The risks of liquid restaking tokens

1. TL;DR

A liquid restaking token (LRT) tokenizes a restaked position rather than a plain staked one, so its holders inherit ordinary staking risk, the slashing risk of every service the underlying capital is opted into, and the redemption risk of a still-young protocol — all stacked on top of each other. The clearest evidence this is not a free lunch arrived on 23–24 April 2024, when Renzo’s ezETH depegged from about $3,049 to as low as $700 in under an hour, triggering over $56 million in liquidations. Solana has no direct equivalent market yet; its closest analogue is Jito’s Tip Router routing MEV tips to JitoSOL-backed vaults, without a comparable liquid, tradeable “restaked receipt” token in wide circulation.

2. Explain it simply

Analogy

A plain staking receipt is like a single certificate of deposit from one bank. A restaked receipt is like a fund-of-funds: it bundles certificates of deposit from several different banks, each with its own rules, into one tradeable share. The share is more convenient to trade, but if any one of the underlying banks fails, the fund’s value drops — and a buyer of the share often cannot tell, just from the price, exactly which underlying bank is causing the trouble.

ELI5

When you lock up your money to help run a network, you sometimes get a stand-in token so you can still use your money elsewhere. A restaked stand-in token goes one step further: your locked money isn’t just helping one network — it’s also volunteering for extra jobs at other services, each with its own way of punishing mistakes. The stand-in token you’re holding bundles all of that together into one number. The problem is that the token can only be exchanged back for your real money as fast as the slowest of those extra jobs allows, and if any of those extra jobs go wrong, the token’s price can drop suddenly, well before you get a chance to cash out at the price you expected.

Step-by-step walkthrough

Scenario: a user, Sam, holds 10 ETH worth of a staking receipt token and deposits it into a restaking protocol to receive a liquid restaked receipt.

Deposit (before: 10 ETH-equivalent staking receipt → after: 10 ETH-equivalent restaked receipt token). Sam tenders the staking receipt to the restaking protocol and receives a new token representing a claim on a basket of restaked positions across several services and operators.
The protocol allocates Sam’s capital (before: undifferentiated deposit → after: capital split across multiple services with different rules). The protocol’s managers decide which services to opt into and which operators to use — a decision Sam does not individually control.
A distribution announcement disappoints the market (before: token trading near its backing value → after: many holders want out at once). Holders start selling the restaked receipt on the open market rather than waiting for the protocol’s redemption process, because direct redemption is not available or is slow.
Selling outpaces available buyers (before: token near par value → after: token trading at a steep discount). Because the token cannot be redeemed instantly for the same reason it exists — capital is tied up across several services with their own unlock schedules — a wave of selling pushes its market price well below what it is actually backed by, even though no service has actually been penalized.

Common misconceptions

Myth: A restaked receipt token dropping in price means the underlying services were slashed. Reality: In the largest real-world case to date, the depeg was driven by a market reaction to an unpopular token-distribution announcement and an inability to redeem directly, not by any actual slashing event (DLNews, 2024-04, secondary).
Myth: Liquid restaking tokens and plain staking receipt tokens carry roughly the same risk. Reality: A comparative analysis finds restaked positions differ from plain staked positions along duration, default, and portfolio-construction dimensions — restaked withdrawals are “rate-limited by the AVS itself, the L1 protocol, and the restaking protocol” simultaneously, not just one of the three (Neuder & Chitra, ethresear.ch 18799, 2024-02).
Myth: More services opted into means more yield with proportionally more safety. Reality: Portfolio construction for a basket of restaked positions “requires managing multiple AVSs, node operators & their mapping to AVSs, different yield rates/variabilities/denominations, and various risk profiles,” a complexity the same authors say “likely far exceeds the design space” of an ordinary staking-token basket (Neuder & Chitra, 2024-02).
Myth: This is only an Ethereum problem. Reality: The same structural risk — MEV or restaking rewards flowing through an extra distribution layer before reaching token holders — exists wherever a chain routes MEV back to stakers through an intermediary, including Solana’s Tip Router.

