Ethereum Staking Deep Dive: Understanding Consensus Layer Rewards

After nearly three months since it’s most recent update, we feel it is important to take stock of the action so far and dig into the vast amount of data generated by Ethereum's consensus protocol - the Beacon chain. To do so, Block Scholes has teamed up with Twinstake, the market-leading institutional staking provider. Combining our deep, analytical crypto-research framework with Twinstake’s staking experience & rich staking data sets, we will build a solid understanding of Ethereum’s staking incentives, mechanisms, performance and risks from the ground up. Armed with this foundation, we intend to explore the future of institutional Ethereum staking.

This report is the first instalment of a three-part series, where we will dive deep into:
Part 1: Predictable workings of Consensus Layer rewards.
Part 2: Exciting Execution Layer rewards, and the drivers of the high variance of their performance.
Part 3: The future of Ether staking a building block, on which further financial and technological innovations will be constructed.

We hope that you’ll enjoy this series and that our collaborative efforts traversing the realms of theory and practice will inspire you to join the exciting journey of Ethereum staking.

Consensus Layer Rewards

Where do staking rewards come from?

Stakers can expect to earn a 3.78%* APR from Consensus layer rewards alone.

*From an observed validator set size of 650,471, recorded at epoch 213,000 - 04:00 UTC 6th July 2023.

‍Since the Merge in September 2022, the Beacon chain and its validators have been responsible for creating new blocks, validating them, and adding them to the blockchain. By staking ETH tokens and taking part in this operation, users are given the opportunity to earn an expected APR of 3.78%* on their stake. This rate is earned by performing several different activities, each corresponding to an action that helps the performance of the consensus protocol.

Our Goal

We aim to break down the total rewards earned by stakers on Ethereum’s Beaconchain since the execution layer was merged into the consensus layer - then compare them with the theoretical maximum rewards that are promised by the protocol. To do so, we will make a distinction between the rewards that a staker earns into two categories:

1. Rewards that are paid to a staker from the consensus layer. These rewards are payments of newly minted ETH to stakers for performing duties that keep the protocol operating correctly and confirming and finalising blocks when they are proposed. We will refer to these as “consensus layer” rewards*:

*From an observed validator set size of 650,471, recorded at epoch 213,000 - 04:00 UTC 6th July 2023.

2. Rewards that are earned through transactions on the execution layer. These rewards are either paid by an ETH user paying a higher transaction fee to incentivise the proposer to include their transactions, or payments received from block builders to propose their preferred block. These include:

• Priority Tip Rewards**: Making sure to add specific transactions that have paid priority incentives to the block proposer.
• MEV Rewards**: Using specialised external block builders to order and bundle transactions in a way that maximises value from transactions.

** Execution layer rewards are to be analysed in the sequel to this report.

We will first explore the source of and drivers behind each Consensus Layer reward available to stakers. In the sequel, we will explore the art of maximising Execution Layer rewards, which promise much higher APRs for the skilful staker.

An Annualised Percentage Rate

To fairly compare each source of staking reward, we will first recover the raw, ETH-denominated payment to a staker for each activity for every slot and epoch observed in our data. We then calculate the return that this generates on an assumed balance of 32 ETH - the minimum required stake by an Ethereum staker. This figure is annualised by multiplying it by the number of periods that a validator can be expected to receive that reward per year.

The rewards granted for making votes, Chain Head, Source, and Target votes are available once every epoch and the size of the reward or penalty granted is dependent only on the validator’s performance. For these rewards, we multiply by 82125 (the number of epochs in a year), which assumes no compounding on rewards.

Other rewards, such as those earned by proposers and sync committee members are only available at random opportunities. In these cases, we will instead display the distribution of rewards scaled by the expected number of opportunities to earn them in a year and quote the unconditional distribution’s mean and standard deviation by applying the laws of total expectation and variance respectively.

Empirically, we recorded the following statistics of the rewards earned since theMerge. Some historical values are larger than the values quoted above because there has been an increasing number of validators activating nodes on the Beacon chain, and some are lower due to less-than-optimal performance by validators.

