AI-managed deposits could push up consumer loan rates

The Federal Reserve Bank of Dallas published an analysis on Aug. 25 that finds AI-managed, instantly transferable deposits could reduce banks’ capacity to absorb interest-rate risk by roughly $700 billion in 10‑year Treasury equivalents. The study used U.S. commercial-bank balance sheets as of July 15.

The Dallas Fed estimated about $7 trillion of asset-side interest-rate exposure expressed in 10‑year equivalents. Of that, roughly $5.84 trillion is supported by the duration characteristics of deposits other than large time deposits. The analysis models how faster and programmable shifts of deposit balances would weaken the stable funding that helps banks hold long‑duration assets.

The authors approximate effective deposit duration as weighted average life multiplied by one minus the deposit beta. Deposit beta measures how quickly banks must raise deposit rates when short-term rates rise. In a sensitivity case that assumes a 10% increase in deposit price sensitivity and a four‑year weighted average life, modeled aggregate duration‑risk appetite falls by about $700 billion in 10‑year equivalents. A separate scenario that cuts weighted average life by 10% reduced modeled maturity‑transformation capacity by roughly $580 billion.

A 10‑year equivalent converts an exposure into the interest-rate risk of a comparable position in 10‑year Treasuries; the analysis notes the ultimate credit impact depends on how banks change asset and funding mixes. Possible responses include issuing more term debt, which would likely raise borrowing costs for households and firms, or holding larger reserves and Treasury holdings to guard against faster outflows, which would reduce capacity for less liquid lending.

The paper points to technological developments that could speed deposit moves: instant settlement, programmable tokenized deposits and agentic AI that can move funds automatically. In June 2026, The Clearing House announced an initiative to develop 24/7 interoperable tokenized commercial‑bank money that includes automated and agentic‑commerce uses.

The Dallas Fed cites international evidence that faster payment rails can alter bank liquidity behavior. A 2025 Central Bank of Brazil paper found heavier use of its instant‑payment system led banks to hold more liquid assets and lower liquidity transformation, though the Dallas Fed notes that example is not a direct comparison to U.S. tokenized deposits.

The authors emphasize tokenized deposits remain early in development and the magnitude of possible effects is uncertain. They add that their views should not be attributed to the Dallas Fed or the Federal Reserve System.