A recurrent question in legal discussions about crypto activities is whether we should create bespoke regimes or try to fit crypto within existing rules at the risk of disregarding the nuances of a novel and developing technology. This theme is also evident in the discussions around the taxation of staking and lending in decentralised finance (DeFi), which was the subject of a recent Call for Evidence published by HM Revenue and Customs (HMRC), the UK tax office.
HMRC has acknowledged industry concerns that in certain circumstances the current rules for capital gains tax (CGT), when applied to DeFi staking and lending, are inconsistent with the substance of the activity. This issue arises where the rules treat transactions as tax disposals although the effective economic ownership of cryptoassets is retained. This can result in a tax liability where no corresponding gain has been realised. To address this, HMRC has proposed to either apply the existing tax regime for securities lending to DeFi staking and lending, or create new tax rules specifically for DeFi activities. As argued in this article, the first option would ignore the substance of liquid staking and create undesirable knock-on effects.
This article focuses on liquid staking on a Proof of Stake (PoS) blockchain such as Ethereum – and, more specifically, on the concerns that under current rules stakers could be liable for CGT when (i) depositing staked tokens in exchange for receiving liquid staking tokens and (ii) subsequently exchanging liquid staking tokens for an equivalent value of staked assets. Liquid staking on Ethereum has grown exponentially over the past year and is expected to further accelerate following the successful transition from the Proof of Work consensus mechanism to PoS in September 2022 (the Merge).
The first two sections of this article seek to clarify the terminology around staking, providing liquidity, and DeFi lending, before describing different PoS staking models. To determine whether there is a tax disposal in liquid staking transactions, section 3 considers the possible legal categorisation of such transactions under English law. We conclude that liquid staking transactions should not be treated as involving tax disposals, although they do not fit into any existing legal category, making this conclusion uncertain. Given this uncertainty, we support HMRC’s efforts to create clear tax rules for staking and other DeFi activities. We argue that a new regime based on the “no gain no loss” approach proposed by HMRC is a better way to achieve this aim than extending the current rules for securities lending to DeFi activities. This article is written from the perspective of English law and is based on Consensys’ response to the Call for Evidence.
1. Staking Must Be Distinguished From Lending and Providing Liquidity
As a starting point, it is useful to clarify the terminology used to describe different DeFi activities in the Call for Evidence. In our experience, DeFi “lending” refers to the process of sending tokens to a decentralised protocol that pools them with tokens of other users and automatically enables another party to borrow the tokens in exchange for collateral. The lender receives interest generated from loans and, if the borrower fails to repay the loan, the lender automatically receives the collateral.
The Call for Evidence uses the term “staking” to refer to providing liquidity. In our view, using the term “staking” in that context is overly broad and imprecise; instead, to avoid confusion, the term “providing liquidity” should be used. As with lending, a liquidity provider sends tokens to a decentralised protocol, usually in return for some incentive to do so in the form of payment of fees or additional tokens. In these instances, liquidity protocols serve as the basis for other users to borrow from, such as the popular Aave protocol. In general, these liquidity protocols add the depositor's tokens to a liquidity pool, which, depending on the protocol, may facilitate a broader range of activities beyond lending, such as trading, market making, or providing liquidity for decentralised exchanges. The tokens initially deposited to the protocol form the basis of the liquidity provider’s position in the given liquidity pool, where they expect the position to increase in value due to fees or rewards generated by the protocol from deploying the deposited tokens.
We use the term “staking” in this article to refer to the process of “locking up” tokens to participate in transaction validation on a PoS blockchain. Whereas lending and providing liquidity involves moving tokens between different parties, in staking the tokens are locked in a smart contract as collateral and risk being forfeited if the validator acts maliciously or is inactive. The locking up of collateral acts as an incentive for validators to act honestly when proposing new blocks or verifying and attesting to the validity of blocks proposed by others, thereby increasing the security of the network. Rewards generated by staking are therefore directly dependent on the quality of the validator’s work in light of the specifications of the PoS network. If a validator does not conform to the rules of the network, some or all of the staked tokens corresponding to that validator will be forfeited, or “slashed”. In contrast, revenue generated through lending or providing liquidity is (in broad terms) a function of supply/demand. It is not contingent on the work of the lender or liquidity provider and users do not risk losing their deposited tokens (except for a technical issue or an unexpected fall in the value of the borrower’s collateral).
