When US banks pay savers more, stablecoins grow more slowly. That is the finding of a working paper published by the Bank for International Settlements on 13 October, which tests whether stablecoins and bank deposits compete for the same dollars. Using weekly US data from January 2019 to June 2025, Iñaki Aldasoro of the BIS and Rashad Ahmed of the Andersen Institute for Finance and Economics estimate that when deposit rates rise by 1 basis point in a week, stablecoin growth over the next ten weeks drops from about 15% to about 12%, and to about 6% under their second method. The paper’s own appendix puts the effect on USDC at 4.8 percentage points and on Tether’s USDT at 1.9.
The paper has been picked up across the crypto press over the past week, sometimes as a threat to stablecoins. It does not say that. It measures a slowdown in growth, not a fall, and its authors write that their data speak to demand for stablecoins, not to bank balance sheets.
What it does show is a lever. In the authors’ words, banks “retain some influence over the pace of that adoption through deposit pricing”. That is why the paper lands in the middle of the fight over stablecoin yield, which runs from the GENIUS Act in Washington to Article 50 of MiCA in Brussels.
Stablecoin bank deposits: what the BIS working paper measured
Working Paper No 1382 asks a narrow question: do people hold fewer stablecoins when bank deposits start paying more? If they do, the two are substitutes in the economist’s sense, because demand for one responds to the return on the other. The authors chose that question because the bigger one, whether stablecoins drain bank funding, cannot yet be answered with the data available. Stablecoins are still small next to the US banking system, and nobody knows precisely where their holders live.
The outcome they track is the market value of reserve-backed stablecoins, taken from CoinMarketCap: USDT, USDC and smaller dollar tokens. That figure grew from under $5 billion in 2019 to more than $220 billion by mid-2025, according to the paper. The rate on the other side is an average of six national US savings and money market deposit rates compiled by RateWatch, for balances from $2,000 to $25,000.
The model holds a long list of other things constant: the federal funds rate, three-month Treasury bills, the ten-year Treasury yield, the bitcoin price, the dollar index and the VIX volatility index. The point is to isolate the part of a deposit rate change that is not simply the Federal Reserve moving all rates at once. The sample has 339 weeks of data. The first, simple estimate is a standard regression (ordinary least squares, or OLS in the paper’s tables): it measures how stablecoin growth moved in the weeks after deposit rates moved, with those other factors held fixed.
One unit needs care. Every result is expressed per basis point of deposit repricing, one hundredth of a percentage point, within a week. That sounds tiny, but deposit rates move slowly: the paper says a 1 basis point weekly change is about 1.2 standard deviations of the usual weekly move. To wipe out the average ten-week growth entirely under the first estimate would take a rise of about 5 basis points in a single week, which the authors call a six-standard-deviation event.
How the economists separated cause from coincidence
Rates and stablecoins could move together for reasons that have nothing to do with savers switching. Banks might raise deposit rates precisely because stablecoins are pulling money away, which would hide the effect the authors are looking for. They say this reverse link biases the simple estimate towards zero, so it probably understates the real effect.
Their fix uses a quirk of US banking, and it is easier to follow than its name. Treasury bill yields follow the federal funds rate almost one for one. Deposit rates do not: below a fed funds rate of about 3%, banks barely pass increases on to savers, and above it they reprice much faster as competition for deposits heats up. That kink, documented in earlier research on the 2022 to 2023 tightening, gives the authors a source of deposit rate changes that comes from banks’ own pricing habits and not from anything happening in crypto. Economists call such a source an instrument. The second method, an instrumental variable (IV) estimate, keeps only the deposit rate moves explained by the kink and checks whether stablecoins reacted to those. It comes out roughly three times larger than the simple one. The authors attribute part of the gap to the bias described above and part to the fact that the kink only bites when rates are high, when savers have the most reason to switch.
Two further checks back the result. The first is a placebo, a test run where no effect should appear: the same instrument does not predict the market value of bitcoin, ether and other non-stablecoin cryptocurrencies, so the effect is not a broad crypto cycle in disguise. The second compares deposit tiers. When banks raise rates on large money market accounts (from $25,000) faster than on ordinary savings accounts (from $2,000), stablecoin growth also slows, by about 0.7 points after ten weeks. That is a small number, but it comes from a different source of variation and points the same way.
