What is a stablecoin, and what holds the peg

“Stable” describes a target, not a property. A stablecoin is a token whose issuer or protocol intends to hold at a fixed value against a reference asset — almost always the US dollar — and the intention is the least interesting part. What matters is the mechanism holding it there, because that mechanism is what gets tested when the token is under pressure, and the three common designs fail in three completely different ways.

How current stablecoin supply divides between those designs, by peg and across chains, is read from a dated snapshot in where stablecoin supply lives.

Fiat-backed: the peg is a promise to redeem

The largest stablecoins work by holding reserves. The issuer takes dollars, issues tokens, and commits to giving the dollars back. The peg is not maintained by magic or by markets — it is maintained by the credible expectation that a token can be exchanged for a dollar with the issuer.

That expectation does most of the work even for people who will never use it. Retail holders usually cannot redeem directly; redemption is typically available to vetted institutional counterparties, often above a minimum size. What keeps the price at a dollar on an exchange is arbitrage: if the token trades at ninety-nine cents and a large counterparty can redeem at par, buying the discount is free money, and the buying closes the gap.

This has a consequence worth sitting with. The retail price holds because somebody else’s redemption right is credible. If that credibility weakens — because redemptions are paused, or the reserve composition is in doubt — the arbitrage stops, and the price is then held up by nothing but sentiment.

Which is why reserve disclosure is the whole argument for this category, and why it deserves the same scrutiny as any other attestation. What an attestation actually proves is narrower than the marketing around it: a snapshot of assets at a moment, usually without the liabilities that would make it a solvency statement.

Because tokens are issued against deposits and redeemed against them, the supply of a reserve-backed stablecoin follows those flows rather than a schedule, which makes it a different object from a token whose supply is set in advance by caps, emissions and unlocks.

Crypto-collateralised: the peg is an incentive to liquidate

The second design backs the token with crypto rather than cash, and solves the obvious problem — crypto moves — by demanding more collateral than the tokens are worth. Lock up well over a dollar of volatile assets, mint a dollar of stablecoin, and if the collateral falls too far, the position is liquidated automatically and the tokens are bought back.

The peg here is held by liquidation incentives rather than by a redemption desk. It is more transparent, because the collateral is on-chain and anyone can check it. It is also more fragile in exactly one circumstance: a fast, correlated market fall, when liquidations fire simultaneously into a market with no bids. The mechanism that protects the peg in normal conditions is the same mechanism that stresses the market in abnormal ones.

Algorithmic: the peg is confidence, and confidence is reflexive

The third design holds the peg with neither cash nor collateral, but with an arbitrage loop against a second, floating token: when the stablecoin trades below target, holders are given an incentive to destroy it in exchange for the floating asset, reducing supply until the price recovers.

This works while the floating asset has value. It stops working precisely when the peg is under pressure, because the thing being offered in exchange is falling at the same time and for the same reason. Terra’s UST failed this way in May 2022, and the failure was not a bug in the code — the code did what it was written to do. The design assumes confidence in the moment confidence is gone.

Treat “algorithmic” as a description of where the risk sits, not as a technical detail.

What a depeg actually is

The word covers two very different events, and conflating them causes bad decisions.

The first is a secondary-market discount: the token trades below par on an exchange while the redemption mechanism still functions. This is usually a liquidity event, and it usually closes.

The second is a failure of the mechanism itself: redemptions suspended, collateral insufficient, or the incentive loop inverted. The price chart can look similar in the first hours. The outcomes are not similar at all.

The question to ask when a token slips is therefore not “how far has it moved” but “can it still be redeemed, and by whom”. A discount with a working redemption desk behind it is a different object from a discount without one.

What European law now requires

The EU’s Markets in Crypto-Assets regulation does not treat stablecoins as one category. It splits them by what the token references: an e-money token is pegged to a single official currency, and an asset-referenced token references anything else — a basket, a commodity, or a mix.

E-money tokens carry the stricter regime. The issuer must be an authorised credit institution or electronic money institution, and under Article 49 a holder may demand redemption from the issuer at any time and at par value, a right that cannot be waived or made conditional. That converts the redemption promise described above from a commercial practice into a legal obligation — for tokens issued under that regime, which is not all of them.

Why EU rules bar issuers and providers from paying interest on one

This is the provision most people are surprised by, and it is unambiguous. Article 50 prohibits issuers of e-money tokens from granting interest in relation to them. The second paragraph extends the same prohibition to crypto-asset service providers when they provide crypto-asset services related to e-money tokens — so an exchange or custodian offering those services may not pay interest on the token either, even though it did not issue it.

The definition is drawn deliberately wide. Any remuneration or benefit connected to the length of time a token is held counts as interest, including net compensation or discounts with an equivalent effect, whether it comes from the issuer or from a third party. Restructuring a yield as a reward, a rebate or a loyalty benefit does not move it outside the rule while the benefit is related to how long the token is held; the test is that link, not the label.

So when a product does advertise a return on a dollar stablecoin, the honest reading is that you are being offered a different thing than the token: a credit arrangement, a lending market, or an unregulated venue — each with a counterparty who can fail. The yield is not a property of the stablecoin. It is payment for taking somebody’s credit risk.

Where this leaves a holder

A stablecoin is a claim, and the three designs differ in who the claim is against — an issuer, a pool of collateral, or the continued confidence of other holders. Knowing which one you hold tells you what to watch: reserve disclosure and redemption access for the first, collateral ratios and market depth for the second, and for the third, the price of the asset that is supposed to absorb the shock.

Where you keep it is a separate question with its own trade-offs, covered in exchange or self-custody. The token’s design decides what can go wrong with the peg. Custody decides what can go wrong with your access to it.