Making Votes‍

Rewards are most commonly earned from Chain Head, Source, and Target votes.

‍Every 32 slots, the full set of active validators are spread evenly across each slot that makes up the epoch.

A new block is created in each slot and proposed to the validators that have been assigned to it. This means that each validator will be required to perform this vote (and has the opportunity to earn its rewards) in one slot per epoch. In that slot, they are asked to make two types of votes:

*From an observed validator set size of 650,471, recorded at epoch 213,000 - 04:00 UTC 6th July 2023.

The Finality Votes (source & target votes) are sent together as a pair of votes that are rewarded separately.

Chain Head Votes - Voting on the most recent valid block.

This vote (which we call the Chain Head Vote) asks validators to choose the latest valid block that they have heard proposed so far. They will do so after listening for anew block that is proposed in their slot and considering if:

• They have heard the newly proposed block
• They have heard the newly proposed block in time
• The transactions in the block are valid
• They agree with the choice of an existing block to which to append the new block

If the new block satisfies all conditions, they will vote for it as the head of the chain.Otherwise, the validator can vote for any other block proposed in the past. The validators who voted for the winning block are rewarded when the votes they cast are included in the block that is proposed in the very next slot (that itself agrees with the view of the vote).

Their vote must be heard by the proposer within 12 seconds to be included in the very next block and earn a reward. It doesn’t matter if their vote was not included in this block because the vote was late, the proposer in the immediately subsequent block did not propose a block, or the block that was proposed was on a minority chain or otherwise invalid: the chain head vote is worthless and earns no rewards for the validator who made it.

If the validators make their vote within 12 seconds and it is successfully included in the block that was proposed in the immediately subsequent slot, then they are entitled to the maximum reward, which is dependent on the number of validators that were active, N, during that epoch and the proportion of validators that agreed with their vote.

Finality Votes - What stops validators from reverting lots of transactions?

To stop validators from suddenly choosing blocks very far in the past as the head of the chain (thereby undoing all of the transactions in the blocks that have been added to the chain since that block) the Beacon chain also asks validators to periodically “lock-in” chunks of blocks that have already been validated - and calls these blocks finalised.

This means that the network cannot later pick a block upstream of this block as the head of the chain without incurring a significant penalty to staked ETH balances. This does not stop reversions from happening - but it does make them prohibitively expensive. Sent to the network at the same slot in the epoch as their vote to determine the head of the chain, the two finality votes asked of validators are:

1. A Source Vote that picks the block at the start of the chunk of blocks to be finalised.
2. A Target Vote that picks the end of the chunk of blocks to finalise.

Validators are rewarded for making these votes in the same way that they are for chain head votes, but if they do not include a correct vote within the required number of slots they can be penalised by having ETH taken and burned from their staked balance. The size of the rewards for valid source and target votes is the same as the size of the penalties for late or invalid votes.

To earn the source reward, the vote message must include a valid source vote and be included in a block proposed in one of the next 5 slots - otherwise, the validator is penalised by the source reward amount. To earn the target reward, the vote message must include both a valid source vote and a valid target vote, and be included in a block proposed in one of the next 32 slots. If the source vote is incorrect, then the validator receives both penalties and no rewards.

The theoretical maximum reward paid for source & target votes (currently 2.36%*) is noticeably higher than those paid for chain head votes (now 0.83%*). This corresponds to the importance placed by the protocol on agreeing on finality for the Beacon chain.

*From an observed validator set size of 650,471, recorded at epoch 213,000 - 04:00 UTC 6th July 2023.

Sync Committees - Lightening the load on nodes.

Ethereum’s Beacon chain allows some nodes to run light clients which require nodes to store far less data. Those clients do not have access to data regarding all validators on the network and so are unable to verify blocks based on validator votes. Instead, light nodes can verify the head of the chain simply by checking the signatures of the members of a Sync Committee.

