The Anatomy and Future Prospects for Staking

This article intends to be a quick guide for the current staking sector with a focus on Ethereum, the associated misconceptions and what the future of staking may look like.

Key Takeaways:

  • Ethereum’s staking yield is akin to the FED Funds rate for DeFi
  • Proof of Stake is not a consensus mechanism
  • Staking “yield” is mostly a made-up concept for an easier mental model
  • Liquid staking tokens provide tangible advantages over solo-staking
  • Squad staking will help solo-stakers and promote greater decentralisation
  • Ethereum may put a limit on the total number of validators
  • Restaking (and Liquid Restaking) are potential game changers

In 2021, Morgan Stanley predicted that staking could be a $40bn industry by 2025 and as the understanding around Proof of Stake mechanisms evolves, this forecast seems more likely. Below we show the current size of the staking market by blockchain.

With the successful completion of the Shapella upgrade, Ethereum is now a fully-fledged Proof of Stake network and is by far the largest. There are currently 575k validators helping to secure the Ethereum network with an additional 56k patiently waiting to join over the next month. Now that withdrawals have been enabled, it can be argued that the Ethereum staking yield is akin to the FED Funds Rate — it essentially sets a floor rate for the borrowing of ETH within the DeFi ecosystem. Why would one lend their ETH (an inherently risky activity) for less than the staking yield? If borrowing rates are lower, participants can borrow ETH for a lower rate, stake the ETH and keep the spread, this will drive up the demand for borrowing until the rates reach near par. As shown below, the issuance rewards provide an average annual yield of ~4.2% and this will continue to drop as more validators join the network.

But before we proceed further, what is Proof of Stake? Many believe it is a type of consensus mechanism but in reality, it is only part of one.

Proof of Stake is a Sybil Resistance Mechanism

Proof of Stake (like Proof of Work) is used to protect against Sybil attacks. A Sybil attack is a form of assault against a computer network service in which the attacker creates many fictitious identities and uses them to wield a disproportionately large influence over the service. An example of this attack would be bots voting on a poll on Twitter. Twitter’s Sybil resistance mechanism is to require an email address when signing up or other forms of KYC (obviously not effective enough as bots can do this as well). However, Proof of Work requires proof that an enormous amount of work has been done before adding to the chain. Ethereum’s Sybil resistance requires validators to stake 32 ETH (also not an easy thing to do). Combine the fact that bad actors will see their stake slashed if they misbehave, leaving a more crypto-economically secure protocol.

Now that we’ve tackled one misconception, let’s quickly cover another — “staking yield”.

Misconceptions of a Staking “Yield”

From a high-level view, stakers can expect to receive ETH rewards from two sources:

i) the consensus layer

ii) the execution layer.

As the names imply, the consensus layer rewards validators who complete consensus-related tasks while the execution layer rewards validators for the execution of transactions. For greater detail on these rewards see here but for consensus tasks, over the long term, a validator’s expected proportion of rewards earned for each task are as follows:

However, despite what many believe validators don’t receive a “yield”, instead, validators are randomly chosen (proportional to how much ETH they have staked) to perform certain tasks. Depending on the task and the timeliness of its completion this leads to the validator being rewarded with a nominal amount of ETH regardless of how much ETH they have staked. The fact that the rewards are completely separate from the amount of staked ETH makes the notion of a yield unfit as a technical descriptor. The ETH itself is not being used to generate returns, rather it’s used as “access rights” to be allowed to perform tasks in exchange for rewards. However, if a validator can probabilistically expect to earn 2 ETH per year by locking up 32 ETH then the idea of a “yield” makes for an easier mental model when referencing a longer time span. However, the “pooling” of staked ETH muddies the water even further.

Staking Pools

The idea of a staking pool is not well defined but we will use the term here to describe any type of staking that explicitly relies on more than one participant. For the first type of pooled staking, we will reference centralised staking providers (think CoinBase or Binance).

Centralised Staking Providers

Centralised entities make it easier to contribute towards staking on Ethereum through an easy-to-use interface, eliminating the need for specific hardware or know-how to run a validator and lowering the amount of ETH needed to contribute towards staking (from 32 ETH to as little as a fraction of 1). Another overlooked advantage of these centralised services is the smoothing of rewards — if any of the validators in this pool are randomly chosen, the rewards are distributed evenly among all participants. This makes the payouts less sporadic and provides a more consistent revenue stream — this is yet another reason why the term “yield” is thrown around so much.

