Macro · Research note
The Shadow Monetary Policy of Stablecoins: Threats and the Vulnerability of Multi-Issuance Architectures
The stablecoin market is roughly $313 billion, not the several trillion usually quoted. Getting the scale right changes the diagnosis: the danger is structural, not size.
Almost every warning about stablecoins begins by describing a multi-trillion-dollar parallel financial system. That figure is wrong, and the error is not small: total stablecoin supply stood at roughly $313 billion in mid-2026. The trillions belong to two other things, namely forecasts of where the market may be in five years, and the market capitalisation of crypto-assets as a whole.
The correction matters because it changes the diagnosis: at $313 billion, stablecoins are not large enough to threaten the banking system by sheer weight. They are large enough to do so through structure, and this piece argues that the structural risk sits in two places: the concentration of supply in two issuers, and the legal design of stablecoins issued across multiple jurisdictions against ring-fenced reserves.
Getting the number right
Figure 1 sets out the distribution. Tether holds roughly $184.7 billion of supply and Circle roughly $73.8 billion, together about 83% of the market. Approximately 99% of all stablecoin supply is denominated in dollars. Ethereum holds around $170 billion of it, close to 60% of global supply on a single chain.
Note what this concentration means: a system of $313 billion spread across fifty issuers would be a curiosity, whereas the same total held 83% by two entities, 99% in one currency and 60% on one chain is a system where a single failure becomes a system-wide event. Concentration, not aggregate size, is what makes the market systemically interesting.
Figure 2 places the market against the assets that back it. Stablecoin reserves are invested largely in short-dated Treasury bills, and the stablecoin float is a modest fraction of bills outstanding. Hence claims that stablecoin issuers now dominate the front end of the Treasury curve overstate the position considerably, even though issuers collectively hold more Treasuries than several sovereign nations.
The transaction figures deserve the same scepticism. Headline volumes of $18.3 trillion for USDC and $13.3 trillion for Tether in 2025 are raw on-chain totals: they include automated market-making, arbitrage loops, bridging and internal transfers, and they are not comparable to card-network volume despite frequently being compared to it. The direction of travel toward real payments use is genuine; the magnitudes routinely quoted for it are not.
The transmission threat, and the evidence for it
The conventional monetary transmission mechanism runs through commercial banks: a central bank moves its policy rate, banks adjust deposit and lending rates, and aggregate demand responds. Stablecoins interfere with this at the deposit level.
When a household converts a bank deposit into a stablecoin, total system deposits need not fall; what changes is composition. A granular, insured, sticky retail deposit becomes a large, uninsured, wholesale deposit held by one corporate issuer at a partner bank. Under Basel III liquidity coverage rules, wholesale deposits from financial institutions carry substantially higher assumed run-off rates than retail deposits. The partner bank must therefore hold more high-quality liquid assets against the same nominal balance, and those assets cannot fund maturity transformation.
This was theory until early 2026, when Michael Junho Lee and Donny Tou at the Federal Reserve Bank of New York tested it directly in Staff Report 1185, linking on-chain stablecoin flows to wholesale interbank payment data.
The results in Figure 3 are specific. Partner banks servicing stablecoin issuers saw interbank payments rise roughly 67%, intraday reserve volatility increase by 38% to 55%, and reserves-to-assets ratios rise about 147% relative to peers. Most consequentially, their loans-to-assets share fell by 14 percentage points after the partnership began.
| Traditional deposit funding | Stablecoin-mediated funding | |
|---|---|---|
| Deposit character | Granular, insured, retail, sticky | Concentrated, uninsured, wholesale |
| Settlement | T+1 or T+2, batched, predictable | Atomic, 24/7, continuous |
| Liquidity buffer required | Standard run-off assumptions | Elevated, driven by minting and burning |
| Effect on lending | Deposits fund long-term credit | Loans-to-assets fell 14 pp at partner banks |
| Policy transmission | Rates steer deposit and loan pricing | Liquidity management dominates behaviour |
Source: Lee and Tou, “Stablecoin Disintermediation,” Federal Reserve Bank of New York Staff Report No. 1185, February 2026.
That last row is the finding that matters: the banks in the study were beneficiaries of stablecoin growth, and they still lent less. Therefore disintermediation does not require deposits to leave the banking system; it requires only that they change character on arrival.
The mechanism has an obvious amplifier: because stablecoins settle atomically and continuously, a partner bank cannot net its exposure overnight the way it can with conventional payment flows, so it manages liquidity in real time and holds reserves accordingly.
Multi-issuance: where the structure breaks
The disintermediation problem is chronic and gradual; the acute problem is a legal one.
A cross-country multi-issuance scheme has a single stablecoin jointly issued by several entities in different jurisdictions. Typically one entity is domiciled in the European Union under the Markets in Crypto-Assets regulation, and an affiliated but legally distinct entity sits offshore. To a holder the two tokens are identical: same ticker, same contract, fully fungible, interchangeable on any exchange. To a lawyer they are not identical at all, because each entity holds reserves only for the tokens it issued, in its own jurisdiction and under its own rules.
