Every few weeks somebody asks me a version of the same question. Their stablecoin pays them yield just for sitting in a wallet, and they want to know where the money actually comes from. It is the right question, and the answer depends entirely on which kind of yield-bearing dollar they hold, because the two dominant designs share a naming convention and almost nothing else. One is a savings account with extra steps. The other is a hedge fund trade wearing a dollar costume. Both can run smoothly for years, both can break, and they break in completely different ways, so it is worth walking through the machinery before you park anything meaningful in either.
The savings-wrapped dollar
The sDAI pattern is the easier one to reason about. You deposit a stablecoin into a vault, the vault routes it into the protocol's savings module, and the savings rate is funded by whatever the protocol earns. Historically that meant interest paid by borrowers taking overcollateralized loans. In many deployments a large share comes from short-term government debt held through real-world asset structures, so in practice you are collecting something close to a T-bill rate with a DeFi wrapper around it.
The wrapper is usually an ERC-4626 vault, which matters for one practical reason. Your share count stays fixed while the exchange rate between shares and the underlying climbs, so the token is designed to drift above one dollar over time. People see the wrapped token quoted at a premium to the underlying and assume something has gone wrong. Nothing has, the premium is the accumulated yield. The number worth watching is the gap between the market price and the redemption value the vault itself reports, because that gap is the real discount or premium.
What can break it. The rate is set by governance, so it can be cut to zero whenever protocol revenue thins out. The real-world asset leg drags in custodians, brokers, legal wrappers, and in some cases bank accounts, which means part of your on-chain dollar depends on off-chain entities honoring off-chain paperwork. And redemption, while usually atomic, relies on a liquidity buffer inside the protocol. If enough people leave at once, the buffer drains and the protocol has to unwind those traditional-market positions on traditional-market timelines, which means days, while you wait. None of this is exotic, it is roughly the risk profile of a money market fund with smart contract risk layered on top and some bank counterparty risk swapped out.
The funding-backed dollar
The second design earns its yield from a completely different place. The protocol holds spot collateral, often staked ETH, and simultaneously shorts an equal notional amount of perpetual futures. The price exposure cancels out, which is what keeps the token dollar-flat, and the income is the funding rate that perp longs pay to shorts whenever the market leans bullish, plus the staking yield on the collateral leg. In a hot market this combination can pay several times what any savings-wrapped design pays, which is exactly why people reach for it.
The catch is that funding is a market price, and market prices go negative. When the crowd flips short, the protocol's hedge starts paying funding instead of collecting it. Staking yield buffers some of that, and most designs keep a reserve fund to absorb the rest. The question you should be able to answer before holding one of these is how big that reserve is relative to total supply, and how long a stretch of negative funding it can absorb before the backing itself starts eroding below par. Historically funding has been positive most of the time, but the negative stretches cluster in exactly the weeks when everything else is on fire, which is also when redemptions spike. The design gets tested at its weakest moment by construction.
Two other failure surfaces are worth naming. The hedges live on centralized derivatives exchanges, usually with collateral parked at off-exchange settlement custodians, so an exchange failure forces the protocol to unwind and re-establish hedges under stress, and during that window the token is quietly directional. And if the collateral leg is a liquid staking token rather than raw ETH, any dislocation between that token and ETH itself punches a hole in the delta-neutral math. Neither of these shows up in the advertised yield, but both are part of what the yield is paying you for.
Why the token can leave a dollar while the backing is fine
A peg is maintained by an arbitrage loop. Somebody has to be able to mint at par and redeem at par, and their arbitrage against the open market is what pulls a drifting price back. Almost every yield-bearing dollar restricts that loop somehow. Minting and redemption often sit behind KYC whitelists, cooldown periods, per-day caps, or queues, which is fine on a calm day. In a panic, secondary-market sellers show up faster than gated redeemers can, and the price on the open market slides even though every dollar of collateral is sitting exactly where it should be. The discount appears because the exit door is narrower than the crowd trying to fit through it, and it can persist as long as the gate does.
The second-order effect is nastier. These tokens get looped as collateral in lending markets, because the yield makes looping look free. Once a lending market's oracle reads the secondary price, a discount starts triggering liquidations, the liquidations sell into thin pools, and the selling deepens the discount that caused the liquidations. The backing can be untouched through the entire spiral. If you hold the token outright and can afford to wait out redemption, this is survivable and sometimes even a decent entry. If you are levered against it, the market price is your price, and the fact that redemption value never moved is a detail for the post-mortem.
The checklist I actually run
Before parking size in any yield-bearing dollar, I want written-down answers to six things.
- State the yield source in one sentence. T-bill interest, loan interest, perp funding, something else. If I cannot say it plainly, I stop here.
- Ask what happens when that source pays zero or negative. Who eats the loss, in what order, and how big the buffer is relative to supply.
- Trace my own exit path as if I am redeeming during a bad week. Direct redemption or secondary market only, KYC, cooldowns, caps, queues, and how much can actually leave per day.
- Check secondary liquidity against my position. If my exit is a pool, what does my full size do to the price on a quiet day, and then assume a fraction of that depth on a loud one.
- Map the counterparties. Exchanges, custodians, real-world asset intermediaries, oracle providers. Each one is a way to lose money that has nothing to do with the yield mechanism.
- Look at how much of the supply is looped in lending markets, because that is the forced-seller overhang that turns a small discount into a deep one.
The rule of thumb underneath all of it is that the spread between what one of these pays and the boring T-bill rate is the fee you are collecting for a specific list of risks, and I try to be able to recite the list from memory. When I hold a savings-wrapped token I am mostly being paid for smart contract and governance risk on top of a money-market position, and I size it accordingly. When I hold a funding-backed dollar I treat it as a carry trade that happens to have a stable price, and I size it the way I would size a trade, assuming the yield can vanish and a temporary discount can show up in the same week. None of that makes me avoid them. It just sets my sizing, and it means I know which machine I am holding before the week arrives that makes the difference obvious.