Bank deposit rates are not a background variable in the stablecoin supply curve. They are a parameter inside it. That is the entire content of a BIS working paper that has been circulating through crypto media for the past week, and almost nobody reporting on it has run the second-order math.
I spent three hours last Thursday trying to trace the original document. I could not find the working paper number, the author list, the sample window, the data frequency, or the elasticity coefficient. Every secondary report of this paper — including the one that reached my desk — is a restatement of a restatement. So let me be blunt about the evidentiary basis first, because this is exactly the kind of story that gets amplified into "BIS says stablecoins are dying" while the actual paper might say something far narrower.
What the secondary reporting does claim is consistent across sources: when US bank deposit rates rise, stablecoin supply growth slows, and USDC is the most sensitive of the major stablecoins. Three claims. One mechanism. Let me take them apart.
The Identification Problem Nobody Is Talking About
Here is what bothers me as someone who has spent years staring at supply curves that turn out to be demand curves in disguise. The paper's core claim — deposit rates up, stablecoin growth down — admits at least two completely different causal structures, and they have opposite policy implications.
Structure one: Stablecoins are a deposit substitute. When banks pay more, rational holders move money back into insured accounts. Stablecoins are money.
Structure two: Stablecoins are a risk asset derivative. When the Fed hikes, risk appetite compresses, crypto market cap contracts, and stablecoin minting — which is largely downstream of crypto trading volume and collateral demand — mechanically slows. Stablecoins are an artifact of crypto activity, not an alternative to banking.
Both structures generate the same correlation. Only one of them justifies treating stablecoins as monetary instruments requiring bank-style regulation. The secondary reporting gives me no way to distinguish between them because the identification strategy is not disclosed. I am not going to pretend I know which one the authors found. Anyone who claims to know without reading the paper is guessing.
But I can tell you how to test it yourself, and I did. Pull USDC 30-day supply change from DefiLlama, regress it against the effective fed funds rate, then regress it against total crypto market cap. If the fed funds coefficient collapses to insignificance once you control for market cap, the paper is measuring crypto beta, not deposit substitution. If it survives, you have something real.
The Composition Effect Behind "USDC Is Most Sensitive"
This is the claim that gets repeated most confidently and deserves the most suspicion.
USDC's holder base is disproportionately US-domiciled, institutionally adjacent, and banked. Those holders have a live alternative: a Treasury account, a money market fund, an FDIC-insured deposit paying 4-5%. USDT's holder base is disproportionately in emerging markets, in capital-controlled jurisdictions, in offshore trading desks — places where the local deposit rate is either below dollar inflation or simply not a safe store of value at all.
So of course USDC is more rate-sensitive. The sensitivity difference may reflect holder demographics, not a causal elasticity parameter. This is a textbook composition effect masquerading as a behavioral finding. If the paper does not stratify by holder geography — and the secondary reporting gives no indication that it does — then "USDC has the highest elasticity" is a statement about who holds USDC, dressed up as a statement about how USDC behaves.
I want to be careful here: this does not make the finding wrong. It makes it less portable. A finding that USDC demand is rate-elastic because its holders are American institutions is a finding about a specific distribution of wallets, not a structural law of stablecoins. Over time, as USDC's emerging-market footprint grows — and it is growing, through payment corridors — that elasticity should decline. Watch for it.
The Part That Actually Matters: Circle's Two-Way Rate Exposure
Here is where I want to spend real analytical energy, because this is the layer the secondary reporting completely misses. It is also the layer where actual money is at stake.
Circle's revenue is overwhelmingly reserve interest income. By my reading of the last few quarters of disclosures, the reserve income line runs north of 95% of total revenue. That makes Circle — functionally, actuarially — a levered long position on short-duration dollar rates, packaged inside a payments company.
Now run the two-way exposure.
| Rate direction | Supply volume | Unit reserve yield | Circle total revenue | |---|---|---|---| | Rates up | Growth slows (paper's finding) | Rises | Indeterminate | | Rates down | Growth accelerates (implied) | Falls | Indeterminate |
Volume and price move inversely with respect to rates. The market values Circle and the broader stablecoin complex almost entirely on supply growth — "stablecoins will reach trillions" — which is a volume-only heuristic. That heuristic systematically misvalues the business in both directions. In a cutting cycle, supply looks great and earnings quality gets worse. In a hiking cycle, supply growth looks anaemic and unit profitability is exceptional.
