The thing that trips people up about a DAI-style stablecoin is that there is no dollar sitting anywhere. When you hold a fiat-backed token, the mental model is easy. Somewhere there is an account with a matching balance, and the peg is a promise that you can get that dollar back. A collateralized debt position stablecoin does not work that way at all. The dollar on the screen is minted into existence by someone locking up crypto and borrowing against it, and the price near one dollar is held there by a handful of economic levers rather than a custodian. Once you see which lever pushes which direction, most of the weird behavior around these coins stops being mysterious.
I want to walk through the levers one at a time. The interesting part is not that they exist. It is that each one only works under specific conditions, and knowing where each one breaks tells you when a peg is actually at risk versus just wobbling.
Minting, redemption, and the two directions of a break
Start with how the supply gets created. Someone deposits collateral, usually crypto worth more than the loan they take out, and mints stablecoin against it. The overcollateralization is the whole safety margin. If you have to lock up roughly a dollar fifty or more of volatile collateral to borrow one stable dollar, the system can absorb a fair amount of collateral price decline before the loan is underwater. When the collateral falls too far, the position gets liquidated, the collateral is sold, and the borrowed stablecoin is bought back and destroyed.
That already tells you something. A peg can break in two directions, and they are not symmetric. When the stablecoin trades above a dollar, minting new supply is profitable. You lock collateral, mint at par, sell into the market above par, and pocket the difference. That arbitrage naturally pushes the price back down toward one. When the stablecoin trades below a dollar, the fix runs the other way. Anyone who has an open debt position can buy the cheap stablecoin on the market, repay their loan at face value, and unlock their collateral for less than they borrowed. That buying pressure pushes the price back up.
The catch is that the below-peg repair depends on there being open debt to repay. If demand for leverage dries up and almost nobody wants to hold a position, the natural buyers of a discounted stablecoin thin out too. That is the first place the theory and the reality drift apart.
The stability fee is the throttle on supply
The stability fee is the interest rate borrowers pay on the stablecoin they mint. It is easy to read it as just a revenue mechanism, but it is really the main throttle on how much supply exists. Raise the fee and borrowing gets more expensive, so fewer people mint and some existing borrowers close their positions to stop paying. That shrinks supply. Lower the fee and the opposite happens.
Now connect that to the peg direction. If the stablecoin is trading persistently below a dollar, that is a sign there is too much of it relative to demand. Raising the stability fee is the slow lever that pulls supply back in. Borrowers repay, supply contracts, scarcity nudges the price back up. If the coin is trading above a dollar, cutting the fee makes minting cheaper and encourages more supply to flow out into the market, which drags the price back down.
The word to sit with is slow. The stability fee is a governance lever. Someone has to propose a change, it gets voted on, and it takes effect after a delay. During a fast dislocation this lever is almost useless, because the market has already moved by the time the vote clears. The fee corrects a persistent drift over days and weeks, not a peg falling apart in an afternoon.
The peg stability module and where the ballast comes from
Because arbitrage through minting and liquidation is slow and lumpy, most mature CDP systems bolt on a peg stability module, or PSM. The idea is simple and it is the closest thing these systems have to a bank account. The PSM lets anyone swap a whitelisted fiat-backed stablecoin for the CDP stablecoin at a fixed one-to-one rate, in both directions, usually for a tiny fee or none at all.
That fixed swap is a hard rail on the price. If the CDP coin drifts to ninety-nine cents, you buy it cheap on the open market and swap it through the PSM for the other stablecoin at par, taking a near-riskless cent. That buying holds the floor. If it drifts above a dollar, you swap the other direction and sell the freshly minted CDP coin above par, holding the ceiling. In practice the PSM does more day-to-day peg defense than any other mechanism, precisely because it is instant and requires no governance vote and no liquidation.
The uncomfortable part is what the PSM actually is. When people swap fiat-backed stablecoins in, the module fills up with those stablecoins, and the CDP system ends up holding a large reserve of them. So the more you lean on the PSM for stability, the more your decentralized stablecoin is quietly backed by someone else's centralized one. You have imported their custody risk, their freeze risk, and their regulatory risk. The peg feels rock solid right up until the reserve stablecoin has its own problem, and then the ballast becomes the leak.
Where it strains, in order
Put the levers together and you can reason about failure modes instead of just reciting them. Here is roughly how I think about which lever matters in which situation.
- Slow drift below peg from oversupply. Stability fee is the right tool. Raise it, let borrowers repay, wait. If the PSM has capacity, arbitrage through it does the fast work while the fee does the structural work.
- Slow drift above peg from excess demand. Cut the fee to encourage minting, and the PSM lets holders of the fiat stablecoin mint CDP coin at par and sell the premium. This direction is usually the easier one to fix.
- Sharp collateral crash. This is the dangerous one. Collateral falls fast, positions go underwater, liquidations fire all at once. If liquidation auctions cannot clear fast enough or at fair prices, the system takes on bad debt, meaning stablecoin exists that is no longer fully backed. That is a peg threat that no fee change fixes.
- Mass redemption or a run on the reserve. If everyone wants out through the PSM at once, the module empties of its reserve stablecoin. Once it is dry, the hard rail on one side of the peg is gone, and the price can gap.
The collateral crash case has a nastier second layer. When you liquidate volatile collateral into a falling market, the selling pushes the price down further, which triggers more liquidations. If a large share of the collateral is correlated, a single asset or a cluster of similar assets moving together, the whole book can cascade. This is why the collateral mix matters more than the headline overcollateralization ratio. A system collateralized entirely by one volatile asset is more fragile than one at the same ratio spread across assets that do not all crash on the same day.
A practical way to read peg risk
When I am sizing up whether a CDP stablecoin's peg is genuinely sound or just calm for now, I run through a short checklist rather than trusting the current price.
- How much of the backing is the PSM reserve, meaning other stablecoins, versus actual crypto collateral. High PSM dependence means the peg is borrowed from someone else's balance sheet.
- How concentrated and correlated the crypto collateral is. Diversified and liquid collateral survives a crash. Concentrated collateral does not.
- Whether liquidation auctions have historically cleared cleanly under stress, or whether the system has taken on bad debt before.
- How much headroom the stability fee has. A fee already pinned high has little room left to defend a below-peg drift.
None of these show up in the price. A peg can read as one dollar flat right up to the moment one of these conditions gives out. The price tells you the levers are holding now, not that they can hold under the next shock, and the four questions above are what separate the two.
If you want to watch this in real conditions, the useful signals are on-chain, since collateral positions, liquidation events, and PSM balances are all public. That is a lot of what we track on Blockcircle. Reading the collateral book and the reserve mix directly beats reading the peg price, because by the time the price moves, the interesting decision was already made a few blocks earlier.