Even though the minimum amount of ETH required to stake on the network is 32, the “maximum” is also 32 ETH. What this means is that a minimum of 32 ETH is required to spin up a validator but staking more than 32 in one validator, while possible, has no advantage — a second validator would need to be created in order to make use of the additional 32 ETH. This means that ETH itself isn’t “pooled” but rather the validators are. However, it should be noted that different centralised staking providers operate their services differently. Some may indeed pool ETH, choose when to pay out rewards or adjust any other dynamics they see fit. The regulatory concerns surrounding increased deviation from Ethereum’s native protocol are out of the scope of this report.

Liquid Staking Tokens (LSTs)

Liquid staking tokens offer a similar service to their centralised counterparts — easy to use, no hardware or operator knowledge needed, removing the 32 ETH requirement and smoothing of rewards. In addition to these benefits, it is also permissionless (anyone can buy the token on the secondary market) and can be used within DeFi (as collateral or otherwise). Additionally, rewards from staking are then used to create more validators, effectively compounding the rewards which a solo staker with “only” one validator would be unable to do.

Similar to centralised staking providers, the paths taken by the competing Liquid Staking DAOs may differ. There are different incentive mechanisms, fee take rates, operator selection methodologies and governance procedures. In fact, Lido’s recent upgrade introduced a staking router architecture which allows for an even more diverse group of staking operators who can each further differentiate their services even within the protocol. This was enabled through DVT which will be discussed in greater detail later on. Holders of Lido’s stETH can also natively exchange their stETH for ETH much more quickly than a solo-staker withdrawing from the Beacon Chain thanks to an internal buffer. The combination of all the benefits stated above is why LSTs are the most popular form of staking right now (~36% market share).

Solo Staking

Around 15–25% of all staked ETH is operated by individuals or institutions. These entities need to be tech-savvy enough and possess at least 32 ETH in order to self-stake. Below we outline the benefits and drawbacks of solo staking vs “pooled” staking.

Despite the disadvantages, solo-stakers are arguably the most important validators since they contribute the most towards the decentralisation of Ethereum. In order to prevent staking from predominantly becoming an industrialised enterprise, solo-staking needs to be easier and less risky. This is where Squad Staking enters the picture, trying to achieve even greater decentralisation for both solo and pooled staking.

Squad Staking

The term “Squad Staking” emerged from the arrival of a novel technology known as Distributed Validator Technology or DVT. DVT, as the name suggests, is a technology that allows the process of validation to be performed on a distributed number of devices as opposed to a single machine. It allows for multiple participants to come together and share control over one (or more) validator(s) hence the term “squad”. This reduces the requirement of 32 ETH since, for example, 4 participants each with 8 ETH can now cooperate with each other and spin up one validator. Secondly, it reduces downtime risk since if one of the nodes goes offline there are three more backups. This reduces the chances of missing rewards or being penalised and provides greater resilience and stability to the Ethereum network as a whole. DVT uses a combination of four key technologies:

Distributed Key Generation Scheme (DKGs)

The private key is encrypted, split into N parts and distributed among N participants. This ensures that no one group member has full control over the validator.

Shamir’s Secret Sharing

This is an algorithm that allows the reconstruction of the private key without requiring the knowledge of every key share. For example, if there are three operators and the threshold for signing is 2 out of 3, then even if one computer goes offline or is not trustworthy, two nodes remain active and can sign messages.

Multi-party computation (MPC)

Multi-party Computation allows the cluster of operators to sign messages and perform computations without rebuilding the complete private key on a single device. This eliminates the risk of centralising the private key, thereby maintaining the decentralisation of the validator setup.

Istanbul’s Byzantine Fault Tolerance

This algorithm establishes a consensus on a block by designating one of the nodes as the leader who will propose the block to the network. If the leader is offline then another leader will be chosen before the proposal round is finished.

DVTs will allow more individuals to participate in consensus while reducing some of the risks associated with solo-staking. Below we highlight the benefits and drawbacks of DVTs.

As there are multiple nodes assigned to one validator, this increases the surface area for things to go wrong. It also introduces some latency in communication although this can be mitigated by using direct peer-to-peer communication rather than a singular gossip network.

DVTs can also be offered through a liquid staking provider (as Lido and Rocketpool have done) which adds benefits such as increased liquidity, smoothing and compounding of rewards but still mitigates some of the drawbacks of traditional LSTs (no control over keys or assets). Centralised exchanges can even use DVTs to offer their customers staking in a more decentralised (and potentially regulatory-compliant) manner. Obol Network is expected to launch its instance of this technology later this year while the SSV network and Stakewise are already live on the mainnet. Expect DVT or “Squad Staking” to become one of, if not the, most popular form of staking on the Ethereum network.


