Stablecoins Are Not All Stable in the Same Way
USDC fell to $0.87 in March 2023 despite full reserves and regular audits. The word "stable" hides four fundamentally different mechanisms, each with its own failure mode.
In March 2023, USDC fell to $0.87. Not a scam coin or an algorithmic experiment—Circle’s fully reserved, audited, institutional-grade stablecoin lost thirteen cents on the dollar in a matter of hours. The trigger was Silicon Valley Bank’s collapse, which temporarily locked $3.3 billion of Circle’s reserves inside a failed institution. The reserves existed. The audits were real. The peg broke anyway. This wasn’t a failure of honesty or math. It was a failure of mechanism, and mechanism is what determines how a stablecoin breaks. The word “stable” hides fundamental architectural differences. Four distinct designs all produce tokens that display “$1.00” on your screen, but they fail in completely different ways under completely different conditions. Understanding the mechanism is understanding the risk.
Four Ways to Build a Dollar You Can’t Hold
When you hold a stablecoin worth one dollar, you’re holding a claim on something, not the dollar itself. The difference between stablecoins isn’t cosmetic. It’s architectural. Four fundamentally different mechanisms produce tokens that all say “$1.00” on the screen, and each one shifts the risk somewhere else.
Fiat-collateralized stablecoins are the simplest bet: a company holds actual dollars in a bank account and issues tokens against them. USDC and USDT both follow this model, though with dramatically different levels of transparency. Circle publishes monthly attestations showing what backs each USDC. Tether has spent years deflecting questions about whether its reserves are actually dollars or a mixture of commercial paper, loans, and other instruments that merely claim to be worth a dollar. The mechanism works until you need your dollar back and discover the issuer can’t or won’t redeem it. This is counterparty risk dressed up as technology.
Crypto-collateralized stablecoins lock volatile assets into smart contracts and mint stablecoins against them, always requiring more collateral than the stablecoin’s face value. DAI typically demands $150 to $200 worth of Ethereum to create $100 worth of stablecoin. The buffer absorbs price swings, but if the collateral drops fast enough, the position gets liquidated automatically. You’ve traded counterparty risk for liquidation risk and smart contract risk.
Algorithmic stablecoins use no backing at all. They expand and contract supply through code, relying on arbitrage incentives to keep the price at a dollar. When demand falls, the system must convince people to destroy tokens or accept promises of future value. This worked until it didn’t. TerraUSD held its peg for months, then lost it in May 2022 and erased $40 billion in seventy-two hours. The mechanism depends entirely on confidence, and confidence is not collateral.
Commodity-backed stablecoins tie each token to a physical asset, usually gold. They solve the fiat trust problem by anchoring value to something tangible, but they import every problem of physical custody: storage costs, verification audits, and the need to trust that the gold actually exists in a vault somewhere.
| Architecture | Primary Risk | What Breaks First |
|---|---|---|
| Fiat-collateralized | Counterparty and regulatory | Issuer can’t or won’t redeem |
| Crypto-collateralized | Liquidation and smart contract | Collateral drops faster than system can respond |
| Algorithmic | Confidence collapse | Incentive loop fails under selling pressure |
| Commodity-backed | Custody and verification | Physical asset proves unavailable or misstated |
The table clarifies what matters: not whether a stablecoin can hold its peg under normal conditions, but what type of abnormal condition will break it. Each architecture survives different scenarios and fails under different stress. A trader treating all stablecoins as interchangeable cash equivalents is making an unpriced bet on four separate failure modes, usually without realizing it.
The Trust Tax: What Fiat-Backed Stablecoins Actually Promise
When you hold USDT or USDC, you’re not holding dollars. You’re holding a claim on dollars that someone else is supposed to be storing for you. That distinction sounds pedantic until the moment it isn’t.
Both Tether’s USDT and Circle’s USDC promise 1:1 backing with US dollars, but the mechanics and risks differ more than their identical ticker prices suggest. Think of it this way: two insurance policies can both promise to cover your house, but one might be underwritten by a Fortune 500 company with transparent financials while the other is issued by an offshore entity that publishes summaries instead of audits. Same promise, different risk.