If you only remember one thing

Every extra layer a token wraps around your capital — staked, then restaked, then tokenized again — adds a rulebook that can move the token’s price without your capital actually being lost yet, and vice versa.

3. How it works

A five-attribute risk lens

Neuder & Chitra’s comparative framework analyzes restaked positions and their tokenized baskets along five attributes, contrasting them with more familiar non-fungible instruments (sovereign bonds, plain staking positions) and their fungible basket equivalents (bond funds, liquid staking tokens):

Liquidity & leverage. An individual restaked position is illiquid and hard to borrow against directly; a liquid restaking token aims to be fungible and usable as collateral, but “due to their infancy, LRTs are yet to establish sustainable liquidity” (2024-02).
Yield. A restaked position’s interest rate is service-defined and may be “denominated” in a token other than the base asset; an LRT aggregates yield “across many AVSs and node operators,” and unlike a liquid staking token, “the interest rate and yield denomination underlying LRTs are AVS-dependent.”
Duration. Withdrawing requires waiting on the slower of three separate rate limits stacked together: the restaking protocol’s own escrow (EigenLayer specified a 7-day escrow at the time of writing), the individual service’s rate limit, and the base layer’s own exit queue.
Default. A restaked position can be slashed by both the base layer and the service it opted into; the tokenized basket spreads this risk across many counterparties, but a slashing event at any one of them “could cause a ‘race for the exit’ scenario” that drives the token to trade at a steep discount even before losses are finalized.
Portfolio construction. Building a credible LRT basket means choosing which services to opt into, which operators serve them, and how to weight the resulting risk — a design space Neuder & Chitra describe as significantly larger than the analogous choice for a plain liquid staking token (which only has to choose node operators, not node operators and services).

MEV distribution as its own restaking-like layer

On Solana, Jito’s Tip Router applies a structurally similar pattern without calling itself an LRT. Each epoch, validators (operators, in restaking terms) that ran the Jito block-building auction generate MEV tips; the Tip Router — itself deployed as a node consensus network — computes a Merkle tree of tip amounts per validator, aggregates these into a “meta Merkle root” once operators reach two-thirds consensus, and only then releases the on-chain reward payment to JitoSOL vault depositors (Jito docs, n.d.). This is not a liquid restaking token in the Ethereum sense — there is no separate tradeable “restaked JitoSOL” — but it shares the underlying risk shape: a reward stream that depends on an off-chain computation and an on-chain consensus step before it reaches a holder, with a redemption value (JitoSOL’s exchange rate) that only updates once that pipeline completes.

4. Worked numeric example

Ethereum: the ezETH depeg, quantified (matching §2). Before 23 April 2024, ezETH traded near its backing value of roughly $3,049 per token. Renzo announced the distribution terms for its REZ governance token: ezETH holders who had held the token for roughly four months would receive only 5% of REZ’s total supply, versus 2.5% allocated to Binance Launchpool participants who had only committed capital for seven days. Because Renzo had not enabled direct redemption of ezETH for its backing ETH, disappointed holders had only one exit: selling ezETH on the open market. Within under an hour, the price fell 79%, bottoming near $700 — a discount of more than $2,300 per token versus backing value, despite no restaked service having actually been slashed. The forced selling cascaded into on-chain lending markets using ezETH as collateral: Gearbox saw 115 users liquidated for roughly 10,650 ezETH (about $33 million), and Morpho saw 146 users liquidated for roughly $23 million, for a combined total above $56 million across more than 250 users (DLNews, 2024-04, secondary; Protos, 2024-04, secondary).