2. Overview of Staking Models
While this paper focuses on liquid staking, this section describes other commonly used staking models and reproduces parts of an article titled "What is Staking" by Andrew Breslin from Consensys.
Solo staking means running an Ethereum node independently without the support of third parties. It requires a minimum of 32 ETH to activate a validator. Solo stakers are responsible for running their own software, and generating and securely storing validator keys (used to perform the validator’s on-chain duties) and withdrawal keys (used to withdraw one’s staked ETH once the Ethereum protocol is upgraded to enable withdrawals).
Staking as a Service (SaaS)
Staking through a SaaS provider also requires a minimum of 32 ETH. Stakers delegate the responsibility for technical and operational risks to the provider, who typically supports users through the process of depositing 32 ETH and generating their signing and withdrawal keys. Like in solo staking, the user custodies their withdrawal keys themselves, meaning they are the only ones who can withdraw their original 32 ETH deposit and the staking rewards earned by their validator. However, unlike solo stakers, the validator keys are held solely by the staking provider. Typically, users pay SaaS providers a portion of the rewards generated by the user’s validator in exchange for the technical services provided, obviating the need for the user themselves to perform the technical duties associated with solo staking.
Non-liquid Pooled Staking
Staking pools enable users who hold less than 32 ETH to combine their deposits with others and fund a new validator each time the pool reaches a cumulative value of 32 ETH or more. Pooled staking is similar to SaaS in that stakers outsource the technical complexity of operating validator hardware and software to someone else. In contrast to liquid staking pools discussed below, users do not receive a liquid token in exchange for depositing their assets. Currently more than 40% of all staked ETH is being staked via pools.
The most common type of non-liquid pooled staking is through a centralised exchange. Since centralised exchanges custody assets on behalf of their users, they can easily pool user assets and run large numbers of validators.
A notable differentiator between pooled staking and non-pooled staking models is that no single user deposit is tied directly to any single validator in the pool’s validator set. In solo staking and when staking via a SaaS provider, the relationship between the staker and validator can be thought of as a one-to-one relationship (a single user with 32 ETH funds a single validator). In pooled staking, the relationship between stakers and validators can be thought of as a many-to-many relationship (multiple users’ funds are distributed to the multiple validators that make up the validator set).
This many-to-many relationship means there is limited traceability between a user’s deposit and the validator(s) which their deposit is directed towards. Due to this lack of traceability, all of the staking rewards earned and penalties incurred by the pool’s validators must be mutualised across all pool depositors. The mutualisation of rewards and penalties means that no single user will benefit more than others if one of the pool’s validators earns outsized rewards, and similarly, no single user will suffer disproportionate losses if one of the pool’s validators is penalised.
Liquid Staking Pools
A popular subset of staking pools offers what is known as “liquid staking”, which provides depositors to the pool with an ERC-20 token representing their staked ETH. This token is implemented differently by different pools, but broadly, it enables users to track their initial deposit and their share of the staking rewards earned by the pool’s validators. Most of these pools operate using smart contracts that accept and manage user deposits, track each user’s stake, and allocate staking rewards in proportion to each user’s share of the pool.
Liquid staking pools have risen in popularity due to the nature of the Ethereum Beacon Chain, which required users who were participating in early staking to commit to leaving their staked ETH “locked” in the protocol for an extended period of time. The Beacon chain has now merged with the Ethereum Mainnet, and withdrawals will be enabled following the Shanghai upgrade planned for 2023. The time duration of the extended staking period creates a capital inefficiency, which liquid stakers seek to solve. The main liquid staking pool, Lido, has captured over 65% of the ETH liquid staking market.
The ERC-20 tokens that these liquid staking pools issue are commonly called “liquid” staking derivative or liquid staking token (LST) since they provide the staker with a limited form of liquidity on their staked position that they otherwise wouldn’t have while their assets are staked. Beyond exit liquidity, LSTs can be freely traded and used across DeFi as collateral assets, while the staked assets continue to generate staking rewards. LSTs usually trade at a small discount to the underlying locked collateral, representing the time value of liquidity and other considerations such as technical and smart contract risks. Contrary to centralised providers, it is possible to verify, thanks to on-chain data, that LSTs match the value of assets locked in a staking protocol.
