Do Tokenized Deposits Shrink Bank Lending Capacity?
Tokenized deposits can shrink bank lending capacity by making deposits less sticky, because banks fund long-dated assets on the assumption that depositors do not move quickly. A Federal Reserve Bank of Dallas research article published on 25 August 2026 models the effect: a 10-percentage-point increase in deposit rate beta would remove roughly $700 billion of 10-year-equivalent duration risk appetite from US banks, assuming a four-year deposit weighted average life. The mechanism is programmable deposit tokens combined with agentic AI, which let balances chase yield without the holder doing anything. The authors state the scenario is falsifiable: if adoption is large and betas stay flat, the concern is overblown.
TL;DR — Key Takeaways
- ✓The Paper: Federal Reserve Bank of Dallas research article, 25 August 2026, on tokenized deposits, bank liquidity and maturity transformation. Authors' views, not an official Fed position.
- ✓The Number: A 10-point rise in deposit rate beta removes about $700 billion of 10-year-equivalent duration risk appetite, assuming a four-year deposit weighted average life.
- ✓The Mechanism: Programmable deposits plus agentic AI remove switching friction. Deposit stickiness is largely inattention, and software does not get distracted.
- ✓The Precedent: Brazil's Pix — roughly 200 million users, about $650 billion monthly in Q1 2026 — coincided with banks shifting toward government bonds and subprime lending.
- ✓The Test: Falsifiable by design: large-scale tokenized deposit adoption with flat betas would void the estimate.
The First Serious Argument Against Tokenized Deposits Is About Funding
Tokenized deposits have been debated as a settlement question — how fast, how interoperable, how safe. A research article published by the Federal Reserve Bank of Dallas on 25 August 2026 asks a different one: what happens to the bank on the other side, whose lending book is funded by deposits that were previously slow to leave.
The answer is quantified. A 10-percentage-point increase in deposit rate beta would reduce US banks' appetite for 10-year-equivalent duration risk by roughly $700 billion, assuming deposits have a weighted average life of four years. Duration appetite is the capacity to hold long-dated loans and securities, which is another way of saying the capacity to lend long.
The argument does not require tokenization to fail. It requires tokenization to work, and traces the consequence through the funding side of the balance sheet.
“If large-scale tokenized deposit adoption occurs and deposit rate betas remain flat, the $700 billion duration figure evaporates and the concern is overblown.”
— Federal Reserve Bank of Dallas research article, 25 August 2026
Research that names its own disproof is rare enough in this field to be worth reading closely.
Banks Lend Long Because Depositors Are Slow
Deposit rate beta measures how much a bank must raise deposit rates when market rates rise. A beta near zero means deposits stay put regardless of what better-paying alternatives exist. Maturity transformation — funding 10-year assets with overnight liabilities — is only viable because beta is well below one.
Stickiness is not loyalty. It is friction: switching requires noticing the rate gap, opening an account elsewhere, moving balances and updating whatever pays out of them. Retail and small-business depositors routinely leave money in accounts paying materially less than money market funds, and the banking system's capacity to hold duration is built on that inattention.
| Deposit rate beta | What it implies | Effect on long-dated lending |
|---|---|---|
| Near zero | Funding cost barely moves with market rates | Wide capacity to hold duration |
| Rising 10 points | Funding cost tracks rates more closely | About $700 billion less 10-year-equivalent appetite |
| Approaching one | Deposits price like wholesale funding | Maturity transformation stops paying |
The paper's claim is narrow and therefore strong: tokenization does not change depositor preferences, only the cost of acting on them.
Software Does Not Get Distracted
The mechanism is the combination, not either half. Programmable deposit tokens make balances movable by instruction; agentic AI supplies an instruction-giver that monitors rates continuously and never postpones. Together they let money switch banks near-instantaneously without the holder deciding anything.
This is the part that should interest anyone building deposit token infrastructure, because it is an unintended consequence of features that are being marketed as benefits. Programmability, 24/7 availability and instant transfer are the product. Reduced deposit stickiness is the same property described from the bank's side of the ledger.
Note also that no consumer behaviour change is required. Historically, deposit flight needed a trigger — a rate shock or a loss of confidence — and depositors who paid attention. An agent authorised once and left running removes both conditions, and yield-chasing becomes the default state of the balance rather than an event.
The networks that would carry these movements are examined in why two banks' deposit tokens cannot talk to each other, and the interoperability layer being built for them in Swift's shared ledger orchestration approach.
Brazil Already Ran a Version of This Experiment
Pix, Brazil's instant payment system, reached roughly 200 million active users and about $650 billion in monthly transactions by Q1 2026. A 2025 academic study cited in the Dallas analysis found that greater Pix usage pushed Brazilian banks toward government bonds and subprime lending while reducing traditional credit intermediation.
The parallel is imperfect and worth stating precisely. Pix is a payment rail, not a deposit token, and it does not automate yield-seeking. What it did was make moving money nearly costless and instantaneous, which is the same friction removal from a different direction — and the funding-side response was measurable.
