Cryptocurrency Economics and the Taxation of Block Rewards

I. Introduction

Cryptocurrency networks have no dollars or other real currency to be spent on their own maintenance. Instead, newly minted cryptocurrency tokens are used to encourage people to help maintain the network. This report argues that these reward tokens should not be included in gross income when they are first created and that the better approach is to tax them when they are sold or exchanged.

A crude metaphor will help in understanding the way cryptocurrencies actually work and why it matters for taxation. Those who hold a cryptocurrency’s tokens pay for the maintenance of that cryptocurrency network. Token holders do this by (figuratively!) taking a handful of their tokens and putting them in a communal fund — let’s call it a cryptocurrency kitty. A sign on the kitty reads, “These tokens are for those who maintain the network.” Those who maintain the network are allowed to take tokens from the kitty and keep them.

Treasury wants to tax those who take tokens from the kitty. The taxation problem arises most clearly when the token holders who put tokens into the communal fund are the same people who take tokens out. When a token holder both pays into the kitty and is paid out of it, it’s not fair to ignore the extent to which that token holder is effectively paying himself. Existing tax guidance asks this token holder to pay tax on the tokens taken out of the kitty without considering the tokens he put in.

As a matter of fact, a cryptocurrency could be designed to use a kitty as just described. But this is not the way cryptocurrencies are designed. This report explains how they are actually designed, which involves the creation of new tokens. But as the report walks through the mechanics of how cryptocurrencies work, it’s helpful to keep the kitty metaphor in mind.

Proper tax treatment is especially important for the newer proof-of-stake cryptocurrencies, which differ from Bitcoin and proof-of-work cryptocurrencies in how network maintenance responsibilities are distributed. Proof-of-stake technology, as the term is used in this report, allocates the opportunity to maintain the cryptocurrency network according to participants’ ongoing stake, or proportion of ownership, in that network. This means that in proof of stake, those who put tokens into the kitty are the same people who take tokens out.

Cryptocurrency research and development is increasingly focused on proof-of-stake technology. Among the nine new, public cryptocurrencies launched since January 2016 that have a current or recent network value of at least $500 million, seven use proof of stake to allocate network maintenance responsibilities, and the two that don’t incorporate some proof-of-stake principles.¹ These seven proof-of-stake cryptocurrencies have a combined network value of about $6.8 billion, 3 percent of total cryptocurrency market valuation. Ethereum, the cryptocurrency with the second largest network value after Bitcoin, is expected to switch to proof of stake in 2020. This would bring the combined value of these eight proof-of-stake cryptocurrency networks to $26.6 billion, 12 percent of total cryptocurrency market valuation or 35 percent of total valuation if Bitcoin is excluded from the calculation.

Proof of stake is illustrated in this report with the Tezos cryptocurrency. Returning to the metaphor, Tezos token holders pay for the maintenance of the Tezos network by putting about 5 percent of their tokens in the kitty each year. For those token holders who help maintain the network, the amount they get to take back out depends on how many others are also helping with network maintenance. As it stands right now, a little more than two-thirds of Tezos tokens are held by those who are helping with this maintenance, so the way the math works out, these token holders get to take back their 5 percent, plus about 2 percent on top of that. This extra 2 percent comes from those who pay for, but don’t help with, the network maintenance.

Again, this is just a metaphor. But as we'll see, what literally happens is just an accounting trick to achieve what really happens, which is that the helpers gain 2 percent and the non-helpers lose 5 percent. Because the helpers do gain, they will be taxed. But they should be taxed at the appropriate time and on the appropriate amount.

The accounting trick that leads Tezos token holders to either gain 2 percent or lose 5 percent is the same trick used by most every public cryptocurrency, including Bitcoin. To provide an incentive for people to maintain the network, instead of having token holders put some of their own tokens into a kitty, new tokens get created out of thin air. These reward tokens are valuable, but even though we can assume they have an exact dollar value when they are created, their value is not as straightforward as it may appear. This is because the function of reward tokens is to redistribute the share of ownership (or stake) in a cryptocurrency network away from those who don’t participate in the maintenance of the network to those who do. The new tokens dilute the stake of all token holders while on net increasing the stake only of those who participate. By necessity, this dilution effect partially offsets any gains from participating in network maintenance. Including these rewards in gross income when they are received fails to account for this dilution effect and complicates the proper taxation of these tokens.