Solana Tip Router pipeline (illustrative). A JitoSOL vault holds 50,000 SOL-equivalent stake across operators serving the Tip Router NCN. In a given epoch, the network’s validators collectively earn 400 SOL in MEV tips. Operators compute per-validator Merkle roots, a node operator client aggregates these into one meta Merkle root, and a permissionless cranker uploads it once two-thirds of weighted operator votes agree. Only after that consensus step does the 400 SOL become claimable and get folded into the vault’s exchange rate — moving it, illustratively, from 1.041 to 1.049 SOL per JitoSOL-equivalent unit. A holder cannot receive that reward faster than the pipeline resolves, and if operators fail to reach consensus in a given epoch, the reward for that epoch is delayed rather than lost outright.

5. Where it’s used

Ethereum

Renzo (ezETH), Ether.fi (eETH), Puffer (pufETH), Kelp (rsETH) — the major liquid restaking tokens built on top of EigenLayer, each with its own basket of AVSs, operators, and redemption design.
EigenLayer’s native restaking path — stakers can restake directly (via an EigenPod) rather than through a liquid token, avoiding LRT-specific risk while still taking on restaking’s slashing risk; see /staking/restaking/.

Solana

n/a for a widely circulating, tradeable liquid restaking token — no source in this catalogue describes a live, broadly adopted LRT market on Solana comparable to Ethereum’s.
Jito Tip Router — the closest structural analogue: an NCN that computes and distributes MEV-tip rewards to JitoSOL vault depositors through a Merkle-proof pipeline, without a separate secondary “restaked” token layer (Jito docs, n.d.). See /staking/restaking/ for the underlying NCN/vault/operator model.
Jito (Re)staking vaults — mint liquid vault receipt tokens (VRTs) for capital delegated to NCNs generally, which is the primitive an eventual Solana LRT market would likely be built on, but as of this catalogue’s sources, no dedicated liquid restaking token product built on top of it has reached comparable scale to Ethereum’s LRTs.

6. Risks, attacks, and incidents

Real incident: Renzo ezETH depeg, 23–24 April 2024 (as of 2024-04). Detailed in §4 above: a roughly 79% price drop within under an hour, over $56 million in DeFi liquidations across more than 250 users, root cause an unpopular token-distribution announcement combined with the absence of direct redemption (DLNews, 2024-04; Protos, 2024-04; both secondary).
Precedent from plain liquid staking: the stETH depeg, June 2022. Neuder & Chitra explicitly draw the parallel: liquid staking tokens “first became popularized as a way to realize liquidity or take leverage on locked beacon chain positions that were not withdrawable before the Merge,” and stETH’s own 2022 depeg (see /staking/liquid-staking-tokens/) is the direct historical antecedent for the LRT-specific stress events that followed two years later (ethresear.ch 18799, 2024-02).
Speculative capital arriving ahead of the risk it is pricing. LRT total value locked grew from about $250 million to over $4 billion in the two months before the paper’s February 2024 publication, “preceding the launch of restaked services” — meaning most of that capital was speculating on future yield from services that had not yet gone live and therefore had no track record to price risk against (Neuder & Chitra, 2024-02).
Compounded default risk. Because “one/some of the constituent node operators getting slashed on the AVS or on the L1 could cause a default on the LRT depending on the size of the slashing,” an LRT holder is exposed to every operator and every service in the basket simultaneously, not just the base staking layer (Neuder & Chitra, 2024-02).
Redemption design still evolving. The same authors note plainly that “LRT withdrawal designs are still evolving,” meaning the specific rate-limiting rules that determine how fast a holder can exit are protocol-specific, non-standardized, and subject to change (2024-02).

