IOSG researcher Mario Chow has published a quantitative review of EIP-8363, a proposal that would burn an increasing share of validator rewards as Ethereum’s staking ratio rises, reaching a full burn of that portion when 50% of ETH supply is staked. The study is built from onchain data and the EIP text, with data updated through Aug. 24, 2026. Chow wrote that the independently built model tracks public third-party data within 2%.
The headline conclusion is narrow but consequential. In the paper’s framing, Ethereum’s fee-burn mechanism no longer meaningfully regulates supply, leaving issuance as the main remaining lever the protocol can still pull. At current staking levels, EIP-8363 would cut issuance roughly in half rather than take it to zero. The design is also self-limiting: under what the author describes as reasonable staking hurdle rates, the system would settle with 26% to 34% of supply staked and annual issuance running at 0.3% to 0.5%. The report also says it could not detect a meaningful relationship between the yield being cut and ETH price behavior.
EIP-1559 burn has faded sharply, the report argues
The paper starts with EIP-1559. It says the mechanism burned 1.48 million ETH in 2022, but points out that what gets burned is the base fee, which is essentially a congestion price. Once blobs moved rollup data away from L1 and the gas limit was raised, congestion eased and the base fee collapsed.
According to the model, gas usage doubled since 2022 while the average base fee fell 96%, driving a 98% drop in burned ETH. Over the past 12 months, Ethereum burned 25,660 ETH in total. Using the most recent 30-day run rate, the paper says current burn is about 39 ETH per day, or roughly 14,300 ETH annualized.
Set against annual gross issuance of about 1.08 million ETH, that means burn now offsets only 2.4% of new supply. In Chow’s reading, the old “ultrasound money” story no longer describes Ethereum’s actual supply dynamics. The report adds that this is not a demand story but a pricing story: L1 gas usage rose from 3.4 billion to 6.7 billion units per month, while the average base fee fell from 4.00 gwei to 0.17 gwei.
Net supply since the Merge shifts the debate
Looking at issuance and burn together, the report says the picture is starker. Issuance never stopped growing, while the offsetting burn component has nearly disappeared.
Across the 47 months since the Merge, only 13 months were deflationary, and the last of those was March 2024, according to the paper. ETH has now been inflationary for 28 straight months, with the annualized inflation rate rising by roughly threefold over that span, from +0.26% to +0.87%. The report says that happened not because issuance materially surged, but because the offset faded away. Since 2024, issuance itself is said to be up only 4%.
That changes the terms of the argument around EIP-8363. Rather than a broad philosophical trade between staking yield and monetary scarcity, the study casts the issue as a forced one: with demand-driven burn no longer doing much, Ethereum’s supply policy has to run through the issuance curve.
How EIP-8363 would work
The mechanism described in the paper is straightforward. Validators would first earn their full rewards as usual, and the protocol would then burn a fraction b of those rewards.
One technical detail matters. The burn is calculated from the validator’s theoretical full reward, not from what the validator actually received. Chow says that avoids punishing offline validators twice. If a validator misses duties, it already loses those rewards. The proposal does not add a second penalty by shrinking the burn base to actual rewards.
The report also notes an extreme-condition safeguard: if Ethereum enters inactivity leak, the burn tied to attestation rewards pauses. And the proposal touches only consensus-layer rewards. MEV and priority fees earned by node operators are left intact.
Two common misconceptions in the debate
It does not take issuance to zero at today’s staking level
The first misconception, the paper says, is that Ethereum issuance would be cut straight to zero. That is not what the math shows under current conditions.
With about 42.2 million ETH now staked, the burn fraction b is 58.6%, according to the model. To push net issuance all the way to zero, staked ETH would have to climb to 60.25 million ETH, 43% above the current level. The study therefore says the more accurate near-term description is that EIP-8363 would cut issuance by about half.
The proposal includes an 18-month transition
The second misconception is the idea of an immediate yield shock that would puncture DeFi on day one. The EIP includes an 18-month transition period, the paper says.
At launch, the base reward factor used in reward calculations would be doubled to 128 from 64. Chow says that change largely offsets the initial effect of the 58.6% burn. As a result, net issuance on the first day after activation would still sit at about 83% of the current level. Over the next 18 months, the parameter would gradually move back to 64, and issuance would slide toward 41% of the current level.
In other words, the decline in rewards would be spread over a year and a half rather than arriving all at once.
Ethereum’s current issuance baseline
The baseline section of the report says new ETH now comes almost entirely from staking rewards. After the Merge, consensus-layer payments to validators are the only source of new issuance.
The paper breaks that flow into three duty categories with fixed protocol weights: attestation rewards account for 54/64, or 84.4%, equal to 911,672 ETH per year; block proposal rewards account for 8/64, or 12.5%, equal to 135,063 ETH; and sync committee rewards account for 2/64, or 3.1%, equal to 33,766 ETH.