Re-staking is probably one of the most anticipated yet controversial applications coming to the Ethereum ecosystem. Once again, as the name implies, re-staking involves taking staked ETH and staking it again but for another purpose outside of validating Ethereum blocks. Restaking will also be available for liquid staked tokens like Lido’s stETH, Rocket Pool’s rETH and others. This would mean that an Ethereum validator would perform additional services for another customer (besides Ethereum) and in turn be rewarded for these additional services but they would also open themselves up to additional slashing risks if they act maliciously while providing these services. The optionality that re-staking offers is considerably large, it can be used for:

Bridges — instead of trusting a single bridge operator who may steal your funds, one can choose to trust a bridge operated by restakers who will get slashed if anything goes wrong.

More formally, restakers can verify off-chain whether bridge inputs are indeed correct, and if enough restakers agree, then the inputs are accepted. But if they are accepted and wrong then when someone challenges, the input can be reverified and restakers will be slashed.

Scaling Layer-2s — key components of scaling the Ethereum network through Layer-2s are Data Availability and Sequencers. Data availability ensures that enough data is posted to the chain such that the state can be recreated and verified. If not enough data is posted then restakers will be slashed. Sequencers order the transactions coming from a Layer-2 which give them a lot of power to act maliciously if they choose to censor or front-run users’ transactions. Once again, if the sequencer role is performed by a quorum of restakers, then their ETH will be on the line, incentivising good behaviour.

Securing other blockchains — Restaking also enables validators’ staked ETH to be used as crypto-economic security for alternative protocols, in exchange for protocol fees and rewards while maintaining the slashing risks.

The above examples are only a fraction of the possibilities, restaking can be used for strengthening trust in oracles, MEV Management, creating a futures market for blocks, guarantees on inclusion of event-driven actions and much more. The design space for these services is limited by the imagination. Validators can opt-in to whichever service(s) they feel comfortable providing while other validators can delegate their staked ETH to the ones who choose to provide these services. Liquid staking tokens are also likely to adapt to these offerings — even now one can purchase an ETF like LST which derives its yield from a diversified basket of other LSTs. In this same line of thinking, we can imagine LSTs offering different types of restaked products each with different risk/reward structures.

However, there are concerns surrounding the introduction of restaking as the Ethereum protocol would be unaware of this middleware running on top of it. For instance, if there is a network issue and this causes a large portion of restakers to be slashed then this could reduce the security of the Ethereum network as stakers are then force-exited due to having less than the required 32 ETH. The rehypothecation of crypto-economic security is a new concept which requires further formal research and testing to fully understand but we will briefly touch on the high-level concepts here.

As it currently stands, Ethereum is currently securing around $400bn-$500bn with $33bn of staked ETH (implying a 13x security leverage), this works because the total value secured is not what is protected by staked ETH. Rather, the total value transferable out of the system in a short period of time during the “attack duration” is what is being protected by staking. Using this framework one can get provable guarantees on what the crypto-economic conditions are for safety but this is beyond the scope of this report. To further ease any qualms with restaking the concept of dual staking arises, where a separate token is created that takes the leverage security risk away from staked ETH.

Restaking also changes the incentives for staking. Ethereum has an ethos of minimum viable issuance, that is what is the minimum amount of issuance required to keep the network secure. As more validators enter the network more ETH is issued but ETH rewards per validator decrease (ceteris paribus), this reduces the incentive for more validators to join until an equilibrium is reached. If validators can then earn additional rewards through restaking this morphs the incentive structure and may not be ideal for the goal of minimum viable issuance. Although this additional incentive layer may be mitigated by the Ethereum network potentially limiting the total number of validators in the future in an effort for faster finality.


Restaking is clearly both an exciting and challenging endeavour and is currently being pioneered by EigenLayer. The introduction of this middleware to the blockchain landscape will shape the spectrum of trust as traditionally centralised applications can borrow Ethereum’s security to enhance their offerings. Staking is a relatively new industry yet the pace of progress is incredibly fast, in a short period of time we already have delegated PoS, liquid staking and DVTs with restaking, liquid restaking and dual staking coming soon. While there are many questions still to be answered, one thing we can be certain of is the level of innovation in this space continuing to grow.

The Anatomy and Future Prospects for Staking was originally published in CoinShares Blog on Medium, where people are continuing the conversation by highlighting and responding to this story.

