Author: Mario Chow, IOSG
EIP-8363 proposal: As the staking rate increases, a growing portion of validator rewards is burned, reaching 100% burn when 50% of the supply is staked. This paper models its impact on issuance, yields, and staking equilibrium; examines whether ETH’s yield has truly explained its price; quantifies how much of the on-chain economy truly depends on this yield; and presents our conclusions.
All calculations are based on on-chain data and the original EIP text. The model is independently constructed and aligns with publicly available third-party data within 2%. Data updated as of August 24, 2026.
A one-sentence argument
Fee burning is dead, leaving issuance as Ethereum’s only remaining lever on ETH supply. Under this proposal, at current staking levels, issuance is halved—not eliminated; and it is self-limiting: under any reasonable staker yield threshold, the system eventually stabilizes with 26–34% of supply staked and an issuance rate of 0.3–0.5% per year. Meanwhile, the yield reduction it imposes shows no detectable relationship to ETH’s price.
One, the burn mechanism no longer works.
EIP-1559 burned 1.48 million ETH in 2022. EIP-1559 burns the base fee, which is essentially a congestion pricing mechanism; once blobs move rollup data off L1 and the gas limit is raised, congestion disappears: gas usage doubled, while the average base fee dropped by 96%, and burn volume has fallen by 98% since 2022. Over the past twelve months, it has burned a total of 25,660 ETH, and the recent 30-day burn rate is even lower: 39 ETH per day, annualized at approximately 14,300 ETH.
▲ Daily ETH burned from EIP-1559 fees by year: Declines from 8,844 ETH per day in 2021 to 57 ETH per day in 2026—far below the current issuance curve and far below the maximum issuance curve under EIP-8363.
Compared to the total issuance of approximately 1.08 million ETH per year, burns currently offset only 2.4% of new supply. As a mechanism, "ultrasound money" has ended. The migration to L2s and blob scaling has shifted the fee base away from L1: during the same period, L1 gas usage actually doubled (from 34 billion to 67 billion units per month), while the average base fee dropped from 4.00 gwei to 0.17 gwei—this is a price effect, not a demand effect.
Net issuance: What exactly happened to the supply
Burns are only half the ledger. Viewed alongside the issuance, the picture is even grimmer: issuance has never stopped growing, while the offset has vanished directly beneath it.
▲ Comparison of ETH minted on the consensus layer each month since the merge versus ETH burned by EIP-1559. The issuance bar grows steadily; the burn bar shrinks nearly to zero by 2025.
Over the combined 47 months, only 13 months were deflationary: the most recent being March 2024. ETH has been in inflation for 28 consecutive months, with the rate tripling during this period from +0.26%/year to +0.87%/year. The reason is not an increase in issuance (up only 4% since 2024), but rather the elimination of offsetting factors.
This reframes the entire debate. EIP-8363 is often framed as a choice between staking rewards and monetary scarcity. But a more accurate understanding is much narrower and more constrained: monetary policy is now Ethereum’s only remaining lever on ETH supply, because the demand-driven one no longer works. Whether or not anyone legislates it, all supply issues now pass through the issuance curve.
II. What exactly does EIP-8363 do?
Before unpacking the mechanism, we must first clarify the official core motivation: defending network security. The proposal’s authors argue that once the total staking rate crosses the 50% threshold, Ethereum will lose its “social layer defense” and face a systemic parasitic risk from LST oligarchs that are “too big to fail.” Thus, the proposal seeks to lock in a staking cap through forced disinflation. However, grand security philosophies often obscure the real financial stakes on the ledger. Setting aside metaphysical debates about decentralization, what does this mechanism actually mean in the real on-chain economy? Below is a purely quantitative analysis.
How is the money deducted? (Core mechanism)
Pay first, deduct later: Validators initially receive the full reward for all tasks, but then the system directly burns a portion of the reward at a specified ratio (denoted as b).
Deduct based on the theoretical maximum, not double punishment: The key here is that the system calculates the burn amount based on the full reward you theoretically should have earned, not on what you actually received. Why? Because if you accidentally go offline, you wouldn’t earn the reward anyway; if the system also deducted based on your actual earnings, it would be unfairly punitive to those who were offline. Deducting based on the “theoretical value” ensures everyone stays motivated to contribute, while preventing offline users from being penalized twice.
