liquidity stress testing to document the inconsistencies. If the FTP curve assumes a block of current accounts stays for years while the liquidity stress test assumes much of it leaves within weeks, the margin and the liquidity position describe two different banks.

One set of assumptions removes that gap at its source. The whitepaper Funds Transfer Pricing as a Steering Mechanism describes the full overnight process and how one set of assumptions links margin, liquidity and regulatory ratios, so that the bank sees one decision's effect on all three in the same run rather than three versions that someone has to reconcile.

What the Numbers Mean for Institutions

The rate cycle showed how far reported returns can drift from what the underlying business earns, and the next rate move will shift reported margins again. What counts is the ability to take the rate effect out of every contract, every night, on the assumptions the rest of the bank already uses. Institutions that see their real margin every morning will know what stays profitable when rates move. Those working from the reported margin will find out only after rates have moved.

The Mirai Platform is an example of how this works in practice. It brings FTP, balance sheet management and regulatory compliance together on one platform, fed by one data load and run on one set of assumptions, so that the behavioral life assumed for a block of current accounts is the same one that sets its FTP credit and the same one the liquidity metrics read.

For the full framework, from the base rate and FTP spreads to contribution margin by product, area and branch, calculated overnight and tested with what-if strategies, see Mirai RiskTech's whitepaper Funds Transfer Pricing as a Steering Mechanism: how banks measure and steer the real margin.