The argument for how block rewards should be taxed requires a careful explanation of the mechanics and economics of public cryptocurrency networks. This report presents that explanation as clearly and literally as possible. But analogies and metaphors remain an important part of the argument for two reasons. First, cryptocurrency is unique in terms of its tax implications, and neither Congress nor Treasury has addressed these issues. We are left with a variety of potentially relevant policies and principles, so analogy is necessarily a primary means of engaging with existing law.

Second, cryptocurrency itself makes sense only through analogy and metaphor. Cryptocurrency tokens exist, but in calling them tokens, we come to understand them by reference to another, older idea of a token.² Some of these analogies and metaphors are far from perfect, which, as we will see, has led to some pervasive confusion about cryptocurrency. To the extent there are literal truths about how cryptocurrencies work, sometimes those truths are important, but sometimes they’re not. For example, a detail in a cryptocurrency’s code could lead to a particular tax consequence, while a different method of solving the same design problem could lead to a different tax consequence.

This is worrisome. Today there are about 300 cryptocurrencies (or other digital assets of a type that didn’t exist a decade ago) that by at least one measure have a dollar value of $10 million or more.³ New variations are emerging, and even proof-of-stake cryptocurrencies — the focus of this report — make up an evolving and sometimes amorphous category. Unless we’re careful, cryptocurrency will be taxed on the basis of bad metaphors and flawed analogies.

This risk is illustrated by Notice 2014-21, 2014-16 IRB 938, released in March 2014, which states that Bitcoin mining rewards are gross income when received.⁴ This informal IRS guidance is not law,⁵ but as the government’s only pronouncement on this issue to date, and given the legal uncertainty surrounding many cryptocurrency issues, the advice is influential and risks being applied to technology that might not have been understood when it was drafted. On October 9 the IRS expanded on Notice 2014-21 with additional guidance, but the new guidance does not revisit the issue of block reward taxation.⁶

As we’ll see, the notice doesn’t present the reasoning behind its conclusion. But given the state of the technology in 2014, embodied primarily by Bitcoin, the analogies on which the guidance appears to be based were not then unreasonable. Applied to Bitcoin and similar cryptocurrencies, the advice is defensible. But it doesn’t make good sense in light of the newer proof-of-stake technology, and a clear understanding of cryptocurrency economics shows why a single taxation policy can and should apply to both proof-of-stake and proof-of-work cryptocurrencies.

Although cryptocurrency is an evolving technology, the fundamental economics of network maintenance are clear. An appropriately cautious approach to taxation should therefore be based on these fundamental economics. Taxation should not be based on features of particular cryptocurrencies that, on examination, are idiosyncrasies of how those particular cryptocurrencies were designed. Accordingly, this report argues for a single taxation policy with broad application.

The creation of block rewards should not be a taxable event. Instead, reward tokens should be taxed only when they are sold or exchanged. This ensures the fair market valuation and equitable taxation of all tokens and significantly reduces administrative burdens. Of course, how block rewards should be taxed could differ from how they must be taxed under current law. Although cryptocurrency would benefit from legislative and regulatory clarity and certainty, this report also argues that in the meantime no act of Congress or new Treasury regulation is required to ensure the proper taxation of block rewards.

The kitty metaphor helps explain how cryptocurrencies work, but a different metaphor explains how reward tokens should be taxed. This second metaphor is also closer to the literal truth for cryptocurrencies designed thus far, given that we are discussing intangible strings of ones and zeroes: Those who maintain a cryptocurrency network create reward tokens.

There is a meaningful if rarely invoked distinction between property that is received as compensation and property that is created. Received property typically is income when it is received. Created property, on the other hand, typically is not income when it is created. It results in income or a taxable gain only when it is first sold or exchanged. Reward tokens are best understood as property created by those who maintain a cryptocurrency network. In this sense, the tokens are similar to common goods such as crops, minerals, livestock, art, and even manufactured goods.

