Every Stablecoin Law Draws a Perimeter. None of Them Defines a Stablecoin.

By the start of 2026, most of the world's largest financial jurisdictions had a stablecoin rulebook. What none of them had was a definition of a stablecoin.

That sounds like a technicality. It isn't. Each regime defines only the subset it intends to supervise: payment stablecoins under the US GENIUS Act, e-money tokens and asset-referenced tokens under MiCA, specified stablecoins under Hong Kong's ordinance, single-currency stablecoins under the Monetary Authority of Singapore's proposal. Those are perimeters, not categories. A token that falls outside one of them does not stop existing, does not stop clearing, and does not stop showing up on a customer's deposit instruction. It simply stops being counted.

The Wharton Blockchain and Digital Asset Project's Stablecoin Toolkit, published in January 2026 under lead author Kevin Werbach with a global expert working group, is an attempt to fix the counting problem before the supervisory one. Its stated purpose is to describe the market as it actually is rather than as legislation imagines it, and it starts by building a definition from first principles instead of borrowing one from a statute. We've focused here on the parts most relevant to institutional adoption: the definition itself, the stabilisation mechanisms, the economics of yield, and the ecosystem roles that sit between an issuer and a user.

Why the perimeter question got sharp this year

Hong Kong makes the point with unusual clarity, because the whole sequence has now played out inside eighteen months.

The report cites, as evidence that banks were arriving, the joint venture formed by Standard Chartered Bank (Hong Kong), Animoca Brands and HKT to issue a Hong Kong dollar stablecoin. On 10 April 2026 the Hong Kong Monetary Authority granted its first two issuer licences under the Stablecoins Ordinance, one to HSBC and one to that joint venture, now Anchorpoint Financial. Reporting on the round put the count at 36 formal applications against two approvals. On 12 August 2026, Anchorpoint opened beta access to HKDAP, the first issuance of a regulated Hong Kong stablecoin, with HashKey Exchange and OSL Group as authorised distributors and access limited to institutions, corporates and professional investors. Retail access is signalled for late in the year, and HSBC's token is expected in the second half of 2026.

So the licensed perimeter in one of the most active jurisdictions currently contains two issuers and, as of this month, one live token available to a screened set of users. Under HKMA rules, every holder of a licensed stablecoin must be identified unless the issuer can satisfy the regulator that other risk mitigants are effective, which is a stricter posture than most jurisdictions take. The licensed set, in other words, isn't simply a smaller version of the market. It's shaped differently.

Outside that perimeter, the market kept growing. DefiLlama-based aggregates put total stablecoin capitalisation at roughly $313 billion in mid-2026, up about 23% year on year. The report itself, written in January 2026, describes total stablecoin assets as exceeding $200 billion, while eMarketer's reading of DeFiLlama data put the January 2026 figure at $308.55 billion. That gap isn't an error in either direction. It's the definitional problem surfacing in the measurement, and the report concedes as much when it notes that aggregators report varying numbers for stablecoin market value and rankings. What the market is worth depends on what you're willing to count.

The third pressure is already scheduled. The GENIUS Act bars digital asset service providers from offering unlicensed stablecoins or stablecoins from foreign issuers starting in mid-2028. At that point the perimeter stops being a licensing question and becomes an access question.

A definition that doesn't start with the licence

The Financial Stability Board's formulation, a cryptoasset that aims to maintain a stable value relative to a specified asset or basket, is accurate and nearly contentless. It restates the word. Wharton proposes instead: a publicly available, non-central bank issued digital asset, aiming to serve as a stable unit of account through economic mechanisms.

Four elements are doing real work there.

Publicly available requires presence on at least one public permissionless blockchain. That draws the line between stablecoins and the private payment coins banks have run on permissioned ledgers for years. It doesn't require open access in practice, since whitelisting and compliance screening sit comfortably inside it.

Non-central bank issued separates stablecoins from CBDCs by liability rather than by technology. Notably it doesn't require private issuance. Wyoming's Frontier Stable Token, launched by a state commission in August 2025, is government-issued and not a central bank liability, and the definition absorbs it without a special case.

