What Happens When a Stablecoin Depegs?

The word "depeg" covers a remarkable range of outcomes. A stablecoin at $0.998 is technically depegged. TerraUSD went to zero while technically depegged. Treating these as the same category produces confusion about what actually matters — which is whether the mechanism that maintains the peg can repair itself, and whether it has enough to work with.

That's the actual question. Not "is the price below $1" but "what broke, and can the system close the gap."

How Stablecoins Maintain Their Pegs

The peg mechanism varies by design. There are three architectures worth distinguishing.

Fiat-backed stablecoins (USDC, USDT) hold dollar-denominated reserves — cash and short-term treasuries — and allow qualified holders to redeem 1 token for $1 in cash. The price stays near $1 through arbitrage: if USDC drops to $0.98, any institution with redemption access can buy it cheaply and redeem it at face value for a near-guaranteed 2-cent profit. That loop, repeated at scale, pulls the price back. The mechanism is clean and reliable — as long as two conditions hold: the issuer can process redemptions, and the reserves are actually there.

Crypto-backed stablecoins (DAI, LUSD) are issued against on-chain collateral locked in smart contracts, usually at overcollateralization ratios. DAI requires you to lock more than $1 of ETH (or other approved assets) to borrow $1 of DAI. If DAI falls below $1, the system creates incentives to reduce supply: borrowers can repay DAI loans at a slight discount and reclaim collateral, and the protocol's stability fee adjustments reduce new borrowing. If DAI rises above $1, the opposite — new DAI can be minted cheaply, increasing supply. The mechanism works through continuous economic pressure. It can fail under extreme collateral drawdowns.

Algorithmic stablecoins (TerraUSD/UST was the largest) maintain the peg through a mint/burn relationship with a companion token. When UST was below $1, holders could burn 1 UST to receive $1 worth of LUNA. When UST was above $1, anyone could burn $1 of LUNA to mint 1 UST. On paper: supply adjusts in the right direction. In practice: the system has no floor — no hard asset that can be seized and sold if the mechanism fails.

What Actually Happens During a Depeg

The mechanics differ by architecture.

Fiat-backed depegs are usually temporary and most often caused by blocked redemptions, not missing reserves. The USDC depeg in March 2023 is the clearest case study. When Silicon Valley Bank failed, Circle disclosed that $3.3 billion of USDC's reserves were held at SVB — roughly 8% of reserves, uncertain for a weekend. USDC dropped to around $0.87 on some exchanges. But once the FDIC indicated SVB deposits would be fully covered, and Circle confirmed reserves were intact and redemptions would resume, USDC recovered to $1.00 within two trading days.

The arbitrage loop hadn't broken. It had been briefly suspended. That distinction matters.

Crypto-backed depegs are driven by collateral stress. Black Thursday — March 12–13, 2020 — saw ETH fall roughly 50% in 24 hours, triggering a cascade of liquidations across MakerDAO positions. In some cases, zero-bid auctions allowed liquidators to claim collateral for free when the system was overwhelmed by gas spikes and congestion. DAI briefly depegged above $1 (a demand spike as people rushed to buy DAI to repay loans and avoid liquidation) and MakerDAO was left with approximately $5.4 million in bad debt. The solution was a governance vote to mint new MKR tokens and sell them — diluting existing MKR holders to recapitalize the system.

The collateral was real. It just fell too fast for the mechanism to clear.

Algorithmic depegs are a different category entirely. Once confidence breaks, the mechanism designed to restore the peg actively accelerates the collapse. This is what happened to TerraUSD in May 2022.

When UST started losing its $1 peg, holders burned UST to mint LUNA. This increased LUNA supply. But LUNA's price was sustained by belief in the Terra ecosystem — as that belief weakened, LUNA's price fell. A falling LUNA price meant each UST burn now minted more LUNA (since the redemption is for $1 worth of LUNA, not a fixed number of tokens). More LUNA in circulation, falling price, more panic, more UST burns, more LUNA. The loop was reflexive. By the time it stopped, LUNA had gone from roughly $80 to fractions of a cent. UST went with it.

No hard collateral meant no floor. The mechanism wasn't broken — it worked exactly as designed. The design itself was the problem.

Where the Vulnerabilities Are

For fiat-backed stablecoins, the risk is concentrated in banking relationships and reserve liquidity. SVB demonstrated that a single banking counterparty failure can temporarily impair redemptions — a brief panic was the result. Concentration is the constraint, not the reserve model itself. Properly diversified reserves with short-duration, liquid assets are a solved problem in institutional finance. The remaining gap is regulatory: which stablecoins have legally enforceable redemption obligations, and which are operating on trust.

For crypto-backed stablecoins, the constraint is collateral correlation. If all accepted collateral is ETH and ETH drops 50% quickly, every position becomes simultaneously undercollateralized. Diversifying collateral to include uncorrelated assets — real-world assets, other cryptocurrencies with different risk profiles — addresses this structurally, but introduces new dependencies on oracles and off-chain legal structures.

For algorithmic stablecoins without hard backing, the constraint is confidence — which is not a constraint at all. It can evaporate.

What's Changing

After the Terra collapse, pure algorithmic stablecoins without direct collateral backing have effectively exited the serious conversation. Investors who lost money learned the mechanism; regulators who watched added them to priority lists.

Regulatory frameworks are now actively distinguishing between stablecoin types. MiCA in the EU requires fiat-backed stablecoin issuers to hold segregated, liquid reserves and publish regular attestations. The US Stablecoin TRUST Act (still moving through Congress as of 2026) would impose similar requirements on payment stablecoins, formalizing what's currently a matter of issuer policy.

MakerDAO has been reducing its dependence on ETH-only collateral pools by adding Treasury bills and other real-world assets. The result is a more diversified risk profile for DAI — though it also means the system now carries off-chain counterparty dependencies it didn't have before.

The core mechanism hasn't changed for fiat-backed tokens. The regulatory and counterparty infrastructure around them is catching up to the risk.

What Would Confirm Stability

For fiat-backed stablecoins: mandatory reserve segregation with regular third-party audits, combined with clear redemption rights recognized under applicable law, and diversification across banking counterparties so that no single failure blocks redemptions.

For crypto-backed stablecoins: successful navigation of future volatile periods without material bad debt accumulation, and demonstrated stability with diversified collateral types.

What Would Break It

For fiat-backed: Reserve fraud at an issuer (the most significant unresolved risk for some tokens), coordinated regulatory action blocking redemptions, or a banking crisis affecting multiple counterparties simultaneously.

For crypto-backed: A correlated collapse in collateral assets that outruns liquidation mechanisms — essentially a repeat of Black Thursday at larger scale.

For any stablecoin: A loss of confidence that causes a bank run faster than arbitrage can respond. Even well-collateralized systems can face temporary operational impairment under peak redemption pressure.

Timing

Now: The distinction between fiat-backed and algorithmic stablecoins is load-bearing for any risk assessment. Pure algorithmic models with no hard backing are effectively disqualified from serious infrastructure use.

Next: US and EU regulatory frameworks will formalize reserve and redemption requirements for fiat-backed tokens over the next 12–24 months. This changes what's legally required, not just what's operationally sensible.

Later: How crypto-backed stablecoins handle larger-scale stress events — and whether real-world asset collateral diversification holds under systemic pressure — remains a multi-year open question.

Boundary

This covers the mechanism of stablecoin depegs and how different architectures respond. It doesn't assess current reserve adequacy for any specific issuer — that's a present-state question requiring current attestations, not mechanism knowledge.

The architecture is described here. Whether any specific stablecoin's current reserves are adequate lives elsewhere.