Each of these things can be obtained as compensation, which will typically render it gross income under the code and regulations. But each of these forms of property can also come into one’s ownership without triggering a taxable event — namely, when it is grown, mined, raised, painted, or otherwise created in the course of business. The same logic applies to block rewards. New cryptocurrency tokens are indeed created in the course of maintaining the cryptocurrency network. These tokens often do have value at the time they are created, just as wheat has value both before and after its harvest, and just as widgets have value as they come off the assembly line. But like wheat in the field, block rewards should not — and need not — create taxable income until they are sold or exchanged.

As a matter of legal doctrine, the accession to wealth represented by reward tokens must be realized to be included in gross income.⁷ Given the economics of cryptocurrency incentives and in particular the effect of dilution, as well as reward tokens’ nature as created property, as a matter of doctrine, the taxpayer’s new wealth should not be deemed realized until the tokens are sold or exchanged.

* * *

This report does not provide a comprehensive overview of cryptocurrency mechanics or economics. Nor do the arguments apply to all “digital assets” or “virtual currencies.”⁸ The report concerns the economic incentives underlying the maintenance of public cryptocurrencies⁹ and how those incentives should be taxed.¹⁰ The focus is on proof-of-stake cryptocurrency and the example of Tezos, although we return to Bitcoin and proof of work at the end of the report.

As explored in depth later, a proof-of-stake system allocates the opportunity to maintain the network according to the stake, or share, of the cryptocurrency network owned by those who participate in this process. As used in this report, “stake” means a share in a cryptocurrency network, with the emphasis on ownership. The report restricts the term to this meaning for clarity and consistency. There is another common usage, however. When used as a verb or gerund, the word has a slightly different meaning, with the emphasis on commitment and risk: “Staking tokens” refers to a commitment of one’s tokens (such as to secure validation rights) that prevents their use for other purposes, often for an extended period and often with a risk of the tokens’ forfeiture.

This report provides an overview of cryptocurrency technology and explains the consequences of treating proof-of-stake block rewards as gross income as suggested by Notice 2014-21. Using the kitty metaphor as applied to Tezos, this results in taxing seven tokens as income when the validator’s true gain is only two tokens. This is not equitable and could invite creative but economically pointless redesign of cryptocurrencies to better align how they literally function with how they really function in order to prevent improper taxation. The report then explores two possible methods for taxing block rewards fairly. Using the kitty metaphor, the question is how to tax the two tokens of actual gain rather than the seven tokens that would be included in gross income under the notice. The report argues that the fairest method is to tax reward tokens only when they are sold or exchanged, and that this result follows naturally when reward tokens are properly understood as property created by those who maintain a cryptocurrency network. Before concluding, this report returns to Bitcoin to illuminate the key differences between proof of work and proof of stake as they relate to network maintenance and taxation.

Cryptocurrency is a new invention. Its various features, especially when analyzed in isolation, invite analogies to a wide range of tax provisions and policies. Tokens can look a lot like shares in a public corporation, with reward tokens analogous to stock distributions or, for proof-of-stake cryptocurrencies, stock dividends or other distributions paid to shareholders. From a different angle block rewards look a lot like compensation paid to those who help maintain a cryptocurrency network, and this may be the basis for Notice 2014-21, which would deem Bitcoin block rewards gross income at their fair market value when received. But neither characterization quite fits.

Cryptocurrency is unique, and there is no perfect analogy in existing law or policy to determine its tax treatment. That said, the report argues that current doctrine — some of it still implicit over 100 years after the 16th Amendment legalized the federal income tax — provides ample precedent for the proper income tax treatment of cryptocurrency block rewards. As briefly sketched out above, like many other forms of property, reward tokens are best understood as property that is created by those who are the first to own them. For reasons that rarely require articulation, property that is created — or grown, mined, raised, cut, fished, hunted, trapped, painted, coded, assembled, or manufactured — is generally not taxed until it is first sold or exchanged. So too should it be with newly created cryptocurrency tokens.

II. Cryptocurrency Block Rewards

A. Introduction to Block Rewards

A cryptocurrency is a form of scarce, valuable digital property that is maintained on a network of connected computers