The Federal Reserve Bank of Dallas has produced the first central-bank estimate of what the tokenised deposit build-out costs the banks funding it, and the headline figure is being read backwards. The $700 billion in “Tokenized deposits could affect bank liquidity, maturity transformation”, published August 25, 2026 by senior financial economists Rosie Levy and Srini Ramaswamy, is not lending that vanishes, not deposits that walk, and not additive to the $580 billion beside it. It is 10-year-equivalent duration capacity: the banking system’s appetite to hold interest-rate risk.
The distinction carries the story. Duration capacity is a stock that can be rebuilt with different funding; an outflow is money that has gone. The paper is explicit that banks keep a route to holding the loan book intact — term debt — which puts the cost in the price of credit, not its availability.
What the $700bn actually measures
Levy and Ramaswamy duration-weight the whole commercial-banking balance sheet using Federal Reserve H.8 data as of July 15, 2026. Total assets of $25.7 trillion collapse to $7,029.6 billion of 10-year equivalents once each line is scaled by interest-rate sensitivity: cash and reverse repo contribute nothing, while Treasury and agency securities alone carry $3,020.7 billion, 43 per cent of the system’s duration book from one line.
Solving the liability side for a matched position implies aggregate duration of about 2.8 years on the $16.9 trillion “other deposits” pool, a deposit beta near 0.44 at a five-year weighted average life (WAL). The two sensitivities then fall out separately. A 10 per cent cut in WAL removes 10 per cent of the $5.84 trillion of duration deposits carry — hence $580 billion. The $700 billion uses a different assumption set: a 10-point rise in deposit beta at an assumed four-year WAL strips roughly 0.4 years of effective duration from the same pool. Two single-variable stress tests on one base; running them together double-counts the balance sheet.
The contradiction the banks are funding
The finding that deserved the headline is the contradiction underneath. “About 80 per cent of the duration risk taken by banks ($5.8 trillion 10-year equivalents out of $7 trillion total) in the aggregate is supported by the duration characteristics of deposits,” the authors write. Deposits earn that because they are sticky, and the paper is blunt about why: “Sticky deposits rely in part on the existence of frictions preventing rapid reallocation from one bank to another.”
Those frictions are exactly what deposit tokens are engineered to delete. The institutions paying for the infrastructure are paying to erode their own funding advantage, and the build-out is already in production: HSBC and Standard Chartered netting deposit tokens on Swift’s shared ledger, BlackRock issuing 12 tokenised money market share classes on Kinexys, and MUFG settling a JGB repo with a tokenised cash leg on Canton Network. The authors add an accelerant most balance-sheet models ignore: agentic artificial intelligence plus programmable smart-contract deposits “could theoretically allow this switching to occur without requiring direct action from the deposit holder.”
The escape route is the sting. Banks can hold lending composition steady by leaning on term debt, but “the economics of such lending activity funded by wholesale debt would resemble those of non-bank financial firms and would thus likely adversely impact the cost of credit for consumers and businesses.” Banks do not lose the ability to lend; they lose the subsidy that makes lending cheap.
The Pix evidence, and what it predicts
Brazil is the only large-sample precedent. Pix reached roughly 200 million active users and about $650 billion in monthly transactions by the first quarter of 2026, near a quarter of Brazil’s annual GDP. A 2025 Banco Central do Brasil working paper found heavier Pix usage raised bank demand for government bonds and cut credit intermediation, while banks tilted the surviving loan book toward subprime to defend returns.
Read that against Table 2 and a second-order effect appears: the line already carrying 43 per cent of US bank duration capacity is Treasuries, and the Pix response is to buy more of them. Faster settlement also pushes banks toward larger high-quality liquid asset buffers for liquidity coverage ratio purposes as operational deposits reclassify into higher stressed-outflow buckets. The plausible US outcome is a barbelled banking system rather than a smaller one: more sovereign paper, thinner prime credit, wider spreads on what remains.
Watch whether consortium designs concede the point. Project Agorá, the Bank for International Settlements and Institute of International Finance unified-ledger project, is where portability and stickiness get traded off explicitly. Deposit tokens that cannot move between issuers preserve bank duration and fail as payment instruments; ones that move freely work as payments and reprice credit. The Dallas Fed has priced the second option, and no one has yet designed away the choice.
This article is informational analysis only and is not financial, investment, or trading advice. Cryptocurrencies are highly volatile and can lose substantial value rapidly. Past performance and historical patterns do not guarantee future results. Do your own research and consult a regulated financial adviser before making any investment decision.
Karthik Subramanian
Karthik Subramanian is a founder, writer, and technology consultant with nine years in the crypto ecosystem. He covers token economics, L1/L2 infrastructure, DeFi protocols, wallets/custody, and the bridge between crypto and forex—broker technology, liquidity, and macro drivers. Karthik’s writing focuses on clear, practical frameworks that help professionals evaluate new products and on-chain innovation alongside FX market realities.
