The two sensitivities in the paper are not the same size, and the gap between them is the argument. A 10% reduction in the weighted average life of deposits takes $580 billion of ten-year equivalents out of the system [3], which is exactly 10% of the $5.8 trillion of duration risk that deposit characteristics currently support [2] [1]: a one-for-one pass-through. A 10% increase in deposit rate beta takes $700 billion [4], about 21% more from a shock of the same nominal size [2], and equal to a tenth of the entire $7 trillion book banks carry [3]. Repricing pressure, not tenor, is the binding term. The two channels also have different triggers. Round-the-clock rails shorten how long money sits still, moving funds between banks in seconds rather than days [6]; AI agents rerouting programmable deposit tokens on a rate signal, the scenario Rosie Levy and Srini Ramaswamy escalate to [5], is a beta story, and beta is the more expensive one.
The paper's own headline rounds against it: $5.8 trillion of $7 trillion is 82.9%, not 80% [4]. And because the two sensitivities are presented separately, nobody has added them. Deposits that are both shorter and twitchier, on the authors' own coefficients, cost $1.28 trillion of duration appetite, roughly 18% of the total [5]. That is arithmetic rather than their scenario, and the linearity surely breaks somewhere, but it sets the scale of what the consortium designers are trading against.
Which makes the build schedule awkward. The paper notes that deposit tokens only matter if they circulate beyond the bank that issued them [7], and that is precisely what is under construction: the Clearing House network behind JPMorgan Chase, Citigroup, Bank of America and Wells Fargo is aimed at 2027 [8], and 39 state bankers associations have formed the BankChain Alliance so smaller lenders get their own route on-chain [10]. Adoption and outflow run on the same feature.
The defensive case for tokenized deposits was always the balance sheet. On McKinsey's estimate, about 15% of every $1,000 a customer converts into a third-party stablecoin returns to banks as wholesale reserves [11], so $850 does not [6], while a tokenized deposit leaves 100% of the money on the issuer's books [12]. That is a real difference, and it is why banking groups spent months holding up the CLARITY Act over stablecoin issuers paying yield [14] before settling on an instrument that sits inside the existing bank rulebook and can pay interest itself [12]. Paying interest is what raises beta.
The 2027 target is the useful part of the timeline. It leaves room to design the friction back in deliberately, as notice terms or as rate tiers, since the gap between initiating a withdrawal and completing it is what maturity transformation currently rests on [15].