Sync committees are a randomly selected subset of 512 validators that earn extra rewards by signing block headers that are at the head of the Beacon chain and broadcasting it to the network - much like the chain head votes that each validator makes once per epoch. They do this at each slot for 256 consecutive epochs (just over 27 hours) before another 512 validators are selected to perform sync committee duties for the following 256 epochs.

Members can vote for any block they want as the head of the chain without incurring penalties - as long as their signatures are included onchain in the subsequent block. The penalty they receive for missing a vote is equal in magnitude to their potential reward, both of which are dependent on the total number of validators on the network as well as their total effective balance. Unlike chain head and source & target rewards, the size of sync committee rewards increases with the number of active validators. However, the probability of being picked as a sync committee member decreases.

Including Votes in Blocks - Earns larger rewards, but happens less frequently.

The blocks that are proposed to the network of validators in each slot are created by a single randomly selected validator, the Proposer. It is their job to listen to transactions submitted by users, validate them, and include them in a block that is voted on by the other validators. They are also rewarded for collecting chain head, source, and target votes that were made by validators voting on blocks proposed in previous slots and including them in the blocks they propose.

The reward they earn is directly proportional to the rewards paid to validators for making those votes. As most of the rewards earned by validators are dependent on the number of votes made by other validators that agree, so too are the rewards earned by a proposer for recording them. Occasionally there is the opportunity to earn higher rewards for this action, when the previous block was missed and there are more unrecorded votes for the proposer to include.

The rewards earned by proposers are just as deterministic as the rewards that we have discussed so far. It is the probability of being chosen to perform the tasks that earn the rewards that is probabilistic. Validators are chosen for this task according to the proportion of the total staked balance they control on the Beacon chain, so (assuming that each validator has staked 32 ETH each) can expect to have the opportunity to propose a block 4.04* times a year.

*From an observed validator set size of 650,471, recorded at epoch 213,000 - 04:00 UTC 6th July 2023.

The Good, the Bad, and the Data - Stories from the data in our Consensus Layer Analysis

Highest expected APR for validators

Target votes provide the highest expected APR for validators.

• Target votes offered the highest expected rewards of all validating duties, with source and chain head votes close behind with similar distributions.
• The nominal reward available to a proposer for performing proposer duties is much higher than the rewards granted for performing routine voting actions (head, source and target votes).
• The small probability of being chosen to propose a block at all means that the expected APR for including votes was lower than that of chain head votes alone at just 0.51%, compared to 0.84%.
• This is also true of sync committee rewards, which earn validators a nominally larger amount of ETH (if they have been picked) than any other vote they cast on the network.

Validators missed potential vote inclusion rewards the most.

• If a proposer misses their chance to include a valid block in their slot, the votes that it would have included can be picked up and included in a block by a later proposer. Including the extra missed source & target rewards provide a boost to the maximum reward a proposer can earn.
• Whilst blocks are regularly missed, it is not the intended behaviour of the chain - typically, the maximum reward available to a proposer is just under half of this value.
• The variance between those levels shows blocks that have included some-but-not-all of the votes that were missed in the slot before it.

Beacon Chain finality issues

• In May, a bug in a client implementation of the Beacon chain protocol caused validators to disagree on which block was at the head of the chain.
• This meant that fewer validators were able to submit a timely vote and fewer validators correctly voted in the same way as the eventual winner of the slot vote.
• As a result, the average reward paid to validators was lower due to more validators missing their 12-second vote time slot and fewer voters voting for the same blocks, with penalties applied in the most extreme cases of missed source & target votes.

• One cause of the loss in rewards was the drop in the participation rate of validators across the network.
• A lower participation rate on the network indicates that there were fewer validators voting in the same direction as the “winning” votes.

‍‍Proposer missed slots

Proposers missed their slot more often early in the epoch.

• Proposers selected by the network are allocated to produce blocks in one of the 32 slots available in an epoch.
• We found that proposers were more likely to miss their slot earlier in the epoch - with the second slot the most commonly missed slot.
• Three slots in, or after 36 seconds, proposers were much more reliable in completing their proposal duties on time.