For a sector that prides itself on transparency, decentralized finance has been remarkably evasive about one of the more uncomfortable truths of the last cycle: not every liquid staking token deserves to sit on a lending market's collateral list. The industry treats stETH, rETH, cbETH, frxETH, and the newer wave of liquid restaking tokens as if they were interchangeable claims on the same underlying asset. They are not. Each carries its own redemption logic, operator concentration, oracle exposure, and slashing geometry, and pretending otherwise has cost depositors real money during every stress event of the past two years.
The premise of an LST is straightforward enough. A user stakes ETH, receives a derivative token, and goes about composing yield strategies on top of it. The premise of supplying that derivative as lending collateral is far less benign. A money market needs collateral that behaves predictably during the exact moments when nothing else does, which means it has to be redeemable into the unit of account quickly, priced honestly by an oracle that does not lag spot during a wick, and resilient to slashing events that arrive with no warning. That is a much higher bar than the marketing decks suggest, and a surprising number of LSTs simply do not clear it.
Consider the design choices that quietly determine whether a token survives a liquidation cascade or amplifies one. Rebasing tokens distribute yield by adjusting balances, which complicates accounting in any lending contract that was not purpose-built to handle them. Non-rebasing wrappers, such as wstETH, accrue yield through an exchange rate against the underlying and are mechanically easier to liquidate against. Withdrawal mechanics matter even more. Some protocols offer direct, on-protocol redemption with predictable wait times. Others rely on bonded queues. Others still are exit-by-swap only, which means the LST is only ever as liquid as the deepest pool standing in the worst hour of the day. The deeper question is whether the lending markets quoting LTVs against these tokens have actually priced that asymmetry, or whether they are extrapolating calm-market behavior into a regime that has not yet arrived.
Liquid restaking tokens compound the problem. LRTs inherit every risk of the underlying LST and then layer on a second tier of slashing exposure tied to whatever actively validated services the operator has opted into. The yield enhancement is real. So is the correlation risk. A single misbehaving AVS can ripple through a basket of restaked positions in ways that even the protocols themselves struggle to model in advance. The cautious move, which a handful of money markets have already made, is to wall LRTs off in isolation modes with conservative caps. The reckless move is to treat them as full citizens of the collateral universe and hope the oracle holds up.
What is striking is how much of this comes down to risk parameters that lending protocols set unilaterally and rarely revisit. Supply caps, borrow caps, liquidation thresholds, and oracle source choices do more to determine real depositor safety than the headline APY ever will. An LST with mediocre redemption mechanics can be a reasonable collateral asset inside tight caps and conservative LTVs. A high-quality LST can become dangerous if listed with permissive parameters and a shallow oracle source. The token is not the whole story. The wrapping around it is.
None of this is to suggest LSTs are unsuitable for lending markets. They are arguably the most productive collateral the on-chain economy has produced. But productivity is not the same as predictability, and lending is a business that pays for predictability. Depositors who treat every yield-bearing wrapper as a drop-in replacement for ETH are quietly underwriting a risk they have not been compensated for. The protocols that survive the next stress test will be the ones that admitted, before they had to, that the collateral list is a curatorial responsibility rather than a marketing surface.