Stable unit of account is what excludes gold tokens. A commodity token may be well collateralised and reliably redeemable, but it prices in a volatile unit, which disqualifies it from the function stablecoins exist to serve.

Through economic mechanisms is the element most definitions omit. The peg holds because deviation creates arbitrage, not because anyone declares a value. How that arbitrage is constructed is precisely what separates the categories below.

The consequence is that stability is an aspiration rather than a qualification. Terra's UST was a stablecoin while it was collapsing. That is the opposite of how legislation has to work, and it's why the licensed set will always be a subset of the observed one rather than a description of it.

What survives when you deflate the volume numbers

The report is unusually careful with the figures the industry likes to quote. Visa's onchain analytics put stablecoin transaction volume at $34.7 trillion for the twelve months to June 2025, more than double Visa's own annual payments volume. Visa's adjusted figure, which strips out high frequency trading and bot activity, is $7.3 trillion. Of that adjusted number, the report cites analysis suggesting roughly 99% represents cash-like balances sitting in digital asset trading accounts rather than payment for goods or services.

The payments story therefore runs on about 1% of a number that is itself around a fifth of the headline. Independent estimates land where you would expect: Artemis Analytics put stablecoin payments at a $72 billion annualised run rate in February 2025, and Architect Partners estimated $100 billion to $300 billion annually in June 2025.

None of that makes the payments case weak. It makes it early, and it means the comparison most often drawn is the wrong one. The report's own suggestion is that $35 trillion in largely trading-driven volume belongs next to clearing house settlement figures rather than next to card network payment volumes.

Four mechanisms, and the rulebooks recognise one

Wharton identifies four stabilisation mechanisms in use, along with hybrids: off-chain fully collateralised, programmatic overcollateralised, supply-based algorithmic, and synthetic hedged.

The first is the custodial model covering USDT, USDC, PYUSD and essentially every token any regulator has authorised. The second holds its peg by locking excess onchain collateral and liquidating automatically at a threshold, as DAI and USDS do. The third adjusts circulating supply algorithmically, either through a paired volatile token, the seigniorage design that failed catastrophically with UST, or through Ampleforth's rebase approach that changes balances proportionally in every wallet. The fourth, represented by Ethena's USDe, holds spot crypto collateral against offsetting short perpetual futures so the net position stays flat in dollar terms while earning staking rewards and funding payments. Its issuer markets it as a synthetic dollar rather than a stablecoin, in recognition of a different risk profile. Wharton counts it as one anyway.

Only the first sits comfortably inside the major regimes. MiCA and GENIUS exclude or prohibit most of the rest, and Germany's BaFin barred Ethena's German entity from new USDe business in 2025. The report's point is not that regulators are wrong to draw that line. It's that the excluded designs continue to exist, hold meaningful assets, and remain subject to financial crime rules and money transmission licensing even where prudential regimes don't reach them. Which regimes reach which designs, and where the excluded set can still be distributed, is now a jurisdiction-by-jurisdiction question rather than a general one, which is why we maintain the free Stablecoin Regulation Tracker across 200+ markets.

Yield is the arbitrage the rules created

Every tokenised dollar held in Treasury bills, at a central bank, or in repo earns something. Yield isn't a design choice, it's a property of the reserve. The only question is who captures it, and the report identifies four answers: the issuer, the distributor, the holder, or the collateral provider.

The disclosed numbers show the scale. Tether reportedly earned over $13 billion in profit in 2024, mostly from reserve interest, and passed none of it to holders. Circle earned $1.7 billion in interest income on reserve assets the same year and paid $1 billion of it out as distribution fees to third party platforms. As of early 2025, tokens passing yield directly to holders accounted for about 4.5% of total stablecoin assets.

Both the US and EU regimes prohibit paying interest to holders, but not identically. MiCA bans interest tied to holding period across both token classes and extends the ban to indirect compensation from third parties. GENIUS prohibits issuers from paying interest and is silent on indirect payment. The result is that the distributor channel, where a user earns on balances held with an exchange or wallet rather than from the issuer, is a live competitive surface in one bloc and a constrained one in the other. That divergence will shape which tokens win distribution in which market, independently of how the tokens are built.