Emergency protection: If the Ethereum network experiences severe issues (entering an inactivity leak state), this destruction of attestation rewards will be paused.
External snapshot take: This proposal only affects rewards at the consensus layer. The “side income” you earn by running a node—namely MEV and priority fees—will remain untouched and unaffected.
The two most common misconceptions in the community
Misconception one: "Ethereum's issuance rate will be cut to zero directly"
Fact: There isn’t nearly that much. The current staking amount is approximately 42.2 million ETH, at which level the burn rate b is 58.6%.
To bring the inflation rate to zero, the staked amount would need to surge to 60.25 million (43% higher than current levels). Therefore, a more accurate statement is: this proposal at this stage only cuts the inflation rate in half, and there’s still a long way to go before reaching zero.
Misconception 2: "Yield plummets instantly, causing a DeFi crash on the first day of launch"
Fact: The official design included an 18-month "soft landing" period, with the launch day being almost imperceptible.
To prevent sudden shocks, upon launch, the proposal will double the base reward factor to 128. This doubling exactly offsets the 58.6% burn rate mentioned earlier.
That means on the first day of the upgrade, the net issuance rate across the network will remain at around 83% of the current level. Over the next 18 months, the parameter will gradually decrease back to the normal level of 64, causing issuance to slowly decline to 41% of the current rate.
Summary: The decline in yields has been gradually diluted over a period of one and a half years, not a sudden crash. Those concerned about an “immediate burst of the DeFi bubble” are overlooking this buffering mechanism.
Three: Baseline — Where is Ethereum right now?
Supply dynamics
Where does the supply come from?
All come from staking rewards. After consolidation, new ETH has only one source: consensus layer payments to validators, allocated with fixed weights (denominator 64) across three roles: attestation 54/64 (84.4%, 911,672 ETH/year), block production 8/64 (12.5%, 135,063), and sync committee 2/64 (3.1%, 33,766).
The issuance rate is modeled as I(S) = 940.9 · √(S/32) ETH/year, representing the protocol's own reward curve. At S = 42.2 million, this corresponds to a consensus layer APR of 2.560%. Measured priority fees over the first 23 days of August amounted to 2,623 ETH, annualized to 41,500 ETH: equivalent to 0.098% of the staked base. Adding both yields 2.658%, which aligns almost exactly with the reported 2.66%.
Based on measured priority fees, validators receive at least 96% of their income from issuance and at most 4% from fees. Proposer payments exceeding direct priority fees in MEV-boost are not captured, making the fee share a lower bound. Regardless, issuance dominates absolutely, and this proportion is the crux of the entire debate.
Four: Modeling the Proposal
The earliest attempts were designed to build an entire network around "storage," rather than patching existing networks. Two coins led the way on this path, but placed completely opposite bets. A third case was built for banks, not individuals, but belongs to the same family.
Applied at current staking levels, without considering behavioral responses
Supply reduction: −58.6%. Staking APR reduction: −56.4%. Dilution removed: 633,000 ETH/year = $1.55B/year = 0.53% of annual ETH market cap.
▲ Relationship between annual ETH issuance as a percentage of supply and the staking rate. Today’s curve rises steadily; the EIP-8363上线 curve peaks at approximately 1.0% around a 20% staking rate; the permanent curve peaks at approximately 0.5%. Both curves decline to zero at a 50% staking rate.
Full curve (after complete transition)
Supply peaked at around 25 million staked tokens, approximately 0.505% of the total supply, before declining—consistent with EIP's own statements.
Balance—the number that truly ends the debate
Stakers are not passive. If the yield falls below their required return, they exit, which both increases the gross APR and reduces b. Solving for the fixed point:
▲ The relationship between total staking yield and the amount of staked ETH under current rules and EIP-8363. The EIP-8363 curve intersects the 2% threshold at 31.2 million staked ETH and the 1.25% threshold at 40.9 million staked ETH.
Read this table in light of the two most prominent arguments in the debate:
The supply will go to zero—this is only true if marginal stakers are willing to work for a return of about 0.5%. At any reasonable required return, ETH still experiences 0.3–0.5% annual inflation. Supporters are exaggerating.
“Staking will collapse.” At a 2% threshold, the staking rate would stabilize at 26%: below today’s 35%, but roughly in line with 2024’s full-year levels. Critics have also exaggerated.