2023 Global Ether Ownership Overview

Summary of Key Findings

  • Ethereum is not primarily a Western phenomenon
  • The proportion of the population owning ether is consistently reported higher in emerging and frontier markets across independent studies
  • It is probable that more than a hundred million people own ether
  • Ownership is growing rapidly and has been growing rapidly for as long as we have data to measure it
  • Most people use ether as an instrument of speculation
  • A large majority of ether owners are under the age of 45


We believe that the investment case for ether rests on the ability of ether demand for code execution, DeFi collateralisation, and store of value to outstrip the supply created by staking.

But before developing any understanding of where and with whom ether usage is likely to grow in the future, it’s very helpful to start with an understanding of where that ownership and adoption sits today.

We’ve therefore set out to answer the question ‘Who owns ether and why?’ using all publicly available research we can find which bears on the topic. As such, this paper serves as a literature review of current ether (and as you’ll see, crypto) ownership research. We first see what we can say with confidence, what the data more tenuously suggests and finally what is wrong with current research and what needs to be improved in future efforts.

Once we have a better understanding of where demand has already come from, we will be in a better position to make reasonable assumptions about where it might come from in the future. And so we aim to first determine how many people own ether right now, how this number has developed over time, and also say something about who these people are and why they have chosen to own it.


As mentioned, this study is a literature review of all the sources we could find that would in some way inform us about ether ownership. Specific studies on ether however are scarce, and most studies use the much broader blanket terms ‘crypto’, ‘cryptocurrency’, ‘digital assets’, or similar. This has meant that most of our data, especially on ownership, is of a more general kind, and some assumptions have to be made. In total, we found 22 studies, most of them survey-based, that in some way informed us on cryptocurrency ownership, owner demographics, and motivations behind ownership.


Our largest assumption considers the relationship between cryptocurrency ownership and ether ownership. In order to translate the former to the latter, we assumed that the market cap dominance of ether relative to the broader cryptocurrency market is a conservative lower bound of its proportional ownership relationship to cryptocurrency ownership.

The reason we believe using dominance as a proxy like this is conservative is that it effectively assumes people only own one cryptocurrency whereas the real tendency is for people to own more than one. In the few studies, we found that actually ranked the propensity of crypto owners to own ether, their percent figures also tended to be higher than the dominance figures.

Our dominance figures were sourced from CoinMarketCap and averaged by year.


While there are quite a few studies containing some ownership data, the data points are scattered both in time and geography. When collating the data, we parsed out all data points we could find, expressed as a percentage of a given country’s population, dated them by year and added them to our own database. We then did a simple arithmetic average for any given country that had multiple data points within a single year to create single percentage ownership data points for each surveyed country each year.

Country-based percentage estimates were then multiplied by their individual populations and aggregated to form a total estimate of owners per year. It is important to note here that there was only one year where most countries had individual estimates, 2022, with 148 countries represented, covering 7.3bn people. Out of the remaining years, 2018 and 2021 had the second most estimates, both at 21. The populations of the countries sampled in those years amounted to approximately 3bn and 4bn, respectively.

For the year 2020, individual country data points were simply too scarce to reliably inform us on ownership percentages, so we used the global ownership figure from Cambridge’s 3rd Global Cryptoasset Benchmarking Study wholesale and adjusted it for dominance. The estimate given in that study represents their best estimate of a lower bound of global crypto users and is based on a methodology of summing individually confirmed accounts on crypto exchanges globally.

Finally, in order to analyse usage based on economic conditions, we assigned all countries according to the MSCI classification, which focuses on splitting countries into categories based on economic development and market accessibility.

Demographics and Ownership Motivation

Demographic and motivation data were much less structured and here, rather than being able to generally collate and meta-analyse data, we had to take a more artful approach and make some broader and less precise generalisations based on the data we could find.

The one instance where we are able to perform some meta-analysis is in considering the breakdown between different monetary use cases. Here we took a collection of survey data, and bucketed it into our best interpretation of whether the listed motivation best fitted a Medium of Exchange, Speculative or Savings (Store of Value) use case. We recognise that there is an amount of interpretation incorporated into that exercise but we found no better way than to simply apply our best judgement.

Limitations and Sources of Error

Major limitations of our review are the limited amount of available data, the fractured nature of the data available, and the difference in the metrics explored in the source material. Methodologies, when available, are often vague and hard to evaluate properly.

