原文标题：《 深入分析以太坊上海升级后供应冲击与二阶效应 》
Steven Shi is an investment partner at Eco Fund
Leah Yuan, Foresight News
Ethereum may complete the final phase of its conversion to a certificate of interest within a month. More than 16 million ETH pledged on the beacon chain will be able to be withdrawn after the upcoming upgrade of "Shapella" (a portment of "Shanghai" and "Capella"). This number alone has caused anxiety among users, investors and traders concerned about the possible impact on supply and demand for ETH and other related tokens.
In this report, we will break down the details of the Shapella upgrade, provide what to look for before upgrading, and give our best estimate of the impact the upgrade may have on ETH, flow pledged derivatives, and other applications.
The Shapella upgrade, which will allow users to extract some or all of the ETH pledged on the Beacon Chain, is expected to take place between late March and early April.
More than 1 million ETH rewards have accumulated on the beacon chain, which will be automatically withdrawn and converted into liquid assets within approximately 5 to 7 days. Given the competitive dynamics among pledge pool providers, we expect about half of these awards to be re-pledged on beacon chains.
More than 16 million ETH was pledged on the beacon chain. We expect about 2.8 million ETH of this to be proposed, but this is a conservative estimate and the final result may differ significantly from the estimate. We expect that approximately 50% of ETH extracted will be redirected to the label chain via liquidity pledge service providers. Major beneficiaries include Lido Finance, Frax Finance and Coinbase.
In the long term, as pledged ETH will become more secure, Ethereum's pledge participation rate should exceed 40%, becoming the true benchmark rate in the Ethereum ecosystem.
Shapella will also have knock-on effects on DeFi applications including: mobile pledge providers, decentralized trading platforms, lending agreements, meta-governance projects and repledge infrastructure. Shapella may also have an impact on ETH's supply elasticity and its volatility as a trading vehicle.
Beacon chain is the backbone of Ethereum's PoS (Proof of Interest) consensus, allowing users to participate in block verification and receive rewards by pledging their ETH. The chain began accepting deposits in November 2020 and officially launched on December 1, 2020. Users need to pledge at least 32 ETH to perform consensus work in order to receive publishing incentives. Prior to the merger, consensus work consisted only of reaching consensus on the active verifiers and their balances. Merge (Bellatrix After upgrading), the verifier is responsible for ensuring network security, such as block proposal, block authentication, and synchronous committee participation.
But since its inception, the pledged ETH and reward have been locked on the beacon chain. As a result, more than 16 million pledged ETH and 1 million ETH cumulative rewards have been locked on the beacon chain for more than two years.
Next month, Ethereum will complete an upgrade called Shanghai/Capella. [Consensus level upgrades are named after stars (Altair, Bellatrix, Capella), and executive level upgrades are named after cities.]
The most obvious feature of the upcoming Shanghai/Capella upgrade is that verifiers can choose to withdraw their accumulated rewards or the entire verifier balance. As a result, ETH will move from the beacon chain to the executive layer, where ETH is free flowing and available for transactions.
Note that rewards from priority fees and MEV payments accumulate directly to the executive level (the chain of the Ethereum user's current transaction). Thus, until Shapella upgrades, the only reward that is locked in is the consensus tier release reward.
All indications are that the Shapella upgrade will take place around late March to early April. So far, test results for the upgrade have been positive. The developers successfully launched Shapella on the Sepolia test site on February 28, with only minor problems. They have finished testing the shadow forks in the Zhejiang Public test Net and the main net.
The next major milestone is to roll out the Shapella upgrade on Goerli first by mid-March and then on the main web.
After Shapella's upgrade, there will be two types of withdrawals:
1. Full withdrawal: This allows the verifier to completely exit the beacon chain and withdraw their full ETH balance, including their initial 32 ETH balance and accumulated rewards or penalties.
Ethereum has a drain limit for the number of verifiers that can withdraw money per epoch (32 blocks per epoch, approximately 6.4 minutes). This limit increases as the number of verifiers on the beacon chain increases. At present, the loss limit factor is 7. However, as the upgrade progresses, this limit may change to 8. As a result, 1,800 verifiers can withdraw in full every day when the upgrade occurs, which equates to over 57,600 ETH per day.
2. Partial withdrawal: Partial withdrawal allows the verifier to withdraw the balance beyond the principal of 32 ETH. The verifier will continue to be active on the beacon chain while extracting the revenue they have already earned. Because the accumulated reward is idle capital (that is, the reward does not compound automatically), partial withdrawals activate idle capital and improve capital efficiency. Without a partial withdrawal, the verifier would have to exit the beacon chain entirely to claim their reward, and then re-enter again. Therefore, separating partial withdrawals from full withdrawals can help improve network stability and prevent queue congestion.