This mechanism is inherently self-limiting. This is the most interesting property of the design and also the least discussed.
Five: Can staking yield explain ETH's price?
Address the problem behind the problem first.
Is the staking rate related to the yield? Yes: it is fully related, and this relationship is defined by definition, not observed. This must be clarified first, as it determines what the data can and cannot show.
The reward pool for protocol payments is scaled by the square root of the staked balance, resulting in a closed-form solution for the yield per ETH:
Issuance(S) = 940.9 · √(S/32) ETH/year APR(S) = Issuance(S)/S = 166.28 / √S
The more you stake, the more the same pool is divided among more coins. The correlation between the staking rate and the issuance yield is inherently −1. Plotting them together is equivalent to plotting an identity.
The only free variable is the difference between the disclosed yield and the formula value: fee income. It was approximately 1.34 percentage points in 2022 and is now 0.10 percentage points.
Relevance itself
Answer: No correlation exists. 43 months, January 2023 → July 2026. (Data refreshed on August 24 did not rerun this calculation; the window ends in July 2026, and price fluctuations after that do not affect this result.)
Regression results
▲ Relationship between month-end ETH price and staking APR with OLS fit. The fit appears strong, but residuals exhibit significant autocorrelation.
This level of significance at p = 0.006 is “statistically significant”—but it is meaningless. The Durbin–Watson statistic of 0.40 indicates severe serial correlation in the residuals, a textbook hallmark of spurious regression between two trending series. Both variables exhibit trends, so they are correlated; standard errors are underestimated, and the p-value is invalid. Keep this chart as a warning, not as evidence.
▲ Scatter plot of monthly ETH returns versus monthly staking APR changes, with an OLS fit line close to horizontal and wide residual bands.
After differencing to remove the trend, the relationship disappears: p = 0.73, R² = 0.003. The Durbin–Watson statistic is 1.75, indicating this specification is clean. The 95% confidence interval comfortably spans zero in both directions—the data cannot even determine the sign of the effect, let alone its magnitude.
▲ The staking yield change exhibits a 12-month rolling correlation with ETH returns, oscillating around zero and mostly falling within a range indistinguishable from zero.
And this is not a stable relationship hidden within a noisy mean—the rolling correlation repeatedly crosses the zero line and spends the vast majority of its time within a range indistinguishable from zero.
From January 2023 to July 2026, ETH’s staking yield declined from 3.98% to 2.50%, and ETH/BTC fell by 57%. During the same period, the monthly correlation between staking yield changes and ETH returns was −0.05. The yield has been consistently there.
It did not protect the price, and its compression did not cause a decline. If the naturally occurring 37% reduction in the reward curve has had no detectable price effect, then the burden of proof lies with anyone claiming that “one more cut will make a difference.”
Note: The staking sequence is reconstructed from on-chain traffic and is approximately 5% higher than the published data. The direction and shape are reliable, but the absolute levels are not precise.
The supply growth rate also cannot explain it.
If yield doesn’t affect price, then what about the actual change in supply numbers this proposal brings? Same test, same window—replace yield with net supply growth rate.
▲ The relationship between monthly ETH returns and annualized net supply growth. The fitted line slopes downward, but the scatter points are widely dispersed, indicating an insignificant relationship.
Slope: −6.9 (for each additional percentage point in annual supply growth, monthly returns are 6.9 percentage points lower), p = 0.18, R² = 0.044, Durbin–Watson 1.82. The 95% confidence interval for the slope is −17.1 to +3.4.
Honestly assess this result, as it cuts both ways. The relationship is not statistically significant, as the confidence interval spans zero, so it cannot serve as evidence that “reducing supply growth raises price.” However, it is about fifteen times stronger than the yield relationship (R² of 4.4% vs. 0.3%), and the sign aligns with theoretical predictions. If either variable has any marginal effect, the data point to supply, not yield—as precisely the exchange made by EIP-8363.
Six: How reliant is the on-chain economy on ETH yields?
Liquid staking
Lido alone accounts for 48% of Ethereum’s entire $48.5B DeFi TVL. Any claim that “DeFi will be fine” must first withstand this number.
What does the cut in yields mean for each of them?