3. Legal Categorisation of Liquid Staking Transactions
To determine whether there is a tax disposal, the key question is whether there is a transfer of beneficial ownership in the staked assets. Beneficial ownership is the right to the economic interest in an asset and is separate from legal ownership. A person may be both the beneficial and legal owner of an asset, or they may transfer the legal title to another person, such as a trustee or a custodian, who holds and possibly manages the asset for the beneficial owner’s benefit.
If there is no change in beneficial ownership, then there is no tax disposal and therefore no CGT arises. In a liquid staking transaction, we need to consider whether a CGT liability arises when (i) depositing the staked tokens in exchange for receiving LSTs and (ii) withdrawing the staked tokens in exchange for returning LSTs.
The question of beneficial ownership transfer is determined by considering the facts and circumstances, the terms of any agreement between the parties, and the legal substance of the transaction. HMRC notes that there will be no change in beneficial ownership when, for example, the platform1 holds the tokens on trust for the staker, as the trustee relationship implies some pre-defined constraints on the trustee’s ability to deal with the assets. In contrast, HMRC notes that there will typically be a transfer “when the platform is not constrained in what it does with the tokens, including being able to sell them outright to a third party”.
In the case of solo staking or SaaS, there is no change in beneficial ownership as stakers remain in control of their withdrawal keys at all times. There is likely no tax disposal in the case of pooled staking through centralised providers either. Their terms (e.g. Coinbase, Kraken) typically provide that assets are kept in custody by the exchange for the benefit of the account holder, and that title to the assets remains at all times with the staker and shall not transfer to the service provider.
There are no contractual terms governing users’ interactions with decentralised liquid staking protocols.2 We therefore need to consider what is the legal substance of liquid staking arrangements. Under English law, the relevant legal categories are (i) an outright title transfer, (ii) trust, or (iii) bailment. An outright title transfer results in a tax disposal. Under a trust or bailment, the beneficial ownership remains with the staker.
Apart from the substantive requirements of each category, the most important factor when determining which of these categories (if any) applies are the parties’ intentions, which would normally be ascertained from contractual terms (if there were any).3
We argue in the following sections that, although liquid staking arrangements do not neatly fit within any existing legal category, the correct conclusion is that there is no tax disposal. The analysis is nuanced and complex, and taxpayers cannot be expected to carry out this exercise. We need clear guidance from HMRC on this position.
(I) Outright Title Transfer
An outright title transfer enables the recipient to freely dispose of the assets and use them for its own benefit (as opposed to the user’s benefit), restricted only by any contractual obligations.4 For example, in a banker-customer relationship, depositing money to the bank results in a transfer of legal and beneficial title to the deposit, which allows the bank to use customers’ deposits to finance its activities such as lending. Customers only have a contractual claim to receive a sum equivalent to their deposit.5
It is clearly not the parties’ intention for the protocol to deal with the staked assets freely. In fact, it is not even technically possible as most pools are implemented on-chain via smart contracts that accept users’ tokens, pool them with others, and use 32 ETH portions to fund new validators. The smart contracts operate autonomously (i.e. without reliance on humans).
This is subject to two caveats. First, the protocol’s governance community6 may be able to upgrade the smart contracts to enable withdrawals of staked assets following the Shanghai upgrade, or to fix bugs. Any malicious alterations to the code would clearly be contrary to the parties’ original intentions. Second, there is an argument that by pooling/co-mingling assets with other users’ assets, the user relinquishes control over and ownership of the staked assets, resulting in an outright title transfer to the protocol.7 However, this would not be in line with the parties’ intentions and the fact that the protocol’s actions are constrained by code. A liquid staking arrangement therefore should not be treated as involving an outright title transfer.
