Dallas Fed study says tokenized deposits could force banks toward more liquid balance sheets

A new Dallas Fed economics note is putting harder numbers around a question that has hovered over tokenized cash for months: what happens to bank balance sheets if deposits become programmable, portable and effectively always on? The paper does not argue that tokenized deposits are inherently undesirable, and it stops well short of predicting an imminent market shift. But it does make a clear point. If deposits start moving with the speed and flexibility that distributed settlement rails could enable, banks may need to hold more liquid assets, take less duration risk and rethink the economics of lending funded by what used to be relatively sticky deposits.

The study arrives as policymakers and markets spend far more time debating stablecoins than tokenized deposits. Dallas Fed researchers Rosie Levy and Srini Ramaswamy frame the distinction carefully. Stablecoins typically sit outside the traditional deposit perimeter and are often backed by cash and Treasuries, while tokenized deposits remain claims inside the banking system and can continue to pay interest. That makes them attractive as a potential bridge between conventional banking and blockchain-based settlement. But the paper argues that if tokenized deposits circulate beyond a single bank’s own walls, they could also weaken the frictions that have historically made many deposits predictable funding sources rather than instantly contestable balances.

That funding behavior matters because banks do not simply treat deposits as overnight money even when they are legally withdrawable on demand. In practice, banks model deposits using expected life and price sensitivity, then use that funding profile to support longer-dated lending and securities exposure. The Dallas Fed note estimates that commercial banks collectively rely on deposits for roughly 80% of their aggregate duration risk capacity, or about $5.8 trillion of the $7 trillion in ten-year-equivalent exposure it calculates across bank balance sheets. On the paper’s assumptions, a 10% reduction in deposit weighted-average life would shrink maturity-transformation capacity by about $580 billion, while a 10% increase in deposit rate beta would cut duration-risk appetite by about $700 billion.

The mechanism is straightforward even if the balance-sheet math is not. If tokenized deposits can move between banks in real time, and if smart-contract logic or agentic software makes switching automatic, rate-sensitive depositors may become much harder to retain at low cost. The authors also warn that tokenized money market funds could become more direct competitors as settlement frictions fall. In that world, banks may preserve asset composition only by paying more for deposits or by leaning more heavily on wholesale term funding, which would push bank intermediation economics closer to those of nonbank lenders and likely raise the cost of credit for households and businesses.

The liquidity side of the paper is just as important for RWA markets. The researchers argue that real-time tokenized transfers could increase the volatility and uncertainty of deposit outflows even if total deposit balances stay large. That, in turn, would encourage banks to hold bigger buffers of high-quality liquid assets and place a premium on collateral that can be monetized quickly during stress. The paper uses Brazil’s Pix system as a real-world comparison point for always-available payments, citing research that heavier usage increased banks’ demand for liquid assets and reduced credit intermediation. Tokenized deposits are not identical to instant-payment rails, but the message is that faster money changes asset-liability management before it changes headlines.

That is why the Dallas Fed paper lands in the middle of a broader institutional debate rather than in isolation. The Federal Reserve’s FedNow framework has already normalized the idea that domestic payments can move instantly. A recent New York Fed staff report examined the policy tradeoffs between stablecoins and tokenized deposits in blockchain-based finance and concluded that the optimal answer can depend on regulatory costs, risk-shifting incentives and whether policymakers want to preserve bank credit creation. Meanwhile, the Bank for International Settlements’ Project Agorá is explicitly exploring a shared platform that would record central bank reserves and commercial bank deposits in tokenized form to support atomic wholesale settlement. The policy stack is moving from theory toward implementation.

For builders in RWA, the takeaway is not that tokenized deposits are a dead end. It is that tokenized cash should not be analyzed only as a product feature or settlement convenience. Once onchain money becomes easier to reprice, reroute and automate, it starts to alter the funding assumptions underneath the institutions expected to use it. That makes balance-sheet design, liquidity treatment and interoperability at least as important as wallet UX or transaction speed. The Dallas Fed paper is best read as an early warning that the next phase of tokenization will be constrained not just by issuance and regulation, but by how banks absorb the consequences of making deposits behave more like programmable market instruments.