Some withdrawals will be automatic after the upgrade, and the cleaning will take place on average once a week. Each slot will automatically make 16 partial withdrawals, starting with 0 indexes of the beacon chain, or approximately 147,000 verifiers per day.
The upgrade could have a potentially large supply shock. Within a week of the upgrade, more than 1 million ETH will be automatically put into circulation as a result of clearing part of the withdrawals. This represents about 0.8% of ETH's total supply and is worth about US $1.6 billion at today's prices. In addition, 16.8 million ETH could theoretically be extracted.
These figures have caused some consternation. There is talk brewing that the market could see a massive dumping as ETH that has not been liquid for more than two years is allowed to be withdrawn.
But these fears may be overblown. Below, we cover the dynamics of partial and full withdrawals to better illustrate how Shapella's upgraded withdrawals will work, and to illustrate that the supply shock to ETH may not be as extreme as the data above suggests.
Please note that our estimates are rough numbers and our estimates are subject to a number of uncertainties. Our analysis involves many assumptions, but the specific numbers estimated in the article are certainly inaccurate. Nevertheless, we hope that this report will provide food for thought, and we encourage further participation in the discussion and analysis.
In total, more than 1 million accumulated ETH awards will automatically become liquid assets. Since there are 16 verificator queues for partial withdrawals per block, partial withdrawals must first change their withdrawal credentials before receiving these cumulative rewards, so it can take five to seven days for all partial withdrawals to become liquid assets at the execution level.
But even assuming that most verifiers are conservative and want to hold liquid ETH at the executive level, there are some structural dynamics that prompt them to re-pledge these rewards onto the beacon chain.
53% 的累计奖励来自提供流动质押 Token （「LSTs」）的质押池。主要的 LSTs 由相对分散的质押提供者（如 Lido 和 Rocket Pool）或中心化的托管人（如 Coinbase 和 Binance）提供。LSTs 的公平市场价格因提供者而异，Rocket Pool 的 rETH 溢价为 +1.3%，而 Ankr 的 ankrETH 折扣为 -3.0%。大多数 LSTs 的价格在 -0.1% 到 -2.0% 的折扣范围内波动，中心化托管人的 LSTs 通常折价更大。
Source: Project website, Defillama, CoinGecko, 2023.02.24
In our view, the existence of LSTS reduces the potential supply/sales that may result from user withdrawals, as users pledging through these providers already have the option to convert their pledged ETH at a nominal discount.
此外，竞争动态激励重新质押。在所有质押池中占据领先市场份额的 Lido 最近确认了其「APR 最大化」的政策，试图尽可能地重新质押尽可能多的 ETH。由于其 stETH Token 交易的折扣仅为 -0.1%，我们估计 stETH 的赎回需求将很小。因此，Lido 可能会重新质押其累计奖励的所有 224000 枚 ETH（约 3.7 亿美元），相当于总累计奖励的 20% 以上。其他质押提供者也会被迫尽可能质押更多的 ETH，以避免在与 Lido 的竞争中因 APR 不高而失去市场份额。
我们在下表中分享了关于基准情况假设的观点，即估计部分奖励被重新质押的比例。对于具有 LST 的质押池，我们假设更高的折扣将面临更大的赎回压力，以将 Token 恢复到标准价格。
We acknowledge that our assumptions for centralized operators without LSTS are relatively loose, and the reality may differ from our understanding. For example, Kraken is unlikely to repledge in the event of the recent SEC charges and closure of the agreement to provide pledge services to U.S. clients. Celsius may also prefer ETH with high liquidity.
尽管如此，我们通过与 RockX 等机构节点运营商的交流发现，大多数机构节点运营商都希望重新质押他们的大部分奖励以将收益最大化。因此，我们预计以机构为重点的节点运营者，如 RockX 和 Figment，将重新质押他们的大部分奖励。对于具有更多散户用户的节点运营商，例如 Stakefish 和大多数中心化交易平台，我们认为只有 20% 的部分奖励会被重新质押，其余 Token 将会用于满足赎回请求。
Sources: Beaconcha.in, Rated Network, Dune, project websites, Amber estimates, 2023.02.24
Thus, our base assumption is that approximately 52% of the total supply of partial withdrawals will be re-pledged back into the beacon chain, thus greatly reducing the potential supply shock and selling frenzy.