Issuance is modeled as I(S) = 940.9 · √(S/32) ETH per year. At S = 42.2 million ETH, consensus-layer APR comes to 2.560%. The report also measures priority fees over the first 23 days of August at 2,623 ETH, or about 41,500 ETH annualized, equal to 0.098% on the staked base. Together that adds up to 2.658%, almost identical to the published 2.66% figure.
Using those estimates, the paper says at least 96% of validator income comes from issuance and at most 4% from fees. It notes that proposer payments in MEV-boost above direct pass-through priority fees are not captured, so the fee share is a floor, not a full count. Even so, issuance still dominates validator revenue.
Static impact at current staking levels
Holding behavior constant, the model gives three central numbers. Issuance falls 58.6%. Staking APR falls 56.4%. Dilution removed from the system totals 633,000 ETH per year.
At the price used in the paper, that equals about $1.55 billion annually, or 0.53% of ETH market capitalization.
The report also traces the full post-transition curve. Issuance peaks when staked ETH is near 25 million, at about 0.505% of supply, then declines as staking rises further, hitting zero at a 50% staking ratio. Chow says that matches the EIP’s own description.
Equilibrium is the key number
The paper argues that the real question is not the static cut but the response from stakers. If returns drop below what they require, some will exit. That exit lifts gross APR and lowers the burn fraction b, pushing the system toward a new fixed point.
Using two hurdle-rate examples, the report says the EIP-8363 reward curve intersects a 2% required return at about 31.2 million ETH staked, and intersects a 1.25% threshold at about 40.9 million ETH.
That leads the study to reject both extremes in the public argument. “Issuance goes to zero” would only hold if marginal stakers were willing to work for roughly 0.5%. Under more plausible hurdle rates, ETH still inflates at 0.3% to 0.5% a year. On the other side, “staking collapses” is also presented as an exaggeration. With a 2% threshold, staking stabilizes around 26%, below today’s roughly 35% but close to the level seen through 2024.
Chow highlights this as the proposal’s least discussed property: it limits itself by design.
No detectable link between staking yield and ETH price, the paper says
The report devotes a full section to the market question behind the policy debate: does staking yield explain ETH’s price?
It first warns that staking ratio and issuance yield are mechanically related. Because the protocol reward pool scales with the square root of staked balance, each ETH’s issuance yield has a closed-form expression: issuance(S) = 940.9 · √(S/32) ETH per year, and APR(S) = issuance(S)/S = 166.28 / √S. More staking means the same reward pool is shared across more coins.
The meaningful free variable is the gap between published yield and formula-implied issuance yield, which is fee income. That gap was about 1.34 percentage points in 2022 and is about 0.10 percentage points now, according to the paper.
Over the 43-month window from January 2023 to July 2026, the level regression between month-end ETH price and staking APR looks significant at p = 0.006. But the Durbin-Watson statistic is 0.40, which the report says signals severe serial correlation in residuals and a textbook spurious regression between trending series. Chow keeps that chart in the paper as a warning, not as evidence.
After differencing the data to remove the trend, the relationship disappears. The paper reports p = 0.73, R² = 0.003, and Durbin-Watson = 1.75. Its read is blunt: the 95% confidence interval comfortably crosses zero in both directions, so the data cannot even pin down the sign of the effect, let alone its size.
The 12-month rolling correlation also crosses zero repeatedly and spends most of the time in a range statistically indistinguishable from zero. Across that same period, staking yield fell from 3.98% to 2.50%, ETH/BTC fell 57%, and the monthly correlation between yield changes and ETH returns was -0.05.
The study’s conclusion is that staking yield did not support price, and yield compression did not trigger a measurable decline. The report adds one caution: its staking series is reconstructed from onchain flows and runs about 5% above published figures, so the direction and shape are treated as reliable, while the absolute level is not exact.
Supply growth has a somewhat stronger, but still insignificant, relationship
The paper also tests whether net supply growth explains ETH returns better than staking yield does.
Using the same sample window, the estimated slope is -6.9, meaning a 1 percentage point rise in annualized supply growth maps to a 6.9 percentage point drop in monthly return. But the relationship is not statistically significant: p = 0.18, R² = 0.044, Durbin-Watson = 1.82, and the 95% interval runs from -17.1 to +3.4.
Chow’s interpretation cuts both ways. The result cannot be used as proof that reducing supply growth will lift price. At the same time, it is about fifteen times stronger than the yield relationship on R² and has the sign theory would predict. If either variable matters at the margin, the data point more toward supply than toward yield. That is the trade EIP-8363 actually makes.
How much of onchain Ethereum depends on ETH yield
LSTs are large, but the effects are uneven
The report says Lido alone is equivalent to 48% of Ethereum’s $48.5 billion DeFi TVL, so any claim that DeFi would be untouched has to clear that number first.
Its direct revenue hit is easier to quantify. The paper says Lido processes about $602 million in staking rewards each year and takes a 10% fee, or roughly $60 million annually. A 58.6% cut in issuance means 633,000 ETH less reward issuance per year; with Lido holding a 22.8% share, that would reduce its fee income by about $35 million a year, roughly half of that line of business.