7. Open problems

No standardized default/discount mechanism. Unlike Lido’s “bunker mode” for plain liquid staking (see /staking/liquid-staking-tokens/), Neuder & Chitra note LRTs lack an established analogue for pausing or managing withdrawals under stress, since the space is too young to have converged on one (2024-02).
Portfolio construction standards. The paper explicitly frames LRT portfolio construction — which AVSs, which operators, what weights — as an open design problem with a much larger surface area than plain liquid-staking-token curation (2024-02).
Whether Solana needs a dedicated LRT market at all. Because Jito’s Tip Router already routes MEV rewards to JitoSOL holders without a second token layer, it is an open question whether a separate, tradeable “restaked JitoSOL” would add meaningful liquidity value or would mainly reproduce Ethereum’s LRT-specific risks without Ethereum’s AVS ecosystem to justify them.
How regulators will treat LRTs. The paper’s own framing — comparing LRTs to sovereign bond funds — implicitly raises a securities-law question the authors do not resolve, since a basket that aggregates yield from many discretionary, actively-managed allocations looks more like a managed fund than a passive staking receipt.

8. Ethereum vs Solana

Aspect	Ethereum	Solana
Liquid restaking token market	Multiple large tokens (ezETH, eETH, pufETH, rsETH)	n/a — no comparable market in sources
MEV/reward distribution to restakers	Per-AVS reward design, varies by protocol	Jito Tip Router: Merkle-proof-based, epoch-by-epoch
Real stress event	ezETH depeg, April 2024 (~79% drop, $56M+ liquidated)	None recorded in sources
Redemption design	Protocol-specific, several without direct redemption	JitoSOL vault exchange-rate model (see /staking/liquid-staking-tokens/)
Underlying restaking primitive	EigenLayer / Symbiotic (see /staking/restaking/)	Jito (Re)staking NCNs

Ethereum’s LRT market formed because a mature restaking ecosystem (EigenLayer) and a large existing liquid-staking-token base gave holders an obvious asset to bundle and re-tokenize, and it grew fast enough to produce a real, quantifiable depeg within months. Solana’s restaking primitive is younger, more tightly coupled to a single dominant client (Jito), and has so far expressed its MEV-distribution role through an NCN with a direct exchange-rate mechanism rather than a separate speculative LRT market — meaning the specific compounding risk this page describes has not yet materialized there, but also has not yet been tested at scale.

9. Reference doc

The reference

The risks of LRTs (subtitled “or The Commonalities Between Staking, Restaking, and Sovereign Bonds”) — Mike Neuder and Tarun Chitra, 26 February 2024. ethresear.ch/t/18799

Summary of the reference

The post opens with a striking statistic — liquid restaking token TVL grew from about $250 million to over $4 billion in the two months before publication — and states its two goals plainly: help readers understand LRTs by comparison with more familiar assets, and demonstrate that LRTs are risky, with “no free lunch” for the additional yield they offer. The authors build a six-asset case study split into two groups of three: non-fungible yielding assets (sovereign bonds, L1 staking positions, AVS restaking positions) and fungible baskets of those same assets (bond funds, liquid staking tokens, liquid restaking tokens). They walk through five attributes for each — liquidity/leverage, yield, duration, default, and portfolio construction — building the comparison table this page’s §3 summarizes.

Several sections repay close reading. On liquidity, the authors note LSTs are already viewed by lending markets as lower-risk collateral, while LRTs are “yet to establish sustainable liquidity” given their infancy. On yield, they highlight that LRT yield can be “AVS-dependent” in both rate and denomination — a restaked position might pay out in a token entirely separate from ETH, a feature with no parallel in plain liquid staking. On duration, the authors flag that native and liquid restaked ETH may face different rate limits from the restaking protocol itself, and that EigenLayer’s withdrawal process at the time added a 7-day escrow on top of both the AVS’s and Ethereum’s own exit queues. On default, the piece draws a direct historical analogy to the 2023 collapse of Silicon Valley Bank as a “duration mismatch default” for a traditional bond fund, arguing LSTs and LRTs are exposed to a structurally similar mismatch between long-lived underlying positions and demands for instant liquidity. The paper closes candidly acknowledging its own scope: it is a comparative lens, not a formal risk model, and explicitly states its analogies are imperfect but useful.