For an in-depth analysis of the partial withdrawal model, we recommend reading this article from Data Always. The authors conclude that "the decision to automatically make partial withdrawals may be compromised by the clogging of withdrawal queues by verifiers who do not intend to sell, thereby reducing the size of the initial selling frenzy."
We predict that full withdrawals from the top 20 verifier entities will account for approximately 75% of total verifier withdrawals. Then make general assumptions about the other 25% :
A pledge pool operator with an LST will not draw the full amount from any verifier. As can be seen from the above, more than half of the validators offer LST - we believe that all of them can bring their LST back to the same price as ETH by only partial withdrawal. In fact, most pledge pool operators are likely to deploy new validators due to the incentive of re-pledging ETH.
Kraken will withdraw most pledged ETH as it is no longer allowed to provide pledge services to US customers. While it could offer pledges in other ways, the attitude of its US-oriented customer base and regulatory uncertainty could trigger outflows from Kraken's pledge products. Similarly, we expect that Celsius will fully withdraw all of its validators in order to maximize short-term liquidity.
For other centralised trading platforms and node operators, we make 10-25% different assumptions based on our understanding of their target customers (institutional and retail) and operating models (pooled/independent, managed/unmanaged). Based on our communications with various participants in the pledge, we assume that institutional demand for withdrawals is lower than retail demand. Those operators with pooled/custodial models face relatively low pressure for full withdrawals because they can balance out outflows with inflows.
Overall, we estimate that around 2.8 million ETH will be fully proposed from active verifiers after the Shapella upgrade.
However, similar to partial withdrawals, we expect some ETH to be re-pledged. Specifically, many of the pledgers who draw will transfer their ETH back to the LST's mobile pledge provider. This is because suppliers holding LST have several advantages over independent pledges:
Potential for instant liquidity via on-chain pools Use LST as collateral on DeFi, allow users to lend stablecoins, plus leverage pledge ETH, etc provide convenient re-pledge mechanisms such as Eigenlayer.
Sources: Beaconcha.in, Rated Network, Dune, project websites, Amber estimates, 2023.02.24
The re-pledge of ETH by users may reduce the impact of full withdrawals on the total supply shock by approximately 50%.
A number of on-chain indicators support our forecast of subdued withdrawal pressures:
After the Terra / 3AC debacle, the average number of new verifiers continued to rise within 14 days.
主要质押池的 LST Token 以微不足道的折扣交易，非托管的流动性质押提供商的折扣最小（RPL 的 LST 甚至有溢价）。这支持了我们的观点，即质押的 ETH 将从托管性向非托管性的流动性质押转移。
Source: Company websites, Coingecko
Since the pledge began, 39 per cent of the pledgers have made a profit on their dollar books. But a higher percentage of earnings also means more pressure to withdraw.
Source: Dune (@hildobby)
Low-level verifiers voluntarily quit as a precaution against an impending upgrade. If the verifier expects a clogged exit queue after the upgrade, some may try to exit early to avoid the queue. However, it may be too early and this is a key indicator to keep an eye on.
1136 validators have exited the beacon chain, meaning that about 39,000 ETH have entered the circulation market.
Overall, after subtracting ETH that may be re-pledged, our base forecast is that approximately 1.8 million ETH will be converted into liquid assets by way of withdrawal. This represents 1.5% of the total ETH supply and 10.5% of the total ETH supply on the beacon chain. We think this forecast is conservative and the actual number is likely to be lower than this.
We don't want all these ETH to be sold. The data suggest that Ethereum's total supply is increasingly staying on the chain, rather than being sent to centralised trading platforms. We think this trend will continue and expect a significant amount of ETH unlocked from Shapella to remain on the chain.
Predicting Shapella's upgraded selling flow model is beyond the scope of this report, and interested readers can try to build their own models. However, it is important to note that due to the exit queue of the beacon chain, any potential selling flow from full withdrawals is likely to occur over a period of weeks rather than days.
As drawdown uncertainty is finally removed, we expect net new demand to flood the pledge market. Pledged ETH will gradually be seen as the de facto benchmark interest rate within the Ethereum ecosystem, similar to the federal funds rate in the US economy. All pledge yields should be at least in line with DeFi blue chip agreements such as Aave, Compound and Euler.
排除 Token 奖励之后的收益率，来源: Project websites, Defillama, 2023.02.24
In the long term, we expect more than 40% of ETH to be pledged. With a 40% participation rate, the verifier's overall APR was around 2.95%, assuming MEVs and priority charges remained average over a 30-day period.