There is a weakness, and the paper flags it. Economists check how strongly an instrument predicts the thing it stands in for with a score called the first-stage F-statistic, and treat anything under 10 as a warning sign. This one scores 9.78, “suggesting some instrument weakness”, in the authors’ words. In plain terms, the larger IV figure is less certain than the simple one. CoinDesk, which covered the paper on 17 October, noted the same caveat about the preferred estimate.
USDC reacts 2.5 times more than USDT to bank rates
Split by token, the effect is concentrated in Circle’s USDC. In the paper’s coefficient table, a 1 basis point deposit rate increase is followed ten weeks later by growth 4.8 percentage points lower for USDC and 1.9 points lower for USDT, using the simple estimate. The authors link the gap to who holds each coin: USDC is used more by US institutions and wallet.">decentralised finance (DeFi) users, who can easily compare it with a bank account, while USDT is mostly used offshore and in developing economies, where a US savings rate is not the alternative on offer. USDT still moves in the predicted direction, only more weakly.
Further tests try to break the result. Removing the March 2023 Silicon Valley Bank episode, when USDC briefly lost its peg because Circle held reserves at the failed bank, shrinks the estimates but keeps them negative. Controlling for crypto trading volumes changes almost nothing. Controlling for the yields stablecoins can earn in DeFi lending leaves the simple estimate close to the baseline but lowers the IV figure from 9.1 to 6.9 points. Swapping savings rates for certificate of deposit rates produces the largest effect of all under the second method.
| Specification (effect after 10 weeks, per 1 bp of deposit repricing) | Simple estimate (OLS) | Instrumented estimate (IV) |
|---|---|---|
| Baseline, all reserve-backed stablecoins | -2.8 points | -9.1 points |
| USDC only | -4.8 points | not reported in table |
| USDT only | -1.9 points | not reported in table |
| Excluding the Silicon Valley Bank episode | -2.4 points | -6.5 points |
| Controlling for DeFi stablecoin yields | -2.6 points | -6.9 points |
| With a linear time trend | -2.5 points | -4.0 points |
| Certificate of deposit rates instead of savings rates | -2.2 points | -15.5 points |
Source: BIS Working Paper No 1382, Table 2 and Table OA.A.1. Average ten-week stablecoin growth in the sample is 15%.
The certificate of deposit result matters for a specific group. The authors say it has “particular relevance for community banks”, which rely more on time deposits for funding.
What the paper does not show: no measured drop in deposits or lending
The study is about where savers put new money, not about bank balance sheets. The authors write that they cannot identify which holders switch, or which banks and deposit categories are affected. They add a point often lost in the debate: the dollars behind a stablecoin are reinvested, so a change in stablecoin balances can move deposits between banks and types of depositor rather than change the total.
A Federal Reserve staff analysis from December 2025, cited by CoinDesk, made the same argument: money placed in stablecoins can come back to banks as deposits held by the issuer, which changes who funds the bank even when total deposits stay the same. That sits uneasily with the warning from Bank of America chief executive Brian Moynihan in January that as much as $6 trillion of deposits could move into stablecoins and similar products, as CoinDesk reported it. The BIS paper neither confirms nor refutes his figure.
Other limits are written into the paper. The sample covers a single high-rate episode, 2022 to 2025. The estimates describe a world where stablecoins pay holders nothing; the authors note that yield-bearing stablecoins and tokenised money market funds, still a small segment, could change the picture. They also stress that they measure the effect of rate changes, not rate levels. And the data are American: the euro area is not studied.
Stablecoin growth also slows rather than reverses in the estimates. The paper does recall that stablecoin market value fell by about 30% during the 2022 to 2023 tightening, but it attributes that to deposit repricing building up alongside other shocks, including Fed rate hikes, the Terra collapse and the failure of FTX.
A basis point, and why stablecoin holders earn nothing
A basis point is 0.01 percentage point: a savings rate rising from 4.00% to 4.01% has gone up by one. Stablecoins such as USDC and USDT are backed mostly by cash and short-term Treasury bills, and the interest on those reserves goes to the issuer, not to the person holding the token. So a saver choosing between a dollar stablecoin and a bank account earning a few percent is giving up that return in exchange for being able to move the money on a blockchain at any hour. Our beginner’s guide to stablecoins explains how the reserves and redemptions work.