Ownership numbers are also scarcer than the number of data points themselves suggest. For example, although we have over 282 data points for ownership by country, 227 of them are from 2022 only. These 227 are from 4 sources — Triple-A (146), Finbold (47), Kucoin (7) and Finder (26) — and the variance between the sources is substantial, although not extreme.

Out of these four studies, the largest one, Triple-A, links its primary source to a dead web page. It also uses several of the other studies, Finbold, Bank of Canada (BoC), Cambridge, and particularly, Chainalysis, as partial input sources. Chainalysis in turn relies entirely on on-chain data and automated heuristics. The Kucoin methodology is unclear regarding how they arrive at their ownership estimates from their samples, and the report from Finder uses data from a third party ( whose exact methodology we were unable to find. They are however a large, experienced and professional audience-targeting company which gives us some confidence in their methodology.

Even though we have estimates for 146 different countries in 2022, only a couple dozen countries have actually been sampled by surveys. The rest have estimates calculated from a mix between ownership in similar countries, and their weighting in the Chainalysis report. However, out of the top 20 countries by total ownership, only 5 were not directly surveyed in 2022 (China, Turkey, Pakistan, Thailand and Russia), China, with 28m estimated owners, is the most impactful one on the overall global ownership estimate.

From what is claimed in the available methodologies, we suspect the best ownership data comes from Finder and Cambridge. Unfortunately, Cambridge only gives us 3 total data points on ownership, and they are only for the world as a whole. Finder gives us 26 points, all claimed to be survey-based. Interestingly, Finder’s estimates are all in the upper range of our ownership data points.

Overall, we find confidence in the fact that all the different studies are roughly similar. There are no order-of-magnitude outliers and so we find it hard to believe that our results are very far removed from reality.

Results and Discussion

We have looked at ownership from a few different angles: Global Ownership, Ownership by Country, Ownership by Country Classification, and Ownership Demographics.

We consider Global Ownership to be the most informative metric on the overall progression of ether adoption, Ownership by Country and Country Classification tells a story about how political and economic conditions influence adoption, and Ownership Demographics give us some more detail on the type of person who has thus far chosen to become an ether owner.

Global Ownership

Our results suggest that the number of ether users worldwide is growing at a solid pace. For 2022 specifically, the estimate is 115m global users, with a standard deviation of about 12m. This translates to a compound annual growth rate of 30% since 2018. The best trendline fit is an exponential curve, but the R2 is quite poor.

While the data is perhaps too uncertain to say that the numbers (or even the error bars) are very accurate, we believe that it is quite likely that at least a hundred million people own ether. If our estimates are in fact fairly accurate, market cycles seem to play into the adoption curve, causing rapid booms around bull markets and then cooling off in the following bear markets. However, given the fragmented nature of our data, we would be cautious about drawing tight conclusions on a year-to-year basis.

As mentioned in our methodology section, however, the raw data from our selected studies tended to report on ‘crypto’ ownership, not ether specifically. When given only results for crypto, we, therefore, made the assumption that ether ownership as a subset of crypto ownership followed its share of the overall crypto market cap.

Looking at the ether dominance chart we can see that shifts in dominance percentage are one of the factors that play into the seemingly large intra-year volatility of ether ownership and readers should be aware of this as it possibly exaggerates the apparent effects of reduced ownership during the bear market conditions of 2018–2019.

Ownership is Highest in Emerging and Frontier Markets

A key observation is that the reported amount of ether owners in each country per capita is strongly skewed toward emerging and frontier market countries. Out of the top 20, only 4 countries were developed, and the remaining ones were emerging (10), frontier (4) and standalone (2).

It should not come as a surprise that on a percentage basis, ownership is concentrated where the need is the highest. Three of the top five countries have not seen single-digit inflation rates in over five years and one has experienced hyperinflation. Many of the remaining countries on the list have suffered financial repression of varying severity, and have large populations of young, technologically savvy yet often unbanked populations.

To add some further context to the magnitude of these ownership percentages, according to Bloomberg, 3.7% of Indians are invested in equities, whereas our data suggests that almost 2.8% of Indians own ether.

Without any other requirements than an internet connection and a smartphone, Ethereum gives users access to a whole range of financial services such as internet commerce, savings outside of local currency, staking, borrowing, lending, international transfers etc. For a large proportion of the global population, access to financial services is severely lacking, making Ethereum an attractive method of accessing otherwise unavailable necessities.

In this dynamic, we recognise an issue of perception that we’re often faced with when talking to clients in Western audiences. Due to the relatively well-functioning financial systems, we’re all accustomed to, westerners often tend to not immediately understand Ethereum’s beneficial properties, and even if they do, they often see no need for them.