Source: Staking Rewards
Shapella will benefit liquidity pledge providers in several ways. As mentioned above, LSTS are arguably better collateral than ETH. For example, if stETH can be used to borrow stablecoins instead of ETH, the borrowing capital secured against ETH will become less efficient. In addition, drawdown can enhance the anchoring of LST to ETH, thus making it more attractive as collateral. As a result, we expect liquidity pledge providers to gain market share.
相关的是，这些项目维持它们的 LST 和 ETH 之间锚定的成本将变得更低。Shapella 升级之后，LST 的即时链上流动性变得不那么重要。例如，假设 Lido 减半其 Token 发行量以激励 Uniswap 等去中心化交易平台上的流动性。即使锁定的总价值大幅缩小，stETH 仍将与 ETH 紧密锚定在一起。这是因为如果 stETH 真的以远低于锚定的价格进行交易，第三方套利者就会介入。在大多数情况下，提款只需要 4 到 5 天即可完成，所以我们认为，即使与锚定价格相差 10-30 个基点，套利者也会出手介入。( 在 30 个基点时，套利者的年化收益率有望达到 25%)。因此，像 Aave 和 Compound 这样的 DeFi 项目仍然可以自信地将 stETH 用作抵押资产，因为清算人和套利者随时准备购买任何脱离锚定价格的 stETH。
总之，流动质押池很可能面临两种结果：LST 的广泛采用将增加其收入， Token 发行量减少将减少其成本。一个反驳的观点是，这些 LST 协议仍需要激励流动性，以便在 DeFi 生态系统中得到很好的融合，并允许借贷协议进行即时清算。但是，Euler 将 cbETH 作为抵押品进行整合的先例表明，团队意识到即时链上流动性并不能完全反映 Token 抵押品的质量。
We think the biggest beneficiary will be Lido Finance. Because stETH is widely used in different DeFi applications, it provides the most utility in LST. Lido's portfolio of administrative verifiers allows it to accommodate a large number of pledges (over 150,000 ETH was registered in a recent day). Frax Finance and Coinbase will also benefit for similar reasons, although their LSTS are less broad-based. Coinbase also offers higher margins with a 25% return.
The Rocket Pool may benefit less. While we like its decentralized design and philosophy, it faces scalability challenges due to ETH/RPL collateral requirements. And, to attract node operators, its higher profit margins are also a barrier to growth. Still, note that its Atlas upgrade is an important catalyst. Atlas allows node operators to halve their collateral requirements from 16 ETH to 8 ETH, which will triple the capacity of each operator - 16 ETH pledged in one validator can now be spread across two validators, Therefore, the capacity of other users increases from 16 ETH to 48 ETH.
Shapella not only affects ETH and pledge providers, but also has knock-on effects in many industries, such as:
Stablecoin with ETH as collateral. Notably, Liquity (LUSD) and Reflexer (RAI), two stablecoins with ETH as collateral, face headwinds after Shapella. As the pledged ETH reduces risk, users may be more willing to use the liquid pledged ETH as collateral rather than directly using the ETH. Moreover, because Liquity and Reflexer are specifically set to minimize governance, they cannot switch the pledge base to ETH with liquidity pledge. Perhaps it's a good thing that LSTS are still at risk. But these stablecoins already have a very small supply in circulation (LUSD: $230 million, RAI: $7 million) and it is likely that they will see a lower market share after Shapella.
Curve/Convex 和其他去中心化交易平台。Lido 已经花费了近 8600 万自己的治理 Token （今天的价格约 2.3 亿美元）在去中心化交易平台上激励流动性，主要是针对 Curve Finance。这主要是通过创建更深的流动性来使 stETH 成为更有吸引力的资产，让两种资产之间的交易具有低滑点。LDO 发行数量通常至少占提供 stETH/ETH 交易对流动性的 APR 的一半。如果 Lido 减少这些奖励，我们可能会在 Curve 和其他去中心化交易平台上看到较低的 stETH/ETH TVL，可能会导致它们的估值重新评估。为了更好理解，在 Curve 上，stETH/ETH 交易对占总锁定价值的约 32%。
Meta-governance projects such as Convex and Aura, which will centralise voting and receive awards from projects such as Lido, will also see lower awards in this case.
另一方面，Shapella 升级还可能催生一批新的流动质押 Token 。Frax Finance 仅几个月前推出了其 LST，并利用其对 CVX/CRV 的控制来赢得市场份额。Yearn Finance 即将推出一篮子 LST，将遵循类似的策略来获得流动性。在 Shapella 升级后，我们可能会看到一场突如其来的 LST 战争。
Leverage pledge. Since anchoring between LST and ETH is easier to maintain, users may feel more confident about leveraging the ETH through on-chain lending agreements. Primitives such as e-Mode for Aave version v3 allow users to borrow ETH via LST at 90% of LTV, while Gearbox allows up to 10x leveraged trading on stETH, providing users with interesting venues for experimentation (and failure). Therefore, even if no new users want to pledge ETH, existing users may drive ETH growth on the Beacon Chain.