“Substitutes” in the paper means partial competition for the same smart contract holding reserves of two or more tokens, supplied by users, that a decentralized exchange uses to execute trades automatically at a price set by a formula.">pool of money. When deposits pay more, some of the money that would have gone into stablecoins stays in the bank. Nothing in the study points to holders selling stablecoins in a rush.
Why the stablecoin yield ban is the real battleground
If the return on deposits steers money away from stablecoins, the return on stablecoins could steer it back. The authors spell out two regulatory levers in their conclusion. One is whether issuers may pay holders anything at all. The other is whether issuers must keep part of their reserves as bank deposits, which would support bank funding but, they warn, concentrate large uninsured deposits in a small number of banks.
In the United States the first lever is already pulled. The GENIUS Act, signed in July 2025, bars stablecoin issuers from paying interest or yield to holders, and its rules are being written by the OCC, the FDIC and the Treasury, Decrypt reported on 18 October. The law takes effect on 18 January 2027, as set out on our GENIUS Act effective-date page. The ban covers issuers, not every distributor: exchanges such as Coinbase pay USDC holders rewards funded by their share of Circle’s reserve income, an arrangement described in our piece on Circle’s partner payments.
That gap is what banks want closed. According to Decrypt, banks have lobbied to keep stablecoin yield off the table, and the dispute hung over the Clarity Act, the market structure bill that has stalled in the Senate. The BIS paper gives both sides something. Banks can point to evidence that stablecoins compete with deposits on price. Stablecoin firms can point out that the measured effect is on growth, that USDT, the largest token, reacts far less than USDC, and that the study does not measure deposits or lending at all.
Their conclusion ends on that tension: “How that boundary settles will depend on the returns each form of money is allowed to offer.”
In Europe, MiCA already bans interest on e-money tokens
European rules went further, earlier. Under Article 50 of MiCA, issuers of e-money tokens, the category that covers euro and dollar stablecoins such as EURC and USDC in the EU, may not grant interest. Crypto-asset service providers may not either when they offer services linked to those tokens. Any payment or benefit tied to how long someone holds the token counts as interest, including discounts or net compensation with the same effect, whether it comes from the issuer or a third party. Our MiCA glossary entry covers the wider regulation.
Central banks want the ban widened. In their response to the European Commission’s MiCA review, reported by Euronews on 22 September, the ECB and the national central banks of the ESCB wrote that “the payment of stablecoin remuneration should continue to be prohibited”. They asked for the ban to reach crypto lending, borrowing and staking with similar returns, and indirect rewards such as some loyalty-programme benefits, calling it “a clear legislative priority”. They also want authorities to be able to ban new issuance of a token pegged to a foreign currency, and order existing tokens redeemed, when it threatens financial stability. Their recommendations are not binding.
The European Banking Authority, in its own response published on 24 September, judged the existing MiCA rules for stablecoin issuers broadly appropriate. It asked for crypto lending, including access to DeFi protocols, to be regulated, and for changes to contain what it calls the “significant to very significant risks” of third-country multi-issuer schemes. It also wants a review of reserve requirements, notably the minimum share of reserves held as bank deposits, which is the second lever named in the BIS paper. MiCA’s Article 54 currently requires e-money token issuers to keep at least 30% of the funds they receive in separate accounts at credit institutions.
For holders in the EU, nothing changes this week. The Commission’s consultation closed on 30 September, a month later than planned, and EU diplomats told Euronews they expect a MiCA revision in 2027. USDC’s supervision moves to the EBA on 12 November after its designation as a significant e-money token, as we reported in our article on the EBA taking over USDC.
Both responses came out three weeks before the BIS paper. And the paper tested no euro deposit market, so reading its results into the European debate is an extrapolation.
Dates to watch after the BIS stablecoin study
The Federal Reserve meets on 27 and 28 October. In the paper’s mechanism, what counts for stablecoins is not the policy rate itself but how much of it banks pass on to savers, and the authors find that banks pass on much more once fed funds is above about 3%. On 12 November the EBA becomes USDC’s supervisor in the EU. On 18 January 2027 the GENIUS Act takes effect, with the OCC, FDIC and Treasury rules due before then. No vote date has been set for the Clarity Act.
Coindaro found no BIS press release on the paper, and as of 19 October no public comment on it from Circle or Tether. As with all BIS working papers, the views are the authors’ own and not necessarily those of the BIS or its member central banks.