For a trust to arise, there must be, among other things, a certainty of intention.8 It must be clear from the words used, either in the parties’ contract or otherwise, that the parties intended for the trustee to hold the assets on trust for the other party. As mentioned, there are no (natural language) terms and conditions applicable to decentralised liquid staking protocols. Unless there are any direct communications between the parties, there is insufficient information to find an intention to create a trust.9
The Law Commission’s recent paper describes bailment as follows: “Under the current law, a bailment occurs where one person (the bailee) takes voluntary possession of a possessable (that is, tangible) object of property belonging to another (the bailor), usually for a specific purpose. The bailor retains the superior legal title to the object. At the end of the bailment, the bailee must either return the goods to the bailor or deal with them as the bailor directs. Examples of bailments include storage of the transferor’s goods in the transferee’s warehouse and a waiter taking a diner’s coat for safekeeping at a restaurant… Where a bailment relationship arises, the bailor is under a duty to take such care of the goods as is reasonable in the circumstances.” 10
Liquid staking fits this description: the protocol (bailee) takes voluntary “possession” of the staker’s (bailor’s) assets for the specific purpose of locking them in a smart contract as collateral. The staker receives LSTs evidencing their legal and beneficial ownership of the staked assets, similarly as a diner might receive a receipt for leaving their coat at the cloakroom.
However, under the current law, intangible assets, including crypto assets, cannot be subject to bailment because they cannot be “possessed”. Academics have proposed the concept of “quasi-bailment”, which argues that bailment is fundamentally about the imposition of a set of duties on the bailee rather than about holding an asset identical to the asset initially given to the bailee.11 This removes the need for “possession” and rationalises “bailment” of intangible assets, including those that are fungible. When exchanging LSTs back to ETH, a staker will not receive the identical tokens they initially deposited; they would just receive an equivalent amount of ETH.
The Law Commission is currently evaluating responses to a recent consultation on this topic. Its provisional conclusion is that extending bailment to crypto assets, or the creation of an analogous concept, is currently not necessary given the availability of other legal categories. Nevertheless, the Law Commission acknowledged that delegating validator duties to a staking pool “is typically set up and administered by smart contracts automatically but the underlying legal relationship might be best characterised as one of bailment”, and that “as the crypto-token and cryptoasset markets evolve, there might be good reasons for developing a legal mechanism that allows for the imposition of legal duties on a party without the need for a trust relationship to arise and in the absence of a contract. Examples of where we consider that such a legal mechanism might prove helpful include in the context of certain staking arrangements.”
Looking Beyond Existing Categories
Unless a new “quasi-bailment” concept along the lines above is recognised, liquid staking arrangements do not easily fall within any existing category. However, it is important to realise that these categories presume the existence of a relationship between a client and a trusted party (as in the case of a trust or bailment) or between counterparties (as in the case of an outright transfer). The categories have been created to define and impose certain duties on these parties, either by statute or common law. Common examples are fiduciary duties imposed on trustees or the duty of care imposed on bailees. The duties aim to limit the risk of the trusted party misusing or misappropriating the other party’s assets.
Users engaging in liquid staking through decentralised protocols interact with a set of open-source smart contracts, as opposed to a human or corporate counterparty. This eliminates the risks that the imposition of intermediary-style duties under existing law seeks to address. When trying to rationalise new forms of relationships that do not involve a trusted party, we should not limit ourselves to existing legal categories. Perhaps it is most appropriate, at least for the purposes of the tax analysis, to conclude that a staker simply interacts with a smart contract while retaining the beneficial ownership throughout the process.
This raises the question as to whether the interaction creates a contractual relationship. English law can in principle treat smart contracts as legally enforceable contracts if the standard contract formation requirements are satisfied.12 This will depend on the specific circumstances and nature of the arrangement. We expect that the main difficulty will be with satisfying the requirement of intention to create legal relations. In the absence of any natural language communications, the mere existence of code allowing others to interact with it does not necessarily imply an intention to be legally bound on the basis of such code. Instead, the party interacting with the code might merely intend to use the code’s functionality.13 The absence of natural language terms and conditions for decentralised liquid staking protocols implies the latter. There is additional uncertainty as to who would the staker contract with in the case of protocols run by a DAO.
The above analysis shows that there are too many open questions to provide sufficient clarity to taxpayers. This is why we applaud HMRC’s efforts to create bespoke tax rules applicable to DeFi activities. We consider HMRC’s proposals for reform in the following section. We conclude that the ‘no gain no loss’ treatment of liquid staking is the most appropriate option.