The direction of the Brazilian adjustment is the instructive part. Banks did not simply lend less. They changed what they lent against: shorter, more liquid, more collateralised or better compensated for risk. If tokenized deposits move US deposit betas, the visible effect will likewise be a change in the composition of credit rather than an aggregate contraction, and composition shifts are considerably harder to spot in headline statistics.
What the $700 Billion Figure Is and Is Not
It is a modelled scenario under stated assumptions, not a forecast and not an official Federal Reserve position. It depends on a four-year deposit weighted average life and on a 10-percentage-point beta increase that has not happened. Quoting it as a prediction of $700 billion in lost lending would misrepresent the work.
What makes it useful is that it converts a vague worry into a monitorable quantity. Supervisors and treasurers can watch realised deposit betas as tokenized deposit volumes grow, and the relationship between the two series either appears or it does not.
Reading the paper accurately
- It is conditional. The estimate follows from a beta increase; it does not assert one will occur.
- It measures duration appetite, not deposits. The claim concerns capacity to hold interest-rate risk, not deposits leaving the system.
- It is falsifiable and says so. Adoption with flat betas voids the number, on the authors' own terms.
- It is staff research. The views are the authors' and do not represent an official position of the Federal Reserve.
The assumption most worth interrogating is the four-year weighted average life, since it is doing quiet work. Shorten it and the same beta move produces a smaller duration effect; lengthen it and the figure grows.
The Supervisory Question Just Moved
Deposit token projects have been assessed on settlement safety, operational resilience and legal finality. This paper adds funding stability to that list, which is a different kind of question: it is about what the product does to the sponsoring bank rather than what it does to a payment.
For the consortium projects now forming, the awkwardness is structural. A deposit token network is most useful when balances move freely between member banks, and that is precisely the property this analysis identifies as corrosive to funding stability. Interoperability and stickiness are in direct tension, and the tension does not have a technical resolution.
Community bank networks face the sharper version, since smaller institutions rely more heavily on stable retail deposits and have fewer wholesale alternatives when funding costs rise. That asymmetry is worth weighing against the pooling logic set out in whether 3,283 small banks can build what four big ones are building.
For the compliance architecture beneath tokenized money, see smart contract compliance and RWA Layer-0 design, and for the structural overview our institutional guide to RWA tokenization.
Frequently Asked Questions
What did the Dallas Fed actually publish?
A research article dated 25 August 2026 modelling how tokenized and programmable deposits could affect bank liquidity and maturity transformation. Its headline result is that a 10-percentage-point increase in deposit rate beta would remove roughly $700 billion of 10-year-equivalent duration risk appetite from US banks, assuming a deposit weighted average life of four years. The views are the authors' own and are not an official Federal Reserve position.
What is deposit rate beta?
Deposit rate beta measures how much a bank must raise deposit rates when market rates rise, which is a proxy for how readily depositors move money in search of yield. A beta of zero means deposits are completely sticky regardless of rates. A beta of one means deposit costs track market rates one-for-one. Bank profitability and the capacity to hold long-dated assets both depend on beta staying well below one.
Why would tokenization change deposit stickiness?
Because stickiness is largely friction. Depositors leave money in low-yielding accounts because switching requires attention, time and paperwork. Programmable deposit tokens plus agentic AI remove that friction: software can monitor rates and move balances without the holder acting. The paper's mechanism is not that tokenization makes depositors greedier, but that it removes the inattention that bank funding models have quietly relied on.
Is there real-world evidence for the mechanism?
Brazil's Pix system is the closest precedent, with roughly 200 million active users and about $650 billion in monthly transactions as of Q1 2026. A 2025 academic study cited in the analysis found that increased Pix usage pushed Brazilian banks toward government bonds and subprime lending while reducing traditional credit intermediation. That is a payments system rather than a deposit token, but the funding-side effect is the one being modelled.
What would prove the concern wrong?
Widespread tokenized deposit adoption accompanied by flat deposit rate betas. The authors state this explicitly: if large-scale adoption occurs and betas do not move, the $700 billion figure evaporates and the concern is overblown. That falsifiability is what separates the paper from most tokenization commentary, and it gives supervisors a specific series to watch rather than a general anxiety.
What does this mean for institutions building tokenized deposit networks?
That the supervisory question is shifting from settlement safety to funding stability. Deposit token projects have been justified on payment efficiency and 24/7 settlement, and those arguments are unaffected. But a regulator assessing a deposit token network will now also ask what it does to the stickiness of the sponsoring bank's funding, which is a question about product design and yield features rather than about the ledger.
Related Articles
Why Can't Two Banks' Deposit Tokens Talk to Each Other?
The network layer these deposits would move across.
Can 3,283 Small Banks Build What Four Big Ones Are Building?
Community banks facing the same funding question.
Did Swift Just Solve Deposit Token Interoperability?
The orchestration layer for cross-bank deposit movement.
What Is RWA Tokenization? A Complete Institutional Guide
The structural context for tokenized money.