A stablecoin at $0.94 can be minutes from recovery or days from zero. The difference is which of four mechanisms is doing the breaking.
Nadia OkoroPolicy & Consumer Editor · Regulation, tax, stablecoins and the products retail users hold
Stablecoins lose their peg for four distinct reasons: doubt about the reserves, a liquidity crunch where redemption cannot clear fast enough, failure of the stabilising mechanism itself, and friction that blocks arbitrage. The first two usually recover if the backing is real — USDC returned to par within days of the Silicon Valley Bank weekend. The third rarely does, as Terra showed.
A stablecoin trading below a dollar is a signal, not a verdict. In March 2023 USDC fell to about $0.87 and was back at par within days. In May 2022 UST fell through the same level and never returned. Both were called depegs. They had almost nothing in common.
Four distinct mechanisms produce a broken peg, and identifying which one is running tells you whether to wait, exit, or never touch the asset again.
The token is fully backed, but the market cannot verify the backing right now, so it prices in the possibility that it is not. This is what happened to USDC: Circle held $3.3bn of reserves at Silicon Valley Bank when the bank failed on a Friday, the banking system was closed for the weekend, and nobody could confirm the deposit would be recovered. The price fell, the deposit was made whole on Monday, and the peg returned.
Reserve-doubt depegs recover when the doubt resolves and the backing turns out to be real. They do not recover when it is not. The distinguishing question during the event is whether the issuer publishes verifiable reserve composition — which is why reserve quality carries 30% of the weight in our stablecoin rubric, more than any other criterion in any category we rate.
The backing is fine and the issuer is solvent, but everyone wants out at once and redemption is slower than the secondary market. Holders who need immediate exit sell into thin order books, and the discount is the price of immediacy rather than a judgement about solvency.
This is the most common depeg among reserved stablecoins, and it is usually shallow and brief. The tell is the redemption channel: if the issuer is processing redemptions at par for verified counterparties, the discount is a liquidity phenomenon and arbitrage closes it. If redemption is suspended, you are in mechanism four.
Algorithmic and partially collateralised designs stabilise the price with a feedback loop — mint and burn against a volatile sister token, adjust supply against demand, or hedge with derivatives. When the loop runs backwards it accelerates rather than corrects.
Terra is the canonical case: UST was redeemable for a dollar of newly minted LUNA, so falling confidence minted more LUNA, which drove LUNA's price down, which required minting more, and the reflexivity ran to zero in three days. No amount of waiting recovers a mechanism failure, because the mechanism is what is producing the fall.
Modern designs are more careful, and some are genuinely different. Ethena's USDe holds a delta-hedged position rather than an algorithmic loop, and it is explicit that sustained negative funding is the scenario that erodes it. That is a mechanism risk with a named trigger, which is a far better position than an unnamed one — but it is still mechanism risk, not reserve risk.
A peg holds because someone can profitably buy the discounted token and redeem it at par. Remove that path and nothing pulls the price back. Redemption suspended, minimum redemption sizes above what most holders hold, KYC gates that take weeks, or a chain where the redemption contract is not deployed — each breaks arbitrage while leaving the reserves untouched.
This is why redemption access is a scored criterion rather than a footnote. Tether restricts direct redemption to verified counterparties above a six-figure minimum, so an ordinary holder exits through the market and depends on someone larger doing the arbitrage. Liquity's LUSD sits at the other extreme: anyone can redeem for ETH at face value at any time, which is precisely why its peg is enforced mechanically rather than by trust.
Check the issuer's reserve page first and note the date. Then check whether redemptions are open, and at what minimum. Then look at where the discount is deepest: if one chain or one venue is far below the others, it is a local liquidity problem rather than a systemic one. Finally, check whether the token is collateral in lending markets — if it is, forced liquidations can widen the discount well beyond what the fundamentals justify, which is what turned the April 2024 ezETH dislocation from a discount into a cascade.
The practical rule that follows: size a stablecoin position by the mechanism you are exposed to, not by the yield. A reserve-backed token with open redemption and published composition can survive a bad weekend. A mechanism-dependent token pays you extra precisely because it might not.
An accessible breakdown of how fiat-backed, crypto-backed and algorithmic stablecoins each try to hold their peg — and the real failures that show what happens when they don't.
A Merkle-tree snapshot is evidence, not a solvency statement. Here is exactly what an exchange proves when it publishes reserves, and the three gaps every programme leaves open.
No order book, no market makers standing by — just a formula and a pool of reserves. Here's the mechanics behind the constant-product curve that prices most of DeFi.