Kraken’s decision to wind down its staking business in the US (as part of its settlement with the SEC) has added to the uncertainty surrounding ETH market supply dynamics ahead of the Shanghai Fork. Kraken’s staking pool on Ethereum currently makes up around 7% of total staked ETH supply (~35.8k validators), although presumably not all of these assets come from US retail consumers.
Indeed, in the complaint filed by the SEC, the agency alleges that Kraken’s revenues from US investors in its staking program represented only 30.6% (more than US$45M) of its total net revenue earned since its inception in December 2019. (Note that that’s across the 14 digital assets that Kraken provided staking services for, totaling US$2.7B as of April 2022.)
We think that suggests unstaking by Kraken could result in a minimum of 350k up to a maximum of 1.145M additional ETH that could potentially circulate in the market when staked ETH withdrawals are enabled in the Shanghai Fork (expected in March 2023). Because withdrawals will be rate limited to seven validators per epoch (1575 validators per day), that could extend the exit queue from anywhere between 7 days at a minimum to 23 days at a maximum.
What’s unclear is if the unlocking of this additional ETH could be a catalyst for selling pressure. In late December, market players were concerned that the Shanghai Fork represented an important event risk to the downside, but that view shifted in January as general market sentiment improved. Although the regulatory environment has muddled the outlook somewhat, we believe the selling pressure should still be fairly limited, as we believe there are mitigating factors and self-correcting mechanisms that should help control the flow of ETH on the open market.
Softening the impact
First, compounding should not be an issue. That is, validators will not need to fully unstake and redeposit their ETH in order to stake any earned rewards, as there will be both partial and full withdrawals. Partial withdrawals will be automatic (assuming credentials are set properly) for balances exceeding 32 ETH so that only earned rewards are withdrawn. Full withdrawals on the other hand require explicit user action as they constitute an exit from the validation process entirely. Note that there is only one single withdrawal queue, and partial withdrawals should take less time to process than full withdrawals.
1. Distribution of ETH staking by validator pool
At the moment, there are around 1.03M ETH in earned rewards for 516k active validators, which should presumably take about five days to fully distribute (as 16 partial withdrawals can be processed every 12 seconds). This is where we think most of the sell pressure could be concentrated, but there are mitigating factors. For example, more than 67% of all staked ETH is currently staked through third-party entities, which could use those released rewards to set up new validators. Thus, we estimate the total ETH circulating in the market from this source could be closer to 340k ETH ($517M) carried over several days.
If we add the maximum 1.145M ETH to this amount from Kraken’s ETH staking operations, distributed over the respective disbursement periods (for partial vs full withdrawals), that leaves us with potentially ~118k ETH ($103M) being sold per day from these two sources – and that’s assuming the full amounts are in fact being sold. We believe that could easily be absorbed by the market as daily trading volumes for ETH averaged $8.2B over the last 30 days, according to The TIE. (Note however, that the flows are unlikely to be uniform over a set period of time, and our estimates are based on the number of validators that are active on the network today. If the number of active validators increases to over 524,288, the number of allowed validator exits increases to 1800 per day, possibly unlocking more ETH sooner.)
An increase in validator exits should however programmatically lead to higher staking yields for the remaining validators, as the base reward is inversely proportional to the square root of the total balance of all validators. That may attract some forced Kraken unstakers to redeposit their assets again. Many staking pools have no minimums on ETH staking, and there are both centralized and decentralized alternatives. Rocket Pool for example has a decentralized infrastructure with only a 0.01 ETH minimum deposit for stakers. Plus, for its node operators, there is an LEB8 proposal that will lower collateral requirements to 8 ETH (from 16 ETH), which may be a more capital efficient solution for solo stakers who currently need to put up 32 ETH.
The potentially higher base rewards should also improve the case for institutional investors interested in staking their ETH but who were previously concerned about the tradeoff between liquidity and yield in a pre-Shanghai Fork world.
In short, once withdrawals are enabled, we would still expect to see a net increase in ETH staked from its current ratio of 14% of total ETH circulating supply. Compare this to an average 65% (of eligible token supply) observed in other alternative layer-1 networks. (That said, we would expect the ETH staking ratio to be less than that average given the relative amount of Ethereum usage, likely closer to 25-30%.) All in all, we expect high staking demand should offset the amount of forced ETH withdrawals.
2. Staking ratios (% of eligible supply staked) by network
Conclusions
Overall, we think ETH performance around the Shanghai Fork will be less strictly dependent on technicals and more contingent on what risk does at that time. That is, if the macro environment worsens and equity markets are selling off in March, people may decide to unstake and sell ETH just to de-risk, while institutions may not step in as aggressively on the buy side. Conversely, if risk sentiment is positive, we would expect demand to be a more than sufficient counterweight to the ETH being unlocked on the open market. Thus, we think it will be very difficult to disentangle the impact of more ETH coming to the market from the investor sentiment at the time withdrawals are enabled.
