There is a number in BIS Working Paper #1382 that should not exist. The authors report that a one-basis-point increase in bank deposit rates reduces ten-week stablecoin growth by roughly three percentage points. The instrumental-variable estimate is three times larger. Nine percentage points. Per basis point.
Run that arithmetic forward. The Federal Reserve raised rates by 525 basis points between March 2022 and mid-2023. By the paper's own coefficient, stablecoin growth should have fallen by more than fifteen hundred percentage points. Instead, aggregate stablecoin market capitalization fell by roughly thirty percent over the same window.
Both statements cannot be true. Either the units are wrong, or the economics are. The paper's most quoted number is also its least defensible. That does not make the paper worthless. It makes the paper interesting for a reason its authors did not intend.
The stablecoin industry spent the past two years describing a supply-side story. Visa Direct moved into production. Mastercard's Global Money Movement reached 130 countries. Payment volume was cited at more than $400 billion across the first eight months of the year, though the figure arrived without a stated methodology and with a timeline that does not reconcile against the GENIUS Act's July 2025 signing date.
The distribution layer β the last mile between a tokenized dollar and a merchant's acquirer β was being built out with genuine engineering effort. And the architecture underneath it had not moved. A fiat-backed stablecoin in 2025 is the same instrument it was in 2014: a centralized IOU, redeemable one-for-one, with the trust anchor sitting in a bank vault and a legal wrapper rather than in a verifiable on-chain state.
Then GENIUS Act Section 4(a)(11) arrived, and it did something the industry has still not priced correctly. It banned issuers from paying interest on stablecoin balances. The effective date is January 18, 2027. The ban applies to the issuer. It does not apply to third-party platforms, which may continue to offer rewards.
I spent much of 2020 pulling apart Compound's interest rate model, and the lesson I took from that exercise applies here with uncomfortable precision. A yield curve is not a market fact. It is a governance decision wearing the costume of a market fact. The reserve interest on a stablecoin is the same thing. It is the return on the Treasury bills sitting behind the token, and the only live question has always been who gets to keep it.
Before the ban, the answer was negotiable. After January 2027, it is statutory. The issuer keeps it. That is not a technical change. It is a redistribution of cash flows, written into law, dressed as consumer protection.
Here is what the BIS paper actually establishes, once the arithmetic is set aside. The direction is sound and it is not obvious. Stablecoin demand is not purely a function of network utility. It responds to the opportunity cost of holding a non-yielding dollar claim. When a bank deposit pays 5 percent and a stablecoin pays zero, the stablecoin has to justify itself on something other than carry. When the deposit pays zero, it does not.
The paper's sensitivity estimates are where the value sits, even if the magnitudes are suspect. USDC shows a coefficient of roughly 4.8, USDT roughly 1.9. USDC's compliance advantage and its rate fragility are the same property, measured twice. The deeper a stablecoin penetrates US institutional balance sheets, the more its holders resemble depositors, and the more they leave when the deposit rate moves. Tether's user base sits in offshore and emerging markets where the reference rate is not a US deposit but the local currency's inflation. Those holders are buying dollar access, not dollar yield. They are less responsive because they have nowhere else to go.
The structural problem is unnamed. An issuer's revenue is essentially 100 percent reserve interest. Circle and Tether do not earn fees on transfers. They earn the spread on the float. This means the business is exposed to a pincer that no amount of volume growth resolves. In a high-rate regime, the reserve yield is attractive but bank deposits compete for the same balances and suppress growth. In a low-rate regime, growth accelerates but the yield per dollar of float collapses.
The issuer cannot simultaneously maximize scale and unit economics. The two objectives are driven by the same exogenous variable, in opposite directions. That is a structurally hedged business with no hedge.
And the GENIUS Act makes it worse by removing the one tool that could have offset it. Before the ban, an issuer could pass through part of the reserve yield and compete on carry. After January 2027, it cannot. The distribution layer β exchanges, wallets, card networks β retains that option and the issuer does not. Value capture migrates up the stack, away from the balance sheet that carries the credit risk and toward the interface that owns the customer.
The paper has two blind spots that matter more than its data problems.
The first is tokenized deposits. A bank-issued deposit token carries bank credit, deposit insurance eligibility, and β critically β the legal capacity to pay interest. It is the stablecoin's most direct competitor, and the paper does not mention it once. If a major bank ships a yield-bearing deposit token with the same 24/7 settlement properties, the stablecoin's entire competitive position narrows to the unbanked and the offshore. That is a demotion from infrastructure to fringe.
The second is the rate cycle. The sample runs 2022 through 2025 β the most aggressive tightening regime in four decades. The paper measures stablecoin demand against a reference rate that was rising almost continuously. It never asks what happens when the reference rate falls. The answer is that the sticky gravity weakens automatically, and the paper's central finding partially reverses without any change in stablecoin technology.
There is also an institutional point worth stating plainly. The BIS is owned by central banks. Its research on whether banks can defend their deposit franchise is not disinterested. Vested interest distorts the lens of analysis β not through fabrication, but through the questions that never get asked. The tokenized deposit question is exactly the one a central bank research department would prefer to leave for later.
Strip the paper to what survives scrutiny and one claim remains standing: stablecoin growth is a demand-side problem, not a supply-side one. The channels are built. The acceptance is there. What limits adoption is that a non-yielding dollar has to beat a yielding one on something other than price.
January 18, 2027 is the date to watch. Between now and then, the OCC and Treasury will define what counts as a disguised yield. Every rewards program and points scheme offered by a third-party platform sits inside that gray zone. Certainty is a bug in a stochastic world, and the regulator's pen is the only instrument that will resolve it.
The protocol does not lie. The interface does. Right now the interface is a statute, and it has not finished being written.