Key quotes

“The TVL of liquid restaking tokens (LRTs) grew from $250mm to over $4 billion in the past two months.” (tl;dr)

“Highlight that LRTs are risky. There is no free lunch; risk accompanies the additional rewards for holding LRTs.” (tl;dr)

“LRTs differ from LSTs because the withdrawals depend on the AVS details and the restaking protocol in addition to the L1 protocol.” (§ Section 3: Duration)

“Constructing an LRT portfolio requires managing multiple AVSs, node operators & their mapping to AVSs, different yield rates/variabilities/denominations, and various risk profiles. This complexity likely far exceeds the design space of LST constructions.” (§ Section 5: Portfolio construction)

How to read the original

Background needed: what a liquid staking token is (see /staking/liquid-staking-tokens/) and what an AVS is (see /staking/restaking/). The bond-market vocabulary (sovereign bonds, bond funds, duration, default) is explained inline, so no traditional-finance background is strictly required, though it helps. Skip the embedded historical-timeline figure description if reading the text version — it mainly restates dates already given in prose. The hardest part is Section 4 (Default): the Silicon Valley Bank analogy is doing real conceptual work (a duration mismatch, not a credit default, is the failure mode being described), and it is worth pausing to map each element of SVB’s balance sheet onto the LST/LRT case before moving on.

What changed since

The paper’s central warning was validated within two months by the ezETH depeg (April 2024), detailed in §4 and §6 above — a real “no free lunch” event rather than a theoretical one.
EigenLayer’s slashing mechanism, only a design proposal at the time of writing, was later redesigned around “unique stake” and operator sets (see /staking/restaking/, §9’s “What changed since”), changing some of the default-risk mechanics this paper described in the abstract.
LRT withdrawal design has continued to evolve protocol-by-protocol, with some issuers (unlike Renzo in April 2024) later adding more direct redemption paths — though sources in this catalogue do not document a comprehensive, dated survey of which protocols changed what.

Secondary references

Renzo ezETH depeg reporting — DLNews and Protos, April 2024 (secondary) — read for the concrete numbers behind the paper’s abstract warning.
barnabe, “Unbundling staking: Towards rainbow staking” (ethresear.ch 18683, 2024-02) — read for the protocol-design response to LST/LRT concentration risk; see /staking/liquid-staking-tokens/.
Jito docs, “Jito Tip Router” overview (n.d.) — read for a live, non-Ethereum example of the MEV-distribution-as-restaking-like-layer pattern described in §3.
Hasu & Konstantopoulos, “On Staking Pools and Staking Derivatives” (Paradigm, 2021-04) — read for the LST-era precedent this paper explicitly builds on; see /staking/liquid-staking-tokens/.

10. Sources

The risks of LRTs — Mike Neuder, Tarun Chitra — 2024-02-26 — https://ethresear.ch/t/the-risks-of-lrts/18799
Jito Tip Router overview — Jito docs — n.d. — https://www.jito.network/docs/tiprouter/overview/
Jito (Re)staking overview — Jito docs — n.d. — https://www.jito.network/docs/restaking/overview/
Renzo’s ezETH loses Ether peg, drops 79% in under one hour — DLNews (secondary) — 2024-04 — https://www.dlnews.com/articles/defi/renzos-ezeth-loses-ether-peg-drops-79-in-under-one-hour/
Depeg of $3B restaking token ezETH causes over $60M in DeFi liquidations — Protos (secondary) — 2024-04 — https://protos.com/depeg-of-3b-restaking-token-ezeth-causes-over-60m-in-defi-liquidations/
Unbundling staking: Towards rainbow staking — barnabe — 2024-02-15 — https://ethresear.ch/t/unbundling-staking-towards-rainbow-staking/18683
On Staking Pools and Staking Derivatives — Hasu, Georgios Konstantopoulos (Paradigm) — 2021-04-23 — https://www.paradigm.xyz/writing/on-staking-pools-and-staking-derivatives