An exchange-rate token holds the holder's balance fixed and moves the price instead. Rewards accrue into what each unit redeems for, so ten units stay ten units while the ETH behind them grows. rETH and wstETH are two examples in wide use. Because the balance never changes, the token behaves like any ordinary ERC-20 to a lending market, an automated market maker or a bridge, which is why almost every DeFi integration of staked ETH runs through this form rather than a rebasing one.
The redemption rate is published by an oracle, not observed by the contract. Whoever can move that rate can overstate what every unit is worth, so the update path and its governance are the real due diligence surface, not the token code.
Scroll to see the full diagram
The rate is a ratio, and it is published, not derived
Rocket Pool documents the rETH rate as the total rETH supply set against the ETH standing behind it: ETH staked in validators, plus the rETH contract balance, plus rETH's share of priority fees and MEV rewards. The ratio is refreshed roughly every 24 hours from the beacon chain rewards node operators have earned.1 Lido's wrapped token takes the same shape from the other direction, described in its own contract documentation as a value-accruing wrapper whose balance is not changed by an oracle report while its value in stETH is.2
The word doing the work in both descriptions is reported. A token contract on the execution layer cannot see validator balances on the consensus layer. Something has to carry that number across, and in Rocket Pool's case it is an oracle DAO whose nodes relay beacon chain information such as total staking rewards and validator exit status back to the execution layer so the contracts can function.3
So the accounting is clean and the trust is not eliminated, it is relocated. A rebasing token moves supply on an oracle report and an exchange-rate token moves price on the same report. Both depend on the same reporter. The exchange-rate form simply puts the dependency somewhere fewer integrators think to look.
Worked: what ten units become
Rocket Pool's own illustration is the cleanest in the category. Stake at the point where one ETH buys one rETH and deposit ten ETH, so you hold ten rETH. Time passes and rewards accrue. Suppose 128 ETH had been staked through the protocol while the sum of validator balances on the beacon chain reached 160 ETH. One ETH is now worth 128 divided by 160, or 0.8 rETH, and one rETH is worth 160 divided by 128, or 1.25 ETH. The ten rETH redeem for 12.5 ETH.1
Your token count never moved. No transfer event fired. Nothing in your wallet's balance history records the reward, and every accounting system reading balances alone sees a flat position that happens to be worth more.
Run the same arithmetic with a loss and the mechanism reads the same way in reverse. If penalties reduce the backing to 120 ETH against the same 128 rETH, one rETH is worth 0.9375 ETH and the ten units redeem for 9.375. Rocket Pool's documentation frames the ratio as effectively rising over time because rewards keep accruing.1 That is an observation about the inputs, not an invariant the contract enforces, and any product built on the assumption that the rate only moves one way has written a promise the code does not make.
Why lending markets prefer a balance that stays still
The academic systematization of liquid staking tokens gives the mechanical reason. Rebase tokens are largely unsupported on the major automated market makers, and they are not used as lending collateral, because the staking rewards would have to be divided between borrower and lender. With a reward-bearing exchange-rate token, all of the accrual lands on the collateral provider, and the paper records that reward-based tokens consequently became a dominant collateral asset in DeFi lending.4
Lido states the same conclusion as a product decision. wstETH exists mainly as a compatibility layer for integrating stETH into protocols that do not support rebasable tokens, and particularly for bridges to layer 2 networks.2 The protocol that built the largest rebasing token also ships the wrapper that removes the rebasing, and every serious integration uses the wrapper.
Design consequence for a founder choosing between the two. The venue list you need is the specification, and the token model is downstream of it. If lending collateral, layer 2 deployment or generic AMM liquidity is on that list, the exchange-rate form is the one that gets there without a wrapper. The rebasing token entry covers what you give up in exchange.
The tax and accounting question the fixed balance changes
IRS Revenue Ruling 2023-14 holds that the fair market value of validation rewards is includible in gross income in the taxable year the taxpayer gains dominion and control over them.5 A design that credits a balance daily and a design where value accrues into a redemption rate present that standard very differently, because in the second case nothing is credited to anybody until a disposal or redemption occurs.
We are not tax advisers and this is not tax advice. The point is structural: the token model is upstream of the answer, most teams treat it as downstream, and the question reaches counsel after the contract is deployed rather than before. That ordering is the mistake, not the model.