This might not be the case among the over 4bn people worldwide living under authoritarian regimes. It is likely not a coincidence that ether ownership is concentrated in the areas where its properties compare quite favourably to the often substandard local currencies and financial systems. Among these populations, Ethereum’s beneficial properties are much more visceral and intuitive.

In no chart is this more clearly demonstrated than when we rank countries by the total number of owners. Not only are emerging and frontier countries overrepresented in percentage-based usage, but they also represent the vast majority of the global population. The result is that 86% of total ether owners live outside of developed countries.

Ether is Increasingly Used as a Store of Value, and Less as a Medium of Exchange or Instrument of Speculation

While we weren’t able to find good survey data on ether use cases, the nature of Ethereum and its ecosystem allows us to use open-source data to make some informed assumptions regarding use cases based on the volume of use of certain on-chain services.

For example, we can use ether age bands to classify usage into either a Medium of Exchange (MoE), Speculative, or Store of Value (SoV) bucket. In this case, coins that have moved over the last year are considered to be predominantly used as an MoE, coins between 1 and 5 years are Speculative, and coins older than 5 years, staked coins and coins used as DeFi collateral are all bucketed as SoV.

The boundaries we’ve chosen here will inevitably be both blurry and somewhat arbitrary. But again, where no obvious methodology presents itself, the best option is simply to classify these buckets as well as we can and make readers aware of the assumptions we’ve made.

Although ether has been primarily used as a medium of exchange for transactions on the Ethereum network, it has not widely been considered a traditional store of value like gold or Bitcoin. However, some proponents of ether argue that its unique features make it a potentially attractive store of value. Ether’s usage seems to be trending more towards a store of value the longer it exists and proliferates across the globe.

We can also observe from Ethereum coin age bands that the length of time ether owners are holding onto their ether continues to grow. In fact, “1–3 year holders” doubled from 2016 at 20% to now 40% while “3–5 year holders” more than tripled from mid-2020 at 3% to now 11%. As shown below, roughly one-fifth (22%) of holders haven’t moved their ether in over 3 years.

We consider ether that is being staked to fall in the SoV bucket as ether that was staked over the past two years were staked with an indefinite lock-up period. And although un-staking will be available in Q2 2023, those who choose to lock their ether up for an extended period for a low single-digit yield are likely using this process as a savings mechanism. Below we show the progression of ether locked up in the staking contract. It shows that despite yields falling to single digits in early 2021, the CAGR of staked ether has been 123% since then.

We also assign ether being used as collateral to the SoV bucket. Ether owners who use it as collateral to borrow stablecoins or other tokens evidently view ether as preferred capital. As of February 2023, approximately 6% of all circulating ether is used as collateral within DeFi applications.

Ether Usage as an MoE is Still Growing in Absolute Terms

We see that while ether’s relative usage as an MoE is declining compared to as an SoV, its absolute usage is still growing. This may have several reasons but the primary one is that as new use cases arise (DeFi, NFTs, Layer2s etc.), the demand and hence price for ether have both grown exponentially as shown below.

Within the Medium of Exchange bucket, the most popular use cases relate to the simple transferring of ETH and the minting/trading of NFTs (which are also priced in ETH). ERC20 token transfers and stablecoins are also popular, making up more than a quarter of all transactions.

Proposal of Further Research

We believe that ongoing ownership research is an important foundation for both academic, business and policy purposes. However, in order to reduce uncertainty and deliver more value to the research, policymaker and business communities, we believe that this research should be standardised, not only in terms of the questions asked, but also the populations sampled and the frequency with which the surveys are taken.

First and foremost, in order for multiple surveys commissioned by separate actors to add value to each other their questionnaires must incorporate some level of standardisation. The same questions should be asked across all populations and across time.

Moreover, questions should not be so vague that they leave researchers with the necessity of introducing too many assumptions in their work. Use of unspecific bucket terms such as ‘crypto’ should be avoided.

In terms of ownership, surveys should ask specifically whether participants own any of the top five/ten cryptocurrencies by market cap at the time of the survey. They should also ask how much users own and how long they have owned them.

Demographic metrics should also be standardised. When asking about age, the age brackets should be the same for all studies. The same goes for income, education etc.

Questions of motivation should be at least so similar that breaking them down by their overarching use case is trivial. The three predominant overarching use cases for cryptocurrencies are medium of exchange, instrument of speculation, and store of value. All questions of ownership motivation should easily be reduced to one of those use cases.