What the Reserves Actually Contain
Circle publishes monthly attestations from a registered public accounting firm. As of recent reports, USDC reserves consist almost entirely of cash and short-dated US Treasury bills held at regulated US financial institutions. The composition is boring, which is the point. Boring means predictable. Predictable means you can estimate what happens under stress.
Tether’s reserve composition is less straightforward. While the company has improved transparency since facing regulatory pressure in 2021, its reserves historically included commercial paper, secured loans, corporate bonds, and other assets beyond cash and treasuries. The percentages shift. Commercial paper carries credit risk. Secured loans depend on collateral quality and counterparty solvency. These aren’t fatal flaws, but they’re variables you can’t control or fully observe.
| Stablecoin | Primary Reserve Assets | Disclosure Frequency | Third-Party Verification |
|---|---|---|---|
| USDC | Cash, US Treasury bills | Monthly | Attestation by registered CPA firm |
| USDT | Treasury bills, commercial paper, secured loans, cash | Quarterly | Attestation (not full audit) |
| BUSD (discontinued) | Cash, US Treasury bills | Monthly | Attestation by registered CPA firm |
The table shows what you’re actually trusting when you park capital in a stablecoin. Transparency isn’t about morality; it’s about measurable risk. An attestation tells you what exists today. It doesn’t tell you whether those assets will remain liquid when everyone tries to redeem at once.
When the Bank Holding Your Dollars Fails
March 2023 proved that even transparent, well-audited stablecoins carry systemic risk. USDC briefly depegged to $0.87 after Circle disclosed that $3.3 billion of its $40 billion reserves were held at Silicon Valley Bank, which had just been seized by regulators. The math was sound. The reserves existed. But they were temporarily inaccessible.
This wasn’t a run on Circle. It was a run on a bank that happened to hold a portion of Circle’s reserves. The distinction matters because it reveals the actual risk: banking system exposure. You’re not just trusting the stablecoin issuer’s competence or honesty. You’re trusting every financial institution in the custody chain, plus the regulatory environment, plus the speed of government intervention during a crisis.
USDC returned to peg within days after the FDIC and Treasury announced that SVB depositors would be made whole. The system worked. But for 72 hours, the probability distribution was wide open. Traders who needed liquidity sold at $0.87. Others bought the dip and collected a 15% gain in three days. The difference wasn’t information or skill. It was risk tolerance and the ability to wait.
Redemption bottlenecks are the other hidden variable. Most stablecoin issuers allow redemptions only for large amounts, often $100,000 minimum. Below that threshold, you’re selling to another user at whatever the market price happens to be. That works fine when confidence is high. When confidence cracks, the peg is whatever the marginal panic seller will accept. You can be right about the reserves and still take a loss if you need to exit during a liquidity squeeze.
The trust tax isn’t a fee you pay. It’s the risk you accept in exchange for the convenience of moving dollars on a blockchain. That risk is institutional, not mathematical. Circle and Tether can both have perfect 1:1 backing and you can still lose money if the bank fails, regulators freeze accounts, or redemption queues grow faster than the issuer can process them. Knowing that won’t prevent a crisis, but it changes how you size your position and how long you’re willing to stay exposed.
Over-Collateralization as a Shock Absorber
You deposit $1,500 worth of Ethereum into a smart contract. You receive $1,000 of DAI. This isn’t theft or poor exchange rate design. It’s insurance premiums baked into the structure itself.
Crypto-backed stablecoins like DAI use over-collateralization to absorb the volatility that would otherwise destroy the peg. The protocol demands more value in than you take out because your collateral moves. Ethereum can swing 10% in an afternoon. The buffer means that even if your collateral drops from $1,500 to $1,200, the protocol still holds enough to back the $1,000 of DAI you minted. You’re not betting the collateral stays flat. You’re betting it doesn’t fall past the liquidation threshold before you notice.