Liquid staking: Revenue impacted. Lido handles approximately $602M in staking rewards annually, taking a 10% fee (about $60M/year). This 58.6% reduction in issuance equates to roughly 633k fewer ETH in rewards distributed per year; at Lido’s 22.8% market share, this means it loses about $35M in fees annually—nearly half of its revenue from this segment. While significant for Lido, it’s negligible on the broader Ethereum scale. Regardless of yield changes, wstETH remains superior to WETH for any borrower seeking ETH exposure, and its role as collateral remains intact.
LST as collateral for lending and borrowing—the true point of dependency
▲ Percentage of liquid staking tokens in TVL: SparkLend 66.9%, Aave V3 38.7%, Morpho Blue 10.4%, combined total 34.2%.
Of the $31.1 billion in collateral across Ethereum’s three major lending markets, $10.63 billion (34.2%) consists of staking yield derivatives. SparkLend is a classic single point of failure: two-thirds of its collateral is wstETH.
ETF channel, quantify it
The most commonly cited objection is that reducing yields will drain institutional demand, as staked ETH ETFs market yield to investors who cannot access it directly. This channel is real—but it is also very small today.
Products explicitly marketed for yield account for only 5.4% of ETF assets, 0.53% of staked ETH, and 0.19% of total ETH supply. BlackRock’s non-staked ETH product is ten times larger. Whatever is driving institutional capital into ETH, staking yield is not the main attraction—allocational capital overwhelmingly prefers non-staked exposure.
Two factors prevent this conclusion from being an absolute certainty. First, the staking ETF category is still young and growing: Bitwise and Grayscale are already developing staking ETFs for Solana, and Grayscale has also launched one for Hyperliquid, meaning future risks are greater than current AUM suggests. Second, declining yields may slow the rate at which existing non-staking ETF assets convert into staking shares, but this affects growth momentum, not capital outflows. Neither of these points changes the scale: ultimately, this is a $0.5B group vying for a $1.55B annual transfer of value.
Seven: Conclusion and Analysis: When "Security Anxiety" Meets "Redistribution of Benefits"
Set aside the grand security narrative first.
We must acknowledge that the core authors of EIP-8363, such as Justin Drake and Jerome, had an extremely serious intent regarding network security. From a game theory perspective, once the global staking rate crosses the critical threshold of 50%, Ethereum will lose its “social layer defense” capability against extreme attacks and face systemic parasitic risks from LSTs becoming too big to fail. Therefore, this proposal seeks to use forced interest rate reductions as an economic mechanism to lock the staking rate securely within a safe zone.
But behind the philosophy of security, the reality of on-chain data is colder.
Since 2022, Ethereum’s “burn” mechanism has effectively ceased to function: burn volume has plummeted by 98%, now offsetting only a mere 2.4% of issuance. Regardless of your stance on the security intentions behind EIP-8363, an inescapable fact remains: the mechanism that once dynamically adjusted ETH supply in response to market demand has come to a halt. In today’s L2 economics dominated by blobs, expecting L1 fee spikes to revive the burn mechanism is wishful thinking. Ethereum’s monetary policy has entered a state of “autopilot without a steering wheel,” and adjusting issuance is the only trigger we still have left to pull.
Setting aside emotions, the real policy impact lies between these two extreme viewpoints.
Supporters cheer “ending ETH inflation,” while opponents warn of “staking system collapse”—both rhetorics deviate from mathematical reality. With current staking at 42.2 million ETH, this proposal would reduce issuance by approximately 58.6% and lower staking APR by about 56%. To bring issuance to zero? Staking would need to surge to 60.25 million ETH (43% higher than today). More importantly, the mechanism includes a built-in brake: as yields decline, some stakers exit, and the system ultimately stabilizes around a 26% staking rate and 0.48% annual inflation. What it actually delivers is simply halving the dilution—not destroying or overturning anything.
How important is that “half a percentage point” you saved? Numbers are more honest than words.
At the current price, reducing ETH issuance by 633,000 ETH per year preserves $1.55 billion, approximately 0.53% of the total market cap. Don’t dismiss this as a small percentage—it’s roughly five times the current annual L1 fee revenue of Ethereum (about 0.10%/year). For an asset whose fee income has all but dried up, plugging a structural annual leak of 0.5% is far from a “rounding error”—it’s the largest economic lever available right now.
What about the cost? Will DeFi really collapse? Risks do exist.