It also affects the buyer. An institutional allocator with an accounting policy that cannot handle a balance changing without a transaction will not hold the rebasing version at all. That is a distribution constraint, and it is decided at the same moment as the token model.
The three questions a buyer should ask first
Three items, in order. Name the reporter: who publishes the rate, under what quorum, on what cadence, and what the contract does when a report is late or absurd. Name the bound: whether the rate can fall, by how much in one report, and whether a circuit breaker exists. Then publish the redemption path with its queue, because a rate is a claim on capital that has to come back through the base chain's exit mechanics, and the liquid staking token entry covers what happens when that queue lengthens.
The failure we see is a rate that is correct and a redemption path that is theoretical. A secondary-market price and a redemption value converge only while somebody can profitably move between them, and that arbitrage runs through the withdrawal queue.
Underneath all of it, the same anchor. The exchange-rate token is packaging for a staking business. If the fee and issuance flow beneath it is thin, better packaging does not fix that. This page is reference material for design work, not investment advice and not a recommendation about any asset.
Common questions
What is the difference between an exchange-rate token and a rebasing token?
Where the reward lands. A rebasing token grows the holder's balance and keeps the price near the underlying. An exchange-rate token keeps the balance fixed and grows what each unit redeems for, so ten rETH stay ten rETH while the ETH behind them rises.1 The economics match. The ledgers do not, and the fixed balance is what lets lending markets, AMMs and bridges handle the token without special support.4
How is the rETH exchange rate calculated?
Rocket Pool sets it as total rETH supply against the ETH standing behind it: staked ETH, the rETH contract balance, and rETH's share of priority fees and MEV rewards.1 The ratio is refreshed roughly every 24 hours from realised beacon chain rewards. Because the execution layer cannot see validator balances directly, an oracle DAO relays that information to the contracts.3
Can an exchange-rate token lose value against ETH?
Yes. The rate is the ratio of supply to the ETH backing it, and penalties reduce that backing the same way rewards increase it. Rocket Pool describes the ratio as effectively rising over time because rewards keep accruing, which is an observation about inputs rather than a property the contract enforces.1 Separately, the market price can trade below the redemption rate when the withdrawal queue prevents arbitrage from closing the gap.
Why do DeFi protocols prefer exchange-rate tokens as collateral?
Because a fixed balance does not break their accounting. Rebase tokens are largely unsupported on major AMMs and are not used as lending collateral, since the staking rewards would have to be split between borrower and lender. With an exchange-rate token the entire accrual lands on the collateral provider, and reward-bearing tokens became a dominant collateral asset in DeFi lending as a result.4
See LST and LRT Tokenomics Design for how this applies in practice.
Sources
- Staking Overview
Rocket Pool Docs, 2026
The rETH to ETH ratio definition, its roughly 24 hour update cadence, and the worked 128 ETH to 160 ETH example producing a 1.25 ETH redemption per rETH. Read 3 August 2026. - wstETH contract documentation
Lido Docs, 2026
wstETH as a value-accruing wrapper whose balance is unchanged by oracle reports while its value in stETH moves, and its stated purpose as a compatibility layer for protocols and bridges that cannot handle rebasing. - Glossary
Rocket Pool Docs, 2026
rETH defined as a token whose value increases relative to ETH as staking rewards accrue, and the oracle node role relaying beacon chain data such as total staking rewards and validator exit status to the execution layer. Read 3 August 2026. - SoK: Liquid Staking Tokens (LSTs) and Emerging Trends in Restaking
arXiv preprint 2404.00644v3, 2024
The reward LST taxonomy, the reason rebase tokens are unsupported as lending collateral, and the finding that reward-bearing tokens became a dominant collateral asset in DeFi lending. - Revenue Ruling 2023-14
U.S. Internal Revenue Service, 2023
Validation rewards are includible in gross income in the taxable year the taxpayer gains dominion and control over them.
Last reviewed 2026-08
Know the terms but not sure how they apply to your project? That is what an engagement is for. We design, document, and stress-test the whole token economy inside the Tokenomics Data Room.
100+ projects advised. Complete tokenomics in 4 to 6 weeks.