The reasons why stablecoins were unanchored and the factors that really restored them

De-anchoring of stablecoins: Mechanism, Causes and Repair Paths

De-anchoring occurs when the secondary market price of a stablecoin deviates from its US$1 reference price and the arbitrage channel usually used to bridge the spread (i.e. buy at a low price and redeem at face value to the issuer) fails. Whether the exchange rate can return depends entirely on whether the channel can be reopened: in 2023, USDC and DAI will resume within days; in 2022, USDT will resume within weeks; and UST will never resume.

Mechanism: Arbitrage and the situation when it is blocked

According to Chainlink's explanation, stablecoins are roughly divided into three categories: legal currency mortgage type, cryptographic asset mortgage type and algorithm type. Its price stability relies on arbitrageurs continuing to align market prices with the anchored exchange rate. Under normal conditions, this process can be completed in minutes: Eco's guidance states that USDT often drifts to $0.997 or $1.003 during periods of volatility and then rebounds quickly;Chainlink also describes $0.998 or $1.002 in a separate article as a temporary imbalance that can repair itself. In contrast, unanchoring is what Eco calls a "persistent deviation" that traders cannot eliminate through arbitrage because a structural mechanism has broken down.

The pattern of this structural rupture is usually consistent: primary market exit channels (redemption from issuers, or equivalent operations on the chain) become slow, restricted, expensive, or completely unavailable, while secondary market sell-off continues. CryptDaily summarizes it as "half mechanism, half psychology"-mechanisms must provide arbitrageurs with low-risk ways to buy discounted assets, otherwise the discount will persist regardless of whether the underlying collateral is sound or not.

Fiat collateralized stablecoins: Redemption is a relief valve

For fiat collateralized stablecoins, Chainlink explained: If the trading price of the token is less than US$1, institutional holders can buy it at a low price and redeem it from the issuer for US$1, thereby earning a difference-it is this buying pressure that drives the price recovery. This is only effective if redemption channels are truly open and efficient. The USDT case in May 2022 demonstrated how the mechanism works slowly rather than completely: Eco reported that during the Terra/Three Arrows capital crisis, the token fell to approximately $0.95 on May 12, 2022;Tether processed more than $10 billion in redemptions in two weeks; and discounts persisted during this queue and resumed in about two weeks thereafter.

The USDC case in March 2023 showed a different failure point: it was not the redemption queue problem, but the reserve itself. According to Eco, Circle disclosed that its USDC reserves were deposited with Silicon Valley Bank (SVB), which the FDIC had taken over;Eco set the failure as March 10 to 13, 2023. USDC's trading price on Coinbase and Curve fell to about $0.87;Chainlink reported in its article that the token fell below $0.88 in the secondary market. After the U.S. Treasury, Federal Reserve and FDIC announced on March 12 that SVB depositors would be fully compensated, Eco reported that the USDC recovered above $0.99 in 48 hours and fully recovered in about three days.

Crypto-collateralized stablecoins: CDP and anchored stability modules

The white paper released by MakerDAO in 2017 elaborated on the core purpose of such stablecoins: to deal with volatility. The article points out that assets such as Bitcoin can fluctuate as much as 25% in a single day and exceed 300% in a single month, so a collateral-based dollar stabilization tool is needed. According to the white paper, Dai is generated through collateralized debt positions (CDP)-users lock collateral in smart contracts and use it to create Dai as debt; the collateral remains locked until the debt is repaid.

The Maker system subsequently added the Anchoring Stability Module (PSM), which allows holders to exchange Dai for a limited amount of USDC at a fixed 1:1 exchange rate. According to a paper published in 2024 by Conclave researcher Yuval Boneh, the module aims to strengthen the arbitrage mechanism near anchored exchange rates. However, during the SVB crisis, this USDC link instead became a risk exposure: Eco's event table showed that DAI fell to approximately $0.89 on March 11, 2023, directly due to the USDC crisis transmitted through PSM collateral. Although the arbitrage mechanism of DAI is restored, the premise is that the arbitrage mechanism of USDC is restored first.

Another crypto-collateralized design-Liquity's LUSD-sets a redemption floor price instead of a fixed exchange rate. The Conclave/arXiv paper (July 18, 2024) explains that the minimum redemption fee for LUSD is 0.5%, which means redemption from the agreement is profitable only when the LUSD is trading at less than $0.995-the fee itself prevents continued redemption pressure when the price is around $1.