Still, the paper says that does not amount to a systemic break for Ethereum itself. And as long as wstETH retains a positive yield, it remains better collateral than WETH for borrowers who want ETH exposure.
LSTs as collateral are where dependence is more visible
Across the three largest Ethereum lending markets, the report says $10.63 billion out of $31.10 billion in collateral, or 34.2%, consists of staking-yield derivatives.
- SparkLend: 66.9%
- Aave V3: 38.7%
- Morpho Blue: 10.4%
SparkLend is singled out as the clearest concentration case, with two-thirds of collateral in wstETH.
The ETF channel exists, but is small today
The study also examines a widely repeated objection: lower yields could weaken institutional demand because staking-enabled ETH ETFs market yield to allocators who cannot access it directly. Chow says that channel is real, but small at the moment.
Products that explicitly sell the yield angle account for 5.4% of ETF assets, 0.53% of all staked ETH, and 0.19% of total ETH supply, according to the paper. It adds that BlackRock’s non-staking ETH product is ten times larger. The report’s takeaway is that whatever is drawing institutional money into ETH, staking yield is not the main selling point right now. Allocation-driven demand is buying non-staking exposure by a wide margin.
The paper stops short of treating that as final. First, staking ETFs are a young category and still expanding. It notes that Bitwise and Grayscale are already working on staking ETFs for Solana, and that Grayscale has also launched a product tied to Hyperliquid. Second, lower yields could slow the conversion of existing non-staking ETF assets into staking share classes, though that would be a growth effect rather than an outright outflow. Neither point changes the current order of magnitude in the report’s view: the argument is, at base, a fight around a $500 million cohort over a $1.55 billion per year transfer.
Security narrative meets redistribution
In its conclusion, the paper first acknowledges the security case made by the proposal’s backers. Chow writes that core advocates including Justin Drake and Jerome are motivated by network security: if Ethereum’s staking ratio moves past 50%, the chain could lose what they call social-layer defense against extreme attacks and become exposed to “too big to fail” LST concentration risk.
But the report says the harder reality sits on the ledger side. Since 2022, Ethereum’s burn mechanism has largely lost force and now offsets only 2.4% of issuance. In a blob-led L2 economy, a revival driven by surging L1 fees looks remote. That leaves issuance adjustment as the protocol’s last practical macro lever.
On the actual policy effect, the study says the math lands between the slogans. At about 42.2 million ETH staked, EIP-8363 cuts issuance by 58.6% and staking APR by about 56%, not enough to end inflation and not enough to destroy the staking system. The baseline equilibrium in the paper places the system around a 26% staking ratio and 0.48% annual inflation.
The amount being reallocated is not trivial in the report’s accounting. Cutting 633,000 ETH of annual issuance preserves about $1.55 billion a year, equal to 0.53% of market cap. Chow notes that this is roughly five times the size of Ethereum’s current L1 fee economy, which he puts near 0.10% per year.
The report does acknowledge real risk in DeFi, especially where lending markets are concentrated in wstETH. But it separates exposure from dependence. As long as wstETH still carries positive yield, it remains better collateral than plain WETH. What the proposal could break, in the paper’s telling, is the leveraged staking loop. The report says that once base staking yield falls below 1.16%, it no longer covers the cost of borrowing ETH, and those carry loops unwind.
On the claim that lower rewards would tank ETH, Chow returns to the statistical result: over the past 43 months, changes in staking yield and ETH price performance show no significant relationship, with p = 0.73 and R² = 0.003. The paper also notes that staking yield fell from 3.98% to 2.50% while ETH/BTC dropped 57% into July 2026, with no sign that the yield itself provided price support.
Final view: mildly bullish ETH, negative on staking middlemen, low odds of passage
The paper ends with three judgments. First, it is mildly constructive on ETH itself, not because of a scarcity narrative, but because removing roughly $1.55 billion of annual structural sell pressure looks like a favorable trade if issuance is the only lever left.
Second, it is clearly negative on staking intermediaries and related infrastructure, naming Lido, LSTs and LRTs. The report says their valuation logic rests on the staking yield stream that EIP-8363 would cut by 58.6%.
Third, Chow says the proposal is unlikely to pass. In decentralized governance, he argues, measures that impose concentrated losses while offering diffuse gains often fail even when the economics look cleaner on paper. That is the report’s base case for EIP-8363.
The paper lists four conditions that would change that view:
- Onchain data show newly issued ETH is being restaked into auto-compounding LSTs rather than sold into exchanges.
- Staking-enabled ETFs grow tenfold from the current $500 million scale, making incremental demand more important than reduced inflation.
- L1 fees recover enough for the burn mechanism to matter again.
- Future data establish a much clearer negative relationship between supply growth and price, strengthening the case that lower issuance reliably lifts ETH.