The parts of the ecosystem that don't issue anything

The report's most under-discussed section covers everyone who isn't an issuer, and one finding stands out. Citing work by Yiming Ma, Yao Zeng and Anthony Lee Zhang, it reports that USDT had six authorised arbitrageurs in 2024 while USDC had 521. In both cases the top five accounted for the overwhelming majority of activity, 95% for USDT and 85% for USDC.

Direct mint and redemption access, in other words, is concentrated regardless of how many participants exist on paper. Everyone else transacts at secondary market prices. The Hong Kong launch is a live illustration: HKDAP entered the market with two authorised distributors and a model in which those distributors, not the issuer, face the end user. A token four days old already reproduces the structure the report identifies in tokens a decade older.

Add the distributors who control end user access and the yield attached to it, the custodians and treasury managers holding reserves, and the clearing networks forming between issuers and banks, and the picture is that the structural chokepoints sit in access rather than issuance.

What each group takes from this

Banks. There are two distinct entry points and they carry different economics. Issuance is now available under several regimes, but Hong Kong's first round suggests it is being granted narrowly and to institution-backed structures. Distribution and customer access are less constrained, and given how concentrated redemption rights are, the terms on which an institution reaches an issuer matter more than which token it selects.

Regulators. Defining coverage rather than the concept leaves a residual market that doesn't disappear, and that residual currently holds most of the capitalisation given USDT's roughly 60% share. Supervision of it falls to AML and money transmission frameworks by default.

Issuers. The two largest regimes diverge on the specific question of indirect yield, and the disclosed distribution economics show what that channel costs.

Policymakers. The report cautions specifically against sorting the space into risky digital assets and safe stablecoins. Its argument is that stabilisation mechanisms, legal arrangements and business practices produce a spectrum of risk profiles, and that a single threshold applied across all of them will fit some uses badly.

The takeaway: legislating a category doesn't settle the definition

The useful thing Wharton has done is refuse to let the licence write the definition. Once the two are separated, the market splits into a licensed set and a residual set, and the interesting metric stops being total supply. It becomes the share of supply sitting inside a perimeter, which is currently small, and the rate at which that share moves.

The project's second report, published in August 2026, reaches the same conclusion from the legal side. It finds that the major jurisdictions have converged on a common regulated instrument, a fiat-referenced token redeemable at par, fully backed by segregated high-quality liquid reserves and issued by a licensed entity, and states plainly that this regulated core does not describe the entire market. It points to the Financial Stability Board's October 2025 thematic review, which found significant gaps and inconsistencies in how jurisdictions had implemented the agreed standards, with only five having finalised comprehensive stablecoin frameworks as of August 2025. Convergence on the perimeter and coverage of the market are separate achievements, and only one of them has happened.

That framing also explains why measurement disputes keep recurring. A market whose boundary is contested will report different sizes depending on who is counting, which is the same reason the definitional question keeps resurfacing in a market this large, a point we made when the working definition of a stablecoin needed a refresh in 2025. It's also why the concentration of access matters more than the count of tokens, a pattern we traced through the market data earlier this year in a $320 billion market ruled by distribution.

By mid-2028, when the US restriction on offering unlicensed and foreign-issued stablecoins takes effect, the residual set stops being a measurement curiosity and becomes a distribution problem. Anyone whose business touches these instruments has roughly two years to know exactly which side of the line each one sits on, in every market where they operate.

Source: "The Stablecoin Toolkit, Part I: Financial and Market Dimensions," Wharton Blockchain and Digital Asset Project, lead author Kevin Werbach, January 2026. Market capitalisation figures for 2026 attributed to DefiLlama data as compiled in mid-2026 industry reporting and to eMarketer's January 2026 reading of DeFiLlama. Hong Kong licensing details from the Hong Kong Monetary Authority announcement of 10 April 2026 and contemporaneous reporting; HKDAP launch details from Anchorpoint's 12 August 2026 announcement and reporting by Ledger Insights, CoinDesk and The Block. All other figures are drawn from the source report. Methodology note: this analysis focuses on the report's definitional framework, market sizing and ecosystem structure rather than summarising its use case survey, and reorganises the material around the perimeter question rather than following the report's section order.

This article is for informational purposes only and does not constitute financial, investment, or legal advice.