In 2020 I ran an automated yield allocator across Aave and Compound with $150,000 of my own capital. The script hit 45% APR gross and roughly 28% net after gas. What killed it was not the strategy — it was the assumption that yield was a stable parameter. It was a moving target defined by exogenous rate policy. I rebuilt the model around rate regimes and the drawdown profile changed completely. Circle's business has the same structure, one layer up.
Today I would value a stablecoin issuer on revenue per unit of reserves outstanding, decomposed by rate scenario, not on supply. If you are building a position or a model around this sector, that is the change the BIS paper should force on you.
The Institutional Detail That Turns a Rate Channel Into a Wall
There is a legal fact that almost no one connects to this paper, and it is the reason the headline "banks have leverage" is not a metaphor.
US stablecoin legislation prohibits paying yield to stablecoin holders. Circle cannot share reserve income with USDC holders. Read that again with the paper's finding in mind.
If deposit rates rise and Circle cannot respond by passing through higher yield, the competitive gap opens without any offsetting mechanism. The bank does not have to innovate. The bank does not have to build a better product. The prohibition against yield-bearing payment stablecoins functions as a legally enforced pricing shield for the deposit franchise. The "bank leverage" the researchers measured empirically is reinforced by statute.
This is the single most underappreciated structural fact about the stablecoin market, and it cuts both ways. It means the current equilibrium is politically maintained, not technically determined. The moment that prohibition loosens — even narrowly, to large-balance institutional holders — the elasticity the paper measured changes shape. USDC starts competing directly with money market funds, and the rate channel weakens appreciably. If I were building a regulatory tracking sheet, that would be row one.
The Contrarian Read: The Real Threat Is Upstream, Not Downstream
Every analysis I have seen treats the bank-versus-stablecoin competition as a market-share contest in payments. That framing is wrong, and it is wrong in a way that matters for how you allocate.
The stablecoin issuer sits downstream of the banking system in three mechanical ways: reserve custody at commercial banks, settlement through bank rails, and distribution through exchange partners who take a revenue split. The 2023 Silicon Valley Bank episode demonstrated what happens when the middle link breaks — USDC traded at roughly $0.87 for a weekend because a portion of reserves sat at a bank that failed. That was not a design flaw in the token contract. It was the token wrapping a bank and inheriting the bank's fragility.
But the deeper threat is not rate competition. It is tokenized deposits. A bank-issued deposit token is a bank liability, carries deposit insurance, can pay interest, and operates inside the existing regulatory perimeter. It needs no new legislation to exist. If commercial banks commercialize deposit tokens at scale, the stablecoin's competitive position is not squeezed by higher rates — it is displaced from the layer beneath it. The float, the settlement volume, the programmable payment rail all migrate to an instrument that has every advantage a stablecoin has, plus yield, plus insurance, plus a regulator that already likes it.
That is the terminal state if you follow the logic of the BIS paper to its end. Stablecoin issuers spend the next decade fighting for a shrinking regulatory niche while banks quietly ship the same product with a government backstop. USDC's highest compliance posture and its highest rate sensitivity are the same property. Proximity to the banking system is a moat and a ceiling simultaneously.
What This Actually Means for Positioning
The paper is not a trade. Anyone telling you to act on a working paper with an undisclosed methodology is selling you something.
But it is a legitimate model correction, and there is one place it pays off immediately: DeFi liquidity forecasting. USDC is the primary collateral and quote asset across lending markets and automated market makers. If stablecoin supply growth is partly a rate function rather than purely a penetration function, then DeFi total value locked inherits an exposure to Federal Reserve policy that the market does not price. During the 2020-2021 expansion, every decline in DeFi TVL was narrated as a crypto-native event. Some fraction of those declines was monetary policy working through the collateral layer six weeks earlier.
Conversely, in a sustained cutting cycle, stablecoin supply growth could run well ahead of consensus — and that is an underpriced positive for on-chain activity, AMM depth, and lending capacity.
What I want you to do this week is not buy or sell anything. It is to rewrite one model. Take your stablecoin supply forecast and split it into a structural adoption term and a rate-regime term, then run the forecast under three rate paths. If your answer does not move across those paths, your model is measuring the wrong thing.
The paper will not tell you which direction rates go. It will not tell you what Circle earns next quarter. What it does tell you — and this is worth the entire exercise — is that the supply curve you have been extrapolating linearly has a central bank sitting inside it.
Volume screams about adoption. Liquidity whispers about the discount rate. The bank was never watching from the sideline. It was holding the pencil the whole time.