Repledge as a primitive. While EigenLayer's white paper was only released this month, the idea of re-pledging ETH as a basic operation for navigating a new network is already generating buzz in the Ethereum community. The idea is not necessarily new; Avalanche and Cosmos both have variations on the same concept. However, this will further strengthen the position of pledged ETH as a required pledge. Moreover, if the pledge participation rate does reach more than 40 per cent, the base rate of return will compress to 3 per cent or less. In this case, platforms that offer higher returns on the same assets will be very popular.
The elasticity of supply of Ether. The elasticity of supply affects Ether's asset classes for investors and traders. Low elasticity usually increases volatility. One open question is whether Ether's supply elasticity is higher or lower after Shapella's upgrade. On the one hand, allowing more than 16 million ETH to become liquid greatly increases supply elasticity. On the other hand, this risk is mitigated by dequeuing and Ethereum's relatively low pledge rate. Higher ETH pledge rates may increase short-term resilience, potentially increasing volatility in asset classes.
We don't pay for panic. While anything can happen in the short term, the Shapella upgrade is mostly a positive. Among other things, it triggers higher pledge rates of participation, enhances the status of pledged ETH as desirable collateral, improves Ethereum security, and reduces the risk of Ethereum as an asset.
If we must call Ethereum "finished", when can we call it "finished"? By the end of 2023. The merger is complete, the withdrawal is complete. Be able to withdraw from the proof of equity system. Basic capacity expansion is complete. Many basic improvements were made. Enough cryptography was added to Ethereum so that we could support privacy solutions on it. We can stop here if we need to. -- Vitalik Buterin on the Network State Podcast
Most importantly, Shapella will finally end Ethereum's transition to proof of equity. Developers will finally be able to fully focus on other, more urgent agendas on the Ethereum roadmap. After Shapella, note EIP-4844, which reduces the cost of Ethereum L2 Rollup through danksharding, paving the way for Ethereum's expansion path.
Beaconcha. in-Beacon Chain blockchain browser and statistics
Beaconscan-Etherscan's beacon chain blockchain browser
Rated Network - Ethereum verifies the ratings and metrics of the node
Lido Ethereum verifier and node metrics
Pledge Calculator - Beacon Chain
Hildobby Ethereum Pledge Dashboard (Dune)
The Lido extraction is designed to manage withdrawals from the Beacon Chain, which clarifies the overall structure of most LSTs. The stETH holder needs to go through three steps to extract:
1. Request: To extract stETH, the user can send an extract request to Lido to lock the number of stETH to be extracted.
2. Fulfillment: After the user enters the queue, the Lido protocol processes the request in a first-in, first-out order. If there is ETH in the buffer from new deposits or partial withdrawals, it can be used to meet the withdrawals. If there are not enough ETH in the buffer, Lido will request the necessary number of validators to exit. Note that this queue is internal to the Lido protocol and is separate from Ethereum's extract queue.
3. Claim: Users can claim their ETH share.
An interesting choice in the design is that the extraction request cannot be cancelled, but can be transferred. This would create a potential secondary market, where the front line of users requesting Lido withdrawals could sell their positions to those more desperate for liquidity.
There will be two models for Lido withdrawals. Under normal circumstances, Lido withdrawals will run in top speed mode. Users can usually convert stETH to ETH in about 3-4 days, which may be reduced to as little as 1-24 hours if there are enough ETH in the buffer.
However, if a "mass punishment incident" occurs, the Lido will go into cover mode. This can happen when more than 600 Lido validators are penalized at the same time. So far, no Lido verifiers have been punished. In this mode, all extraction requests are deferred until losses are socialized between all users. This is to prevent savvy users from gaining an advantage, since penalties and fines occur asynchronously. Without the bunker mode, savvy participants anticipate mass punishment and thus extract stETH into ETH to avoid losses, converting ETH back into stETH later.
Lidos are structured to give priority to the mortgagee, with the principle of maximising annualised returns. Lidos do not leave more ETH in the buffer in order to facilitate faster withdrawals, but rather ensure that ETH stays in the buffer for as short a time as possible.
This dynamic ETH supply "sinks" from Shapella to the beacon chain. It also forces competitors to do the same, or else their APR will fall and they will gradually lose market share.