Intermediate · Lesson 7
Stablecoins: parity is not solvency
A stablecoin promising one dollar is making two claims: a price and a redemption. They are not the same promise.
A stablecoin is a promise wearing the costume of a number. The number is the price on an exchange; the promise is that somebody will convert the token back into the asset it tracks. Most days the costume fits. The useful questions start on the days it does not.
Two promises, one ticker
Parity is the price the token is supposed to hold. Redemption is the issuer's obligation to turn it back into dollars, or into whatever the token claims to represent. A token can trade at one dollar while redemption is closed, and it can be redeemable at one dollar while trading at ninety-seven cents. Both are stable only in the narrow sense that the price is not moving.
Four questions separate the designs:
- Backing. What assets stand behind the token, at what maturity, held where, and valued how?
- Issuer. Who controls the reserves and the contract, and under what legal arrangement?
- Redemption. Who can redeem, at what minimum, with what fee, and how long does settlement take?
- Liquidity. How deep is the secondary market when you need to sell without waiting for redemption?
A token can score well on one and badly on another: «fully backed» and «always redeemable» are different sentences.
Three designs, three sets of assumptions
Fiat-backed tokens hold cash, bank deposits, government bills or similar assets, and promise to redeem at par. The mechanism is a balance sheet and an operations desk. The risks are custodial: a bank account can be frozen, a bill can be sold at a loss in a forced sale, and a reserve report is a photograph of one day.
Crypto-collateralised tokens are minted by borrowers who lock more value than they take out. The collateral is liquidated if its value falls. The mechanism is a loan with an automatic margin call. The risks are market and contract risk: a fast fall, a congested chain or a stale oracle can turn a solvent position into a liquidation queue.
Algorithmic designs try to hold the peg with rules that expand and contract supply, sometimes against a second token. There is no external reserve to redeem against; the backing is the market's willingness to keep playing. That works while confidence is high and becomes circular exactly when it is tested.
Naming a design does not tell you whether it is safe. Tether and USDC are issued against reserves managed off-chain; DAI is generated on-chain by borrowers. The details are in the terms and the reserves, not the label.
Worked example: the discount that will not close
Suppose a token trades at 0.975 and you hold 10,000 of it. The screen says your position is worth 9,750.
- Sell on the secondary market: 10,000 × 0.975 = 9,750, minus the venue's fee.
- Redeem with the issuer: 10,000 × (1 − 0.001) = 9,990, if you are eligible, meet the minimum and can wait for settlement.
The gap is 240, or 2.4% of par. If redemption were open, instant and free, arbitrageurs would buy at 0.975 and redeem at 1.00 until the gap closed. It survives because redemption needs an account, a minimum, a fee and time.
Now suppose the issuer's assets cover only 98% of its liabilities. A price of 0.98 is not a discount; it is a valuation. Backing is a balance sheet: assets divided by tokens outstanding. With 10,000 tokens outstanding and 9,800 in assets, coverage is 98%. If 400 of those dollars sit in a loan to an affiliate that may not be repaid, marking it at zero takes coverage to 94%.
How a peg breaks
Three failures are often mixed together:
- Collateral impairment. In a crypto-collateralised design, 150 dollars of collateral may back 100 tokens. If the collateral falls 40%, it is worth 90 and covers 90% of the tokens. The contract liquidates, but liquidation sells the same asset that is falling, which can turn a price shock into a queue.
- Redemption failure. The reserves may be there, but the door is shut: eligibility rules, minimums, delays, banking access, a legal freeze or an operational backlog. A claim you cannot exercise is not liquidity.
- Liquidity failure. Market makers widen spreads or leave; a large holder sells into thin depth; the price gaps below par. Arbitrage should close the gap, but arbitrage needs capital, access and confidence.
A depeg can be a liquidity event with a healthy balance sheet, a solvency event with an open redemption desk, or both. The price alone does not tell you which.
What to read before you use one
- The terms: who may redeem, the minimum, the fee and the settlement time.
- The reserves: composition, maturity, custody and the date of the latest attestation. A short-dated bill is not a demand deposit.
- The market and dependencies: how deep the secondary market is, and what contract, oracle or liquidation risk you add by lending or pooling the token.
None of this requires an accusation. A reserve report can be honest and still describe assets that cannot be sold at par on the day everyone asks. The question is not whether the issuer is a villain; it is whether the promise survives the day you need it.
Parity is a price. Solvency is a balance sheet. Redemption is an operations desk. They can fail separately, and the ticker shows only the first.
Check yourself
- A token trades at 0.98 while the issuer says reserves cover 100% of tokens. What else do you need to know before calling it a bargain?
- Why can a fully backed token still trade below par for weeks?
- In a crypto-collateralised design, who absorbs the loss when collateral falls faster than the contract can liquidate?
The terms — stablecoin, liquidity, liquidation — are in the glossary. Next, Liquidity pools, DEXs and bridges applies this to pools, and Tokenomics and governance covers who can change the rules.