Dallas Fed maps the balance-sheet tradeoffs behind tokenized deposits

The Federal Reserve Bank of Dallas has put a sharper analytical frame around one of the next questions in digital money: what happens to ordinary bank funding if deposits themselves become tokenized and transferable with near-instant finality. In a new research note, Dallas Fed economists Rosie Levy and Srini Ramaswamy argue that tokenized deposits could do more than modernize settlement. If adoption becomes meaningful, they say, the same features that make tokenized balances attractive to customers could also make bank funding less sticky, force lenders to carry more liquid assets and reduce the amount of long-dated credit the banking system is comfortable extending.

That matters because deposits are not just a payments product. They are also the raw material banks use to fund mortgages, commercial credit and other longer-duration assets. The Dallas Fed note draws a distinction between tokenized deposits and stablecoins that is increasingly relevant for the RWA stack. Stablecoins generally sit outside the traditional banking perimeter even when they are backed by cash or Treasuries. Tokenized deposits, by contrast, remain bank liabilities inside the existing regulatory structure and can pay interest. But if those liabilities become programmable and easier to move across institutions, the economic behavior of deposits can start to change in ways that ripple through bank balance sheets.

The authors focus on two variables that make deposits valuable funding instruments today: weighted average life, which measures how long balances tend to stay on a bank’s books, and deposit beta, which captures how quickly banks have to pass market-rate changes through to customers. In plain English, sticky deposits let banks fund longer-term loans without repricing liabilities every time short-term rates move. Using Federal Reserve H.8 data, the Dallas Fed estimates that commercial banks collectively carry about $7 trillion of duration exposure in 10-year Treasury equivalents, and that roughly $5.8 trillion of that — around 80% — is effectively supported by deposit behavior. In their framework, a 10% decline in deposit weighted average life would trim aggregate maturity-transformation capacity by about $580 billion, while a 10% increase in deposit price sensitivity would reduce duration-risk appetite by roughly $700 billion.

The transmission channel is straightforward. Once frictions around moving money fall, corporate treasury teams and eventually software agents can optimize deposit placement more aggressively. Banks would then have a harder time treating operating balances as slow-moving funding. The Dallas Fed also argues that tokenization could blur the line between operational and non-operational deposits, pushing institutions to assume faster outflows in stress and to hold bigger buffers of reserves and Treasuries. That does not mean tokenized deposits are inherently destabilizing, but it does mean adoption at scale is likely to show up first in asset-liability management, liquidity planning and loan pricing rather than only in flashy payments demos.

A useful corroborating reference comes from Brazil, where the Central Bank’s Pix system offers a real-world example of what instant, always-on bank transfers can do to balance-sheet behavior even without deposit tokens. The Dallas Fed note cites Pix as an analogue, and a 2025 working paper from the Central Bank of Brazil, Columbia Business School and Wharton finds that heavier Pix usage increased banks’ demand for liquid assets, especially government bonds, while reducing credit intermediation. By the first quarter of 2026, the Dallas Fed said Pix had around 200 million active users and processed roughly $650 billion in monthly transactions, equivalent to about a quarter of Brazil’s annual GDP. The Brazilian evidence does not prove tokenized deposits will produce the same outcome in the U.S., but it gives policymakers and bank treasury teams a concrete data set instead of a purely theoretical warning.

The timing is notable for the onchain finance market because tokenized deposits are moving from concept work toward implementation design. Banks have been experimenting with consortium and interoperable models precisely because single-bank deposit tokens are less useful if they cannot circulate beyond the issuing institution. That puts tokenized deposits on a collision course with several adjacent RWA categories at once: stablecoins used for settlement, tokenized Treasury products used as liquidity instruments and permissioned payment networks being built for institutional flows. The more these products converge, the more important it becomes to understand not just user demand, but how each format changes the funding mix sitting underneath the system.

For RWA builders, the Dallas Fed paper is a reminder that better settlement rails do not arrive as a free efficiency gain. They can reshape the economics of the liabilities that support credit creation. For banks, that means tokenized deposit strategy cannot be owned only by innovation teams; treasury, risk and regulatory functions need to be in the room from day one. For policymakers, it suggests future debates about tokenized money will have to look beyond consumer utility and ask how programmable, instantly mobile bank liabilities interact with monetary transmission, lender-of-last-resort mechanics and the supply of safe liquid assets. The headline is not that tokenized deposits are a bad idea. It is that if they scale, they will likely change the structure of banking before they change the branding of payments.