4. HMRC’s Tax Reform Proposals
The Call for Evidence sought stakeholders’ views on (i) whether it would be suitable to extend the existing tax regime for securities lending to DeFi staking and lending, and (ii) if not, whether it would be more suitable to treat the transfer of cryptoassets for staking or lending as a ‘no gain no loss’ transaction, by treating the disposal value as matching the acquisition cost. We consider each option in turn.
Extending the Existing Securities Lending Rules to Liquid Staking Transactions
Securities lending (also called stock lending) is commonly understood in the UK as “the outright transfer of securities from a lender to a borrower, against a promise by the borrower to re-transfer equivalent securities or the market value of the securities in cash on demand or at the end of a pre-agreed term. The borrower transfers to the lender other securities or cash as collateral and also pays a fee to the lender.”14
The legal and beneficial ownership of the lent assets is transferred to the borrower, which allows the borrower to deal with the assets as it sees fit, including selling or lending them to a third party. Despite their name, the transfers are therefore not a loan, but an outright title transfer. But for the special tax rules discussed below, there would be two disposals for CGT purposes - upon the outright transfer of lent assets from the lender to the borrower, and upon the transfer of those securities from the borrower back to the lender at the end of transaction.
However, from an economic and tax perspective, it is intended for the lender's financial and legal position to remain, as far as possible, as it would have been had the securities not been lent – including its continued exposure to risks and benefits arising from those securities.15 It is this difference between the legal form and the practical economic effect of securities lending that has prompted HMRC to create special tax rules which disregard the two disposals for CGT purposes.16
The transfer of assets from the staker to the protocol and back, including the staker’s continued exposure to risks and benefits arising from the staked assets, makes the economic realities of a liquid staking transaction broadly comparable to that of a securities loan.17 The tax implications of each of these arrangements should therefore be comparable as well.
However, merely extending the existing securities lending tax rules (or an amended version thereof) to liquid staking transactions is inappropriate and should be avoided for two reasons. First, it risks creating confusion as to the potential knock-on effects. In particular, the extension may be misconstrued as treating crypto assets such as ETH as securities for financial regulatory purposes without analysing their characteristics. Such misconstruction is substantively wrong and would muddy the otherwise clear rules as to which types of crypto assets qualify as securities in the UK. Another potential knock-on effect is to imply that liquid staking transactions involve a transfer of beneficial ownership in the same way as securities loans. As discussed in section 4, this is incorrect or at best unclear.
Second, there are conceptual differences between the two arrangements. The function of collateral in a securities loan is to provide assets that the lender can seize if the borrower defaults. The core function and utility of an LST is to act as a receipt for/evidence of legal and beneficial ownership of a corresponding portion of the underlying staked assets, to enable transferability. The liquid staking protocol makes a “promise” to re-transfer the staked assets18 to the staker on demand. This re-transfer functionality is (or will be following the Shanghai upgrade) programmed into smart contracts, and therefore will be available automatically. For this reason, LSTs usually trade at a value broadly equivalent to the underlying locked asset. In contrast, the collateral in a securities lending transaction may fluctuate in value unrelated to the value of the lent assets (subject to any margin calls).
Given the risk of confusion and the different function of LSTs and collateral, extending the securities lending tax rules to liquid staking transactions would be inappropriate. The “no gain no loss” approach proposed by HMRC in its Call for Evidence is more suitable, as discussed below.
“No gain no loss” Treatment
This proposal treats the transfer of cryptoassets for staking as a ‘no gain no loss’ transaction, by treating the disposal value as matching the acquisition cost. According to the proposal, a CGT gain or loss would arise when cryptoassets are disposed of in a non-staking transaction, thereby deferring the tax liability until the cryptoassets are economically disposed of.
This approach is more appropriate as it reflects the fact that, due to their function as a receipt, the LSTs are materially the same in value and in kind as the corresponding staked assets. As the value of LSTs broadly reflects the value of the staked assets, stakers remain exposed to the fluctuations in value of the staked asset throughout the arrangement. The exchange therefore should not result in a realisation of gain or loss.
For example, Lido uses a single “rebasing” token called stETH to track users’ stake and rewards. Users earn interest daily, increasing the amount of stETH they own. Lido maintains a 1:1 ratio of stETH to ETH staked in their validators. StETH is generally trading at around 0.99 - 0.97 of an ETH. The “no loss no gain” approach would simplify users’ tax calculations as they would not need to declare a notional gain of 0.01 per ETH when depositing ETH and receiving stETH in exchange, and the notional loss of 0.01 per when exchanging stETH back to ETH.