The economic motive is regulatory arbitrage: the offshore entity can hold higher-yielding or less liquid reserves, while the EU entity provides market access and the appearance of compliance.
Figure 4 locates the failure. The European Systemic Risk Board has warned about the consequence of this split. Under MiCA, an EU issuer must honour redemption at par and cannot charge a redemption fee. An offshore issuer is bound by no such rule and can impose fees, delay redemption or halt it entirely.
Follow that through under stress. Holders of offshore-issued tokens discover that redemption there is expensive or unavailable, and because the tokens are fungible, they route them into the European Union and redeem against the EU entity instead. The EU issuer now faces redemption claims from the global float while holding reserves for only the portion it issued.
Its only defence is cross-border reserve rebalancing, meaning a rapid transfer of assets from the offshore entity. That defence depends on a third-country authority permitting large outflows during a crisis, which is precisely the moment such an authority is most likely to impose capital controls or ring-fence the assets. If the transfer is blocked, EU holders who believed MiCA protected them are left with claims against exhausted reserves.
The 1992 analogy, and where it fails
The structure resembles a fixed exchange rate regime. In 1992 the Bank of England defended sterling’s peg to the Deutsche Mark with finite reserves against a market that could sell without limit; speculators identified the gap between the commitment and the means of honouring it, and the peg broke.
| Black Wednesday, 1992 | A multi-issuance run | |
|---|---|---|
| The commitment | Sterling’s parity with the Deutsche Mark | Redemption at par under MiCA |
| The constraint | Finite foreign currency reserves | Reserves for domestically issued tokens only |
| The failure mode | Reserves exhausted defending the rate | Reserves exhausted meeting global redemptions |
| Speed | Days of institutional trading | Minutes, via automated redemption and lending |
| Spillover | European exchange rate mechanism | Partner banks selling reserve assets |
The analogy is useful for the mechanism and misleading in two respects, both of which cut against the alarmist reading.
First, the Bank of England was defending a price it had publicly committed to hold, whereas a stablecoin issuer is honouring a redemption claim against assets it actually holds. If the reserves are genuinely there and genuinely liquid, redemption is a balance sheet exercise rather than a defence that can be lost. The vulnerability is created by the jurisdictional split, not by the peg itself.
Second, the speed comparison cuts both ways: automated settlement drains reserves faster, and it also allows arbitrageurs to close a discount faster than any 1992 participant could. A depeg that recovers within hours is not the same event as a currency devaluation, though both look identical in the first minutes.
What this analysis cannot tell you
The multi-issuance risk is structural and has not been observed at scale. No cross-country scheme has yet faced a coordinated run, so the sequence described above is a reasoning exercise from legal documents and regulator warnings rather than an account of something that happened.
Furthermore, the Staff Report 1185 findings come from a small population of partner banks over a short window in a market that was growing quickly. A 14 percentage point fall in loans-to-assets is a large effect, and large effects from small samples in fast-moving markets deserve replication before they carry policy weight.
Finally, the $3.5 trillion projection that anchors most stablecoin risk analysis, including the version this piece originally relied on, is a forecast. Building a systemic argument on a number eleven times the observed market is how the scale error described at the outset propagates. The structural case here does not require the forecast to be right, which is the reason for making it structurally.
Conclusion
Stablecoins do not yet threaten monetary policy by size; at roughly $313 billion they are a rounding error against the deposit base of the banking system.
They threaten it through concentration and design. Two issuers hold 83% of supply, and partner banks lend measurably less after taking issuer deposits, which means transmission is already being altered at the current scale. And the multi-issuance architecture puts a par redemption obligation in one jurisdiction and the reserves in another, which is a fragility that no amount of reserve adequacy fixes, because the problem is not whether the assets exist but whether they can cross a border on the worst day.
Supervisors can address the third problem directly by requiring that redemption obligations and reserve assets sit in the same jurisdiction. Whether they do so before rather than after the first run is the open question.
Update: 10 August 2026
Nothing since 16 June has changed the structural argument; one macro assumption has moved.
The energy shock that frames the opening section has re-escalated. Iranian attacks on shipping resumed in July, the United States reimposed its naval blockade on 12 July, and Brent gained more than 9% in a single session before trading in a wide band into August. The Federal Reserve held its target range at 3.50 to 3.75% on 29 July over a divided committee, with balance sheet reduction still the preferred instrument. Quantitative tightening drains reserves from the same banking system that Staff Report 1185 shows is already holding elevated reserves against stablecoin deposits. Those two pressures work in the same direction and neither is calibrated against the other.
The scale correction at the top of this piece has held. Supply has continued to grow through the summer without approaching the trillions routinely quoted, which supports the argument that the risk is structural rather than a function of size.