On most bank balance sheets, the largest liability has no maturity date. Current accounts, savings accounts and instant-access deposits can be withdrawn on any day, yet banks lend against them, hedge them and hold liquidity for them on the assumption that most of the money will stay for years. How long it stays, and how far its rate follows the market, is written in no contract.
That gap between what the contract allows and what customers actually do has to be estimated, and the estimate rarely stays inside the model that produces it.
This article examines what non-maturity deposits are, how their balances behave, the vocabulary banks use to describe that behavior, and why a single behavioral assumption ends up shaping so many balance sheet decisions at once.
Salaries, household savings and the working cash of companies sit in current accounts and instant-access savings accounts. The European Banking Authority (EBA) found in its February 2025 report on its interest rate risk heatmap that, for half of the 120 institutions it reviewed, these balances made up more than 50% of all liabilities. In the US, the advisory firm Wilary Winn puts them at about 85% of industry deposits.
These balances are known as non-maturity deposits, or non-maturing deposits (NMDs). An NMD is a deposit whose contract sets no end date: the customer can withdraw part or all of it at any time, usually without notice or penalty, and the bank can change the rate it pays whenever it chooses. Fixed-term deposits fall outside the category, because they carry a contractual maturity and a rate agreed for the whole term.
Both sides of the contract therefore hold an option. The customer decides when the money leaves; the bank decides what the money earns. Because neither decision is fixed in advance, every bank has to form its own view of how long the money will in fact stay, and at what price.
Money that can leave today stays for years because customers use these accounts to live and to run their businesses, and moving them is a nuisance. A salary arrives at the end of the month, rent and card payments draw it down, and the next salary refills it. A single account can swing from almost nothing to several times its average within a month. Across thousands of accounts, however, the swings largely offset each other, and the total balance moves around a level that persists for years.
The rate the bank pays is just as slow to move. The European Central Bank (ECB) measured the pattern in its Financial Stability Review of May 2023: over 2007 to 2021, euro area overnight deposit rates passed on about 23% of market rate changes to households and 32% to companies, against about 68% and 86% for term deposits. Deposits that look short on paper behave as long money, and they reprice partly and late.
Every NMD model, report and regulation is written in a small shared vocabulary, and each term names one feature of that behavior:
The first three terms describe how much of the money stays and for how long, the fourth describes what it costs, and the last explains why the answers differ from one group of customers to another.
The UK Prudential Regulation Authority (PRA) rulebook defines core deposits as those "found to remain undrawn with a high degree of likelihood". Contractually, every NMD can leave tomorrow; behaviorally, the core may stay for years. That longer horizon has two dimensions: how long the balance stays, and how long its rate stays unchanged. A balance can remain with the bank for a decade while its rate is reset every few months.
Decay is measured by following a group of accounts and recording how much of the original balance remains after a month, a year or five years. The resulting survival curve falls quickly as the most mobile money leaves, then flattens. Deposit beta covers the price side: a beta of 23%, the historical euro area figure for household overnight deposits, means that a one percentage point rise in market rates lifts the deposit rate by less than a quarter of a point.
Segmentation matters because the differences between groups change the answer. In the EBA's review, the average share of balances treated as core was about 47% for retail NMDs against 37% for wholesale NMDs from non-financial companies. A bank that models its deposits as a single pool averages away exactly the behavior that decides how much of its funding it can count on.
Non-maturity deposits are, for most institutions, the cheapest funding available. Customers accept a low rate, and often none on current accounts, in exchange for a place for salaries to land, cards and payments, and an insured account. In November 2025, the ECB reported an average rate of 0.25% on household overnight deposits, against 1.75% on household term deposits of up to one year. The difference between what a bank pays on these balances and what it earns by investing them is known as the deposit franchise.
This franchise grows more valuable as rates rise, because market rates climb faster than the rate paid on sticky balances. A 2024 analysis by the Federal Reserve Bank of San Francisco put the value of the US deposit franchise at about $1.7 trillion at the end of 2022, against about $1.75 trillion of unrealized losses on bank loans and securities. The two figures nearly cancel.
The offset holds only under one condition, which the same analysis states plainly: the franchise is worth nothing to a bank that fails. Its value depends on the deposits staying.
Consider a bank funding fixed-rate mortgages with current accounts. If it treats the accounts as overnight money, the balance sheet shows a large mismatch, and any rise in rates appears to squeeze the margin and cut the value of the bank. If it treats the core as money that stays and reprices slowly over several years, the same deposits look like long fixed-rate funding and the reported risk falls. Nothing on the balance sheet has changed. Only the assumption has.
That choice runs through both measures of interest rate risk in the banking book (IRRBB). For the economic value of equity (EVE), the behavioral maturity sets the duration of the liability side. For net interest income (NII), the assumed beta and repricing lag decide how quickly deposit costs follow market rates, so a bank assuming a low beta reports a margin that widens as rates rise, while the same bank with a higher beta reports a much smaller gain from an identical book.
The assumption then shapes the balance sheet itself. Treasury typically hedges through a replicating portfolio, treating core deposits as if invested in a rolling ladder of fixed-rate instruments whose average maturity matches the behavioral maturity, so the deposit model sets the size and duration of the hedge book. Through funds transfer pricing (FTP), the same estimate decides how the bank rewards the business that gathers deposits, and projected deposit volumes and costs drive the financial plan.
The sums involved are large. A large European bank disclosed in 2018 that a model update shortening the modeled duration of its deposits by about 0.5 years, across about €185 billion of balances, shifted an estimated €360 million a year in funds transfer pricing, against about €40 million of additional hedging cost.
Half a year of assumed depositor behavior moved hundreds of millions of euros between business lines every year.
The same estimate also sets the liquidity buffer. Under stable conditions, reserves and liquid securities absorb the daily movement of non-core balances. Under stress, the non-core leaves quickly and part of the core can follow, so the bank must hold enough liquid assets to meet withdrawals without forced sales of loans or bonds. Every euro classified as core is a euro the bank assumes it will not need to repay in a hurry.
The two functions therefore pull in opposite directions. To protect the margin against falling rates, treasury wants to invest core deposits at long maturities, which requires assuming that they stay; to survive a run, the liquidity function wants to assume that they may leave. A 2024 working paper from the Bank Treasury Risk Management (BTRM) Programme concludes that "the bank cannot simultaneously hedge its interest rate risk and liquidity risk exposures" from the same deposit book.
Better modeling does not remove that trade-off, although it can be managed coherently when both functions start from the same view of how the deposits behave. In my experience, many institutions still set the interest rate risk and liquidity versions separately, so the same deposits count as long-term money in one report and a flight risk in another.
No regulator tells a bank how its depositors behave. The Basel, EU, UK and US frameworks all leave the estimate to the institution, because only the institution holds the data on its own customers. Supervisors fence it in with caps on core balances and repricing maturity, outlier tests on EVE and NII, minimum run-off rates in the liquidity ratios and requirements to document and backtest assumptions. These limits rule out the most aggressive assumptions without confirming that any assumption inside them is right.
A bank can therefore sit comfortably inside every cap and still hold assumptions its own depositors would not recognize. Because that one estimate sets the buffer, the reported risk, the hedges and the price of deposits at the same time, it belongs on the agenda of the CFO and the asset and liability committee (ALCO), and not only in the model owner's calibration file.
For the full toolkit, from core balance estimation and deposit betas to replicating portfolios, and how Basel, EU, UK and US rules cap these assumptions, see Mirai RiskTech's whitepaper Non-Maturing Deposits Under Pressure: how banks model non-maturity deposits for IRRBB and liquidity.