That threshold is where the game turns dangerous. If your collateral drops below the required ratio, liquidation engines automatically sell your collateral to repay the debt. No warning call. No grace period. The system protects itself, not you. In March 2020, during a network congestion spike, some DAI vaults were liquidated at zero bid because the auction mechanism couldn’t find buyers fast enough. Users lost everything despite being over-collateralized moments before.
| Collateral Ratio | ETH Value at Mint | DAI Minted | ETH Drop Before Liquidation |
|---|---|---|---|
| 150% | $1,500 | $1,000 | 33% |
| 175% | $1,750 | $1,000 | 43% |
| 200% | $2,000 | $1,000 | 50% |
The table shows the tradeoff clearly: higher collateral ratios give you more room to breathe, but they tie up more capital for the same borrowing power. At 150%, you’re liquidated if Ethereum drops a third. At 200%, you survive a 50% drawdown. Neither is bulletproof in crypto.
The peg itself isn’t guaranteed by the over-collateralization. It’s maintained by arbitrage. When DAI trades above $1, arbitrageurs mint new DAI by locking collateral, then sell it for profit until the price falls back. When it trades below $1, they buy cheap DAI, repay their vaults, and unlock collateral. In March 2020, DAI briefly spiked to $1.05 as liquidity dried up and liquidations accelerated. Borrowers who had set tight stop-losses or automatic repayment triggers got caught when the stablecoin they thought was stable suddenly wasn’t.
The risk has shifted. You’re not trusting a bank to hold dollars or a company to maintain reserves. You’re trusting that the liquidation engine fires accurately, that price oracles report the truth, and that the collateral itself remains liquid enough to sell. Smart contract risk replaces counterparty risk. Oracle manipulation replaces accounting fraud. The probabilities change shape but they don’t disappear.
The Death Spiral: When Confidence Is the Only Collateral
In May 2022, TerraUSD fell from $1.00 to $0.03 in seventy-two hours. Forty billion dollars evaporated. The mechanism that was supposed to stabilize the price instead accelerated its collapse, turning a wobble into a systemic wipeout. This wasn’t a hack or a regulatory surprise. It was a structural failure built into the design.
Algorithmic stablecoins don’t hold dollars or Treasury bills in a vault. They maintain their peg through code and incentives. TerraUSD used a dual-token system: burn one dollar’s worth of LUNA tokens to mint one UST, or burn one UST to mint one dollar’s worth of LUNA. If UST traded below a dollar, arbitrageurs could buy it cheap, burn it for LUNA worth a dollar, and pocket the difference. The profit motive was supposed to pull the price back up. It worked perfectly until it didn’t.
The system depended on one assumption: that people would always believe LUNA had value. When UST slipped to $0.98, then $0.95, arbitrageurs hesitated. Minting billions of new LUNA tokens to absorb the selling pressure diluted LUNA’s price. As LUNA fell, confidence in the redemption mechanism cracked. Traders who might have stabilized the peg instead rushed for the exit. The algorithm responded by printing more LUNA, which crashed harder, which eroded confidence further. Classic death spiral.
Network congestion made it worse. Ethereum gas fees spiked above $100 per transaction during the panic. Arbitrage trades that required speed and volume became prohibitively expensive. The very people the system relied on to restore the peg couldn’t afford to execute the trades. The mechanism that promised stability through decentralized market forces couldn’t function when the market moved fast.
| UST Price | Arbitrage Incentive | LUNA Supply Impact | Market Response |
|---|---|---|---|
| $0.98 | 2% profit on redemption | Modest inflation | Traders buy UST, system stabilizes |
| $0.90 | 10% profit, but LUNA falling | Heavy new issuance | Hesitation, smaller positions |
| $0.60 | Theoretical profit, real doubt | Exponential supply growth | Redemptions pause, selling accelerates |
| $0.10 | Negative after gas fees | Hyperinflationary | Full capitulation, no bid |
Confidence isn’t collateral. It’s a psychological state that reverses faster than any algorithm can respond. When the belief that underpins the entire mechanism breaks, the code keeps running but the economics invert. What was designed to buy the dip becomes a machine that prints worthless tokens into a collapsing market. You can’t arbitrage your way out of a crisis if the asset you’re minting loses value faster than you can sell it.
This is the risk profile algorithmic stablecoins carry. They work beautifully in calm markets and fail catastrophically under stress. The odds aren’t hidden. The mechanism is transparent. But the failure mode isn’t a slow leak—it’s a trapdoor that opens all at once when the crowd stops believing.