Algorithm stablecoins: Arbitrage without anchoring

UST's design relies on dual-token arbitrage rather than dollar reserves. To mint a UST, users need to buy a Luna worth US$1 and destroy it in the Terra agreement; if the UST price is lower than US$1, the arbitrageur can buy it at a low price, redeem the Luna worth US$1 at face value, and then sell it for a profit-this buy-and-destroy mechanism should have pushed the price back to US$1. According to a paper published in April 2023 by EPFL researchers Gleb Kurovskiy and Natalia Rostova, before the May 2022 crash, UST's circulation supply was US$18.7 billion, while Luna's mortgage support was US$20 billion, and Luna was trading at close to US$80.

The paper pointed out that two superimposed reasons led to the collapse of the arbitrage mechanism. First, there are built-in limits on UST redemptions, and the effective redemption rate during panic periods has soared from the usual 0.5% to 60%, meaning that traders can only earn 40 cents per dollar when redemying UST; the paper shows that the UST price is mathematically limited to 1 minus the rate, so an increase in the rate will inevitably lead to a decline in the price. Second, even after Terra lifted redemption limits on May 12, 2022, prices had not recovered, which the paper attributed to increased volatility in Luna, lagging price predictors, and a lack of ways to directly convert UST into U.S. dollars or assets outside the Terra system. The paper pointed out that arbitrage profits in the final days of the crash were as high as 200% to 400%-indicating that the carry trade itself was still open, but completing the arbitrage required the destruction of a large amount of Luna value, thereby accelerating the downward spiral rather than stopping it. Eco's records quantify the results: UST fell to about $0.30 on May 11, 2022, and actually fell to $0 on May 13; and Luna's market value plummeted from $40 billion to less than $100 million.

Comparison of six unanchoring cases

Eco's event table lists the low point and recovery time of the largest unanchoring case since 2022 based on CoinGecko's historical price data and media reports:

UST: In May 2022, the low point was about US$0.00 and never recovered. The root cause was algorithm failure.

USDT: May 12, 2022, the low was about US$0.95, recovered in about 2 weeks, the root cause was massive redemptions during the Luna crisis.

USDN (Neutrino): In April 2022, the low was about US$0.78 and never fully recovered. The root cause was algorithm design and the Waves token collapse.

USDC: March 10-13, 2023, the low was about US$0.87 and recovered in about 3 days. The root cause was SVB reserve risk exposure (US$3.3 billion).

DAI: March 11, 2023, the low was approximately US$0.89, recovered in approximately 3 days. The root cause was the USDC crisis transmitted through PSM collateral.

BUSD: Starting from February 2023, the low is about US$0.995, and the exit is slowly contracting. The fundamental reason is that the NYDFS ordered Paxos to stop casting.

BUSD is a special case: On February 13, 2023, the New York Department of Financial Services ordered Paxos to stop minting the token, but the reserves remained intact and redemption channels remained open, so the token itself barely deviated from its face value-Eco reported that in subsequent weeks, the discount placed on Curve and the token was only 0.3% to 0.5%. The actual impact is reflected in supply: Eco said that as holders shifted to USDT and USDC, BUSD supply fell from US$16 billion to less than US$50 million in the fourth quarter of 2024.

Common misunderstandings

To regard every unanchor as evidence of the disappearance of reserves is to regard UST as a template, and it is an exception. In the above-mentioned cases where there was an effective redemption path (USDT, USDC, DAI), once the path was reopened, the discount was eliminated because the underlying claim on the U.S. dollar (or on the USDC) was never really destroyed. UST is different: Kurovskiy and Rostova's paper argues that its redemption mechanism itself collapsed under pressure, not just confidence-which is why prices never recovered even after redemption limits were lifted on May 12, 2022. Eco also pointed out that Ethena's USDe (which holds spot ETH and BTC and is hedging short through perpetual contracts rather than relying solely on paired tokens) has gone through three market cycles as of the first quarter of 2026 without being unanchored-this is Eco's description of its continued performance, rather than a claim that the design is immune to the same failure pattern.

What is not covered in this article

The lows and recovery windows are from Eco's summary table, which is based on CoinGecko price history and media reports rather than raw data directly from the exchange's order book. When data from different sources are inconsistent, this article presents both rather than an alternative: Eco set USDC's SVB weekend low at around $0.87, and Chainlink set it at below $0.88. The two are not the same number. For transaction-level details of why arbitrage collapsed, only the UST case is supported by the paper by Kurovskiy and Rostova; the record of this article lacks equivalent level research on USDC, DAI, or BUSD events, and therefore cannot describe the mechanism with the same precision. Bitcoin's 25% single-day volatility and 300% single-month volatility data come from MakerDAO's December 2017 white paper and are now outdated; they only illustrate the background in which the Dai system was built, not current market conditions. Finally, all this evidence does not support predictions of whether any existing stablecoin will unanchor or the speed of its next recovery-depending on redemption ability, counterparty exposure and the level of market panic, factors that vary by event and cannot be read from historical tables.