Opponents often point to collateral, such as the fact that two-thirds of SparkLend’s collateral is wstETH. But we must distinguish between “exposure” and “dependence”: as long as wstETH continues to generate positive yields, it remains a superior collateral asset compared to standard WETH—this foundational layer remains solid. What EIP-8363 would truly shatter is the “leveraged staking loop.” When the base staking yield falls below 1.16% and can no longer cover the interest cost of borrowing ETH, the leveraged arbitrage capital will unwind. In other words, it’s the leverage bubble that contracts—not the collateral system itself. As for direct protocol losses, Lido loses approximately $35 million annually—roughly half of its fee revenue.
Concerns about "cutting yields causing a market crash" have already been settled by the market.
Over the past 43 months of data, there is no significant correlation between staking yield fluctuations and ETH price performance (p = 0.73, R² = 0.003). The decline in staking yield from 3.98% to 2.50% did not prevent the ETH/BTC exchange rate from crashing 57% in July 2026. Yield is neither a moat for price nor did its compression trigger a sell-off. If a previous 37% drop in yield failed to move the price, those claiming “one more cut will crash Ethereum” need to present stronger evidence.
Why is this debate so intense?
Because this is a zero-sum game where losses are highly concentrated and gains are extremely dispersed.
Setting aside obscure technical jargon and grandiose security rhetoric, the essence of EIP-8363 is a blunt wealth redistribution: currently, stakers take 100% of newly issued ETH, yet they hold only 35% of the total supply. This means they are shifting the cost of inflation onto the remaining 65% of token holders. Reducing this $15.5 billion in issuance is equivalent to forcibly returning $1 billion in implicit wealth annually from stakers (intermediaries) to all non-staking ETH holders.
This is the real reason everyone has gone blind with greed:
The affected parties are highly concentrated: Lido, LST issuers, re-staking protocols, and leveraged traders. This is a small, well-capitalized, and highly organized group of stakeholders. They are acutely aware of how much real money this proposal will take from their pockets (Lido would directly lose half its profits, and the leverage cycle would die outright).
The beneficiaries are extremely dispersed: ordinary holders accounting for 65% of the supply. They avoid 0.5% annual dilution, but when spread across a market cap of nearly $300 billion, this amount is negligible and goes completely unnoticed—no one will take to the streets over it.
This explains why current debates are always filled with empty slogans. When an interest group cannot openly say, “This will take away our $1 billion in annual profits,” they raise the shield of “This will destroy DeFi”; and when researchers seek to forcibly reclaim the faucet of money issuance, the most politically correct weapon is “Defend network security.” Understand the volume of opposition and support as a reflection of the concentration of利益分配, not the mathematical correctness of the proposal itself.
Our final assessment
Strategy: Moderately bullish on ETH-denominated assets, clearly bearish on staking intermediaries/infrastructure. The proposal is likely to be rejected.
At the asset level, our bullish view is not based on the "myth of scarcity," but on common sense: when the only lever capable of regulating supply is rendered ineffective, removing a structural selling pressure of $1.5 billion annually—paid to individuals who never truly paid for the yield—is an exceptionally high-value transaction. It may not be a dramatic reversal, but the compounding effect is significant.
At the intermediary system level, the logic is flawless. The core valuation logic of Lido, LSTs, and LRTs is entirely built on a staking yield that is about to be halved. This is not emotional panic—it’s a real 58.6% contraction in the profit statement.
As for the fate of the proposal? The probability of passage is extremely low. In decentralized governance, the “concentrated harm vs. dispersed benefits” dynamic is the standard script that kills good proposals. Economically sound, but politically stalled—this is our baseline expectation.
Conditions that would cause us to change our perspective (falsification indicators):
On-chain data proves that "newly minted tokens are reinvested, not sold": If fund flows show that newly minted ETH remains within auto-compounding LSTs and does not enter exchanges to be dumped, then our selling pressure assumption does not hold.
Staking ETFs bring massive buying pressure: A current size of $500 million is negligible. But if scaled up tenfold, the marginal demand will surpass the significance of inflation reduction.
L1 fee revival: If the burn mechanism regains dominance in fundamentals, the urgency of artificially manipulating supply will vanish.
The inverse relationship between supply and price has been thoroughly empirically validated: currently, the correlation between the two is very weak; if data over the next year confirms that reducing supply necessarily drives up price, this would become the most robust quantitative pillar for a bullish case on ETH.