Risk Profiles Side by Side
When you hold a stablecoin, you’re not holding dollars. You’re holding a claim on a mechanism designed to behave like a dollar. The mechanism determines everything about how and when that claim fails.
Fiat-backed stablecoins put trust in an institution. Someone holds actual dollars in a bank account, and you hold a token they issued against those reserves. The failure mode is institutional: the bank collapses, regulators freeze the accounts, or the issuer lies about what’s really in the vault. Circle can publish attestations monthly, but you’re still betting on their honesty and their banking relationships. Tether’s opacity makes that bet harder to price, which is why USDT trades at a slight discount during market stress even though it’s used more widely.
Crypto-backed stablecoins replace institutional trust with collateral math. DAI requires you to lock up more value than you borrow, typically 150% or higher. The system liquidates your position automatically if the collateral drops too far. The failure modes are technical and cascading: a flash crash liquidates positions faster than the system can absorb them, an oracle feeds bad price data, or a smart contract bug drains the collateral vault. You’re playing a different game here, one where code reliability and liquidation mechanics matter more than auditor reports.
Algorithmic stablecoins bet everything on confidence and arbitrage incentives. No reserves, no collateral, just a mechanism that’s supposed to expand and contract supply to defend the peg. The failure mode is psychological and self-reinforcing. Once people doubt the peg will hold, arbitrage stops working, redemptions accelerate, and the entire structure collapses inward. Terra’s $40 billion evaporation in May 2022 took 72 hours from doubt to death spiral. The mechanism doesn’t degrade gracefully. It works completely until it doesn’t work at all.
| Type | Mechanism | Primary Failure Mode | Stress Test Question |
|---|---|---|---|
| Fiat-backed | Institutional reserves in bank accounts | Bank failure, regulatory seizure, reserve fraud | Can I verify the reserves exist and match issuance? |
| Crypto-backed | Over-collateralized with volatile assets | Liquidation cascade, oracle manipulation, smart contract exploit | What collateral ratio survives a 50% flash crash? |
| Algorithmic | Supply adjustments via arbitrage incentives | Confidence collapse, death spiral, arbitrage breakdown | What happens when nobody wants to arbitrage the peg back? |
Understanding the mechanism tells you what mental stress test to run. Don’t ask whether a stablecoin is safe. Ask what conditions would break it, and whether you’re exposed to those conditions right now.
The label “stablecoin” creates false equivalence. It groups together instruments that share a price target but almost nothing else. A fiat-backed token is a bet on institutional integrity. A crypto-backed token is a bet on collateral buffers and smart contract execution. An algorithmic token is a bet on sustained confidence and arbitrage participation. Calling them all stablecoins is like calling both a savings account and a parachute “safety equipment” because they both reduce risk. Technically true, but the distinction matters when you’re deciding which one to trust at ten thousand feet.
The practical takeaway isn’t to avoid stablecoins. It’s to treat them as distinct instruments with distinct breaking points. Know what backs the token you’re holding. Know what happens if that backing fails or becomes inaccessible. Know whether you can afford to wait out a temporary depeg or whether you need liquidity on demand. Size your exposure accordingly. If you’re holding six figures in a stablecoin, the difference between USDC’s reserve transparency and USDT’s opacity isn’t academic. If you’re minting DAI against Ethereum collateral, the difference between a 150% and 200% collateralization ratio isn’t a minor optimization. If you’re parking capital in an algorithmic stablecoin, you’re accepting trapdoor risk in exchange for decentralization.
Risk management isn’t about eliminating exposure. It’s about knowing what you’re exposed to and deciding whether the tradeoff makes sense. The mechanism is the risk. Understand the mechanism, and you understand what breaks it.
Keep reading
The Spread Is a Price, and Someone Is Charging It
The spread is a price charged every time you trade, and it's deducted before the market moves. Understanding who collects it and…
Leverage Amplifies the Bet, Not the Edge
Leverage multiplies position size, not skill. A 1% adverse move can wipe out a 100:1 leveraged account, while the unleveraged trader survives…
Where Your Broker Actually Sends the Order
You click buy, the order fills instantly, and you assume it went to the market. Most traders never ask where that order…