A low-hanging fruit for an enterprising academic institution motivated to add value to this type of research would be the creation of question templates and survey implementation protocols. This way, even if surveys are commissioned by a distributed community of funders, so long as certain standards are adhered to, they will all add incremental value to the total body of available data. This will provide the research community with better source material, and enable meta-studies of higher quality, benefitting academic institutions, policymakers and businesses alike.


Error Calculations

For all countries that have multiple sources reporting ownership in the same year, we calculate the standard deviation of the total data points, measuring how far apart estimates are from the mean of all ownership numbers reported for the same country, in the same year. It is effectively the spread among similar measures of ownership, which in this case is provided as a percentage of a country’s population (“similar” in this context means sources reporting on the same country in the same year, not estimates that are close in value).

Among these countries with multiple sources in the same year, we then derived weightings based on their combined populations. Say that multiple sources reported on the US, the UK, and China in 2019, each country then receives a weight based on their unique share of the total population of all three countries combined. We then multiply each country’s population weight by their corresponding standard deviation. The result is a population-weighted standard deviation given on a percentage base against the populations of each of these countries. We then multiply the percent-based standard deviation by the corresponding population of each country to simply change the units of this margin of error from a percentage of ownership to a specific number of owners.

For all the countries that have only one source in a particular year, we apply a fixed standard error figure, namely the average percent-based standard deviation of the reported ownership numbers across all the countries and all years. In other words, the average standard deviation for the ownership dataset as a whole. We then also apply population weightings for these countries separately in each year, deriving our population-weighted standard error percentages, and multiplying them each by their country’s population.

The sum of the population-based standard deviations is then applied in both the positive and negative error direction in our global estimated Ethereum owners chart as error bars. This applies for each year except 2020, where we use the reported Cambridge statistic as our only estimate of crypto owners — an estimate using non-survey measurements.

The demographic results, besides suffering from a low number of sources, also suffer from a lack of metrics standardisation. In other words, the various studies all ask different questions, and so we have been forced to perform a substantial amount of interpretation in order to normalise their results as much as possible.

Furthermore, we consider it likely that many sources suffer from a significant amount of sampling bias. While several impactful sources did not even list their methodologies, those that did often sampled users electronically over the internet, leaving large unknowns regarding the validity of their sampling

Lastly, several studies asked demographic questions we consider to be leading. This went in ‘both directions’, so to speak, with study designers clearly revealing both positive and negative biases towards cryptocurrencies. For reference, and for the independent review of our readers, we have added all of our source materials below.




The information contained in this document is for general information only. Nothing in this document should be interpreted as constituting an offer of (or any solicitation in connection with) any investment products or services by any member of the CoinShares Group where it may be illegal to do so. Access to any investment products or services of the CoinShares Group is in all cases subject to the applicable laws and regulations relating thereto.

This document is directed at professional and institutional investors. Investments may go up or down in value and you may lose some or all of the amount invested. Past performance is not necessarily a guide to future performance. This document contains historical data. Historical performance is not an indication of future performance and investments may go up and down in value. You cannot invest directly in an index. Fees and expenses have not been included.

Although produced with reasonable care and skill, no representation should be taken as having been given that this document is an exhaustive analysis of all of the considerations which its subject-matter may give rise to. This document fairly represents the opinions and sentiments of CoinShares, as at the date of its issuance but it should be noted that such opinions and sentiments may be revised from time to time, for example in light of experience and further developments, and this document may not necessarily be updated to reflect the same.

The information presented in this document has been developed internally and / or obtained from sources believed to be reliable; however, CoinShares does not guarantee the accuracy, adequacy or completeness of such information. Predictions, opinions and other information contained in this document are subject to change continually and without notice of any kind and may no longer be true after the date indicated. Third party data providers make no warranties or representation of any kind in relation to the use of any of their data in this document. CoinShares does not accept any liability whatsoever for any direct, indirect or consequential loss arising from any use of this document or its contents.

Any forward-looking statements speak only as of the date they are made, and CoinShares assumes no duty to, and does not undertake, to update forward-looking statements. Forward-looking statements are subject to numerous assumptions, risks and uncertainties, which change over time. Nothing within this document constitutes (or should be construed as being) investment, legal, tax or other advice. This document should not be used as the basis for any investment decision(s) which a reader thereof may be considering. Any potential investor in digital assets, even if experienced and affluent, is strongly recommended to seek independent financial advice upon the merits of the same in the context of their own unique circumstances.