Rocketpool uses a single “rewards-bearing” token called rETH to track users’ stake and rewards. The amount of rETH users receive in exchange for depositing ETH never changes. As rewards accumulate within the Rocketpool contract, the value of rETH increases. When the user exchanges rETH to ETH at a future time, they will receive more ETH than the user initially deposited. When depositing ETH to Rocketpool, there is no gain as rETH value is equal to the value of the deposited ETH. When exchanging rETH back to ETH, the user would need to separate the value of rETH that reflects the initially staked ETH from the increase in value of rETH due to the accumulation of ETH rewards, with the latter being taxed separately. While the “no gain no loss” approach creates some challenges for this model, it is still the most suitable approach proposed by HMRC as it reflects the economic structure of a liquid staking transaction and promotes conceptual clarity and consistency.
This article is for discussion purposes only and does not constitute tax or legal advice.
Consensys Software Inc.
The Call for Evidence uses the term “platform”, defined as entities (though not necessarily legal persons) that provide DeFi services. In the rest of this article we refer to liquid staking “protocols”, which should be distinguished from platforms providing access to those protocols.
Platforms providing access to staking protocols, such as https://lido.fi or https://rocketpool.net/#header, have terms and conditions governing the use of the platform, but such terms do not govern the use of the protocol itself.
The categories, including the approach to categorising different arrangements, are discussed in greater detail in H Liu, L Gullifer and H Chong, “Client-intermediary relations in the crypto-asset world” (2020) University of Cambridge Faculty of Law Research Paper No 18/2021.
Liu, Gullifer and Chong (2020).
Foley v Hill (1847) 2 HLC 28.
This is typically formed of holders of the governance token issued by the protocol, formed into various groups or committees. A detailed discussion of governance structures is out of scope.
Andrew Burrows (ed), English Private Law (3rd ed, OUP 2013), 16.18.
Practical Law practice note: Creating a trust: bringing the trust into existence.
A trust may also arise by operation of law, either to remedy an otherwise unconscionable outcome (for example, where a person has obtained an asset by means of fraud or by mistake and refuses to return the asset to the rightful beneficial owner) (“constructive trust”), or to return the beneficial ownership back to the person who initially set up the trust in certain limited circumstances (“resulting trust”). None of these situations are relevant for our purposes.
Law Commission, Digital Assets: Consultation paper 256 (2022), accessed 18 November 2021, https://s3-eu-west-2.amazonaws.com/lawcom-prod-storage-11jsxou24uy7q/uploads/2022/07/Digital-Assets-Consultation-Paper-Law-Commission-1.pdf.
Liu, Gullifer and Chong (2020).
As confirmed by the Law Commission in Smart Legal Contracts – Advice to Government (2021), accessed on 14 November 2022, https://s3-eu-west-2.amazonaws.com/lawcom-prod-storage-11jsxou24uy7q/uploads/2021/11/Smart-legal-contracts-accessible.pdf, and the UK Jurisdictional Taskforce – Legal statement on cryptoassets and smart contracts (2019), accessed on 14 November 2022, https://35z8e83m1ih83drye280o9d1-wpengine.netdna-ssl.com/wp-content/uploads/2019/11/6.6056_JO_Cryptocurrencies_Statement_FINAL_WEB_111119-1.pdf.
This is further discussed by the Law Commission in Smart Legal Contracts – Advice to Government (paragraphs 3.71-3.75).
Practical Law – Securities Lending Practice Note.
Under s263B(2) of the Taxation of Capital Gains Tax Act 1992.
Any comparison of liquid staking transactions with securities lending transactions in this article does not imply, and should not be construed as implying, that we believe the proper regulatory framework for ETH or other cryptocurrencies is The Financial Services and Markets Act 2000 (Regulated Activities) Order 2001, the US Securities Act of 1933, or any other similar financial regulatory regime. Any comparison between staking transactions and securities lending transactions is made solely for the purpose of analysing the economic realities of these transactions and their tax implications.
For clarity, the staker will not receive the identical assets that they had deposited, as the assets are fungible.