This document is directed at, and only made available to, professional clients and eligible counterparties. For UK investors: CoinShares Capital Markets (UK) Limited is an appointed representative of Strata Global Limited which is authorised and regulated by the Financial Conduct Authority (FRN 563834). The address of CoinShares Capital Markets (UK) Limited is 82 Baker Street, London, W1U 6TE. For EU investors: CoinShares AM ( is a French asset management company regulated by the Autorité des Marchés Financiers (AMF), registered under number GP-19000015 since 27/03/2019. Its office is located at 25 rue du 4 Septembre, 75002 Paris, France.

Copyright © 2023 CoinShares. All rights reserved.

2023 Global Ether Ownership Overview was originally published in CoinShares Blog on Medium, where people are continuing the conversation by highlighting and responding to this story.

Ethereum’s UpcomingWithdrawal Upgrade

The much-anticipated Shapella upgrades for Ethereum will finally be implemented on the mainnet on April 12. This will enable those who staked their ETH to withdraw their coins for the first time.

The name of the upgrade, Shapella, is a play on the words Shanghai and Capella where Shanghai is the name of the upgrade on the execution layer while Capella is the name of the upgrade on the consensus layer (Beacon chain).

This upgrade is the long-awaited final piece for Ethereum to become a fully-fledged Proof of Stake network. With over 560,000 validators on the network, this withdrawal feature is a welcomed addition to the Ethereum network. As each validator must stake 32 ETH, this equates to roughly 18 million ether or ~15% of Ethereum’s supply, worth roughly $34B.

However, there are concerns that once ether withdrawals are enabled, this will lead to immediate selling pressure and a subsequent decline in the ether price. To estimate the supposed sell pressure we first need to understand who the validators are and what the procedure for withdrawals is. For the withdrawals process, there are two types of withdrawals, partial and full.

Partial withdrawals: Only the staking rewards are withdrawn (everything in excess of 32 ETH). These withdrawn rewards can be spent immediately. The user will continue to validate as expected.

Full withdrawals: The validator will exit and stop validating. The validator’s entire balance (32 ETH principle and any rewards) is then unlocked and allowed to be spent after the exit and withdrawal mechanism is complete.

Partial withdrawals happen automatically once per week while a full withdrawal is only a manual process. For full withdrawals, there are three phases:

1) Wait to be picked up by a validator (16 every 12 seconds or 115,2000 per day) so this will likely not be long.

2) Wait for your turn in the queue — depending on how many validators are on the Ethereum network a different number of full withdrawals is possible. At 560,000 validators, 8 validator withdrawals can be processed per epoch or 1,800 validators.

3) Wait for the fixed protocol delay of ~27 hours (256 epochs).

Now with an understanding of the withdrawals process, we look at who the validators are. We break out the Ethereum validators into three main cohorts.

Cohort 1: Liquid Staking

~60% of ETH staked is done through liquid staking tokens (via a third party) so if owners wanted to exit, they could sell their tokens on the market instantly at a 1:1 ratio.

Cohort 2: Residential Staking

~32% market share and this cohort (those who had 32 ETH+ and were tech-savvy enough to self-stake) is unlikely to be the one who wants to withdraw all their ETH to sell.

Cohort 3: Other

~8% market share for the remaining group of validators.

With the understanding that withdrawals won’t be immediate but rather an orderly queue of first come first serve we can estimate the length of this queue. And while no one can say with certainty how long the queue will be, we outline a framework for estimating its size and time.

Assuming 2% of Cohort 1, 5% of Cohort 2 and 10% of Cohort 3 want to fully withdraw, this equates to roughly 22,000 validators exiting. This excludes partial withdrawals which based on the current balances would take ~100 hours to clear. So, the queue to exit 22,000 validators would take roughly 14 days. Furthermore, accounting for error bands, and reshuffling from both Kraken (SEC lawsuit) and Lido (community wants to see their market share reduced), the queue for withdrawing ETH may range from 1 to 4 weeks. However, we believe the queue will err towards the shorter end of this range.

As soon as the Shapella upgrade is finalised and rolled out, Ethereum developers can switch their focus to the next upgrade, Cancun which is focused on scaling Ethereum by making layer-2s an order of magnitude cheaper. Cancun is arguably a more important milestone for Ethereum which desperately needs to scale its network to achieve its vision of being the world’s settlement layer for the Internet of Value.

Ethereum’s UpcomingWithdrawal Upgrade was originally published in CoinShares Blog on Medium, where people are continuing the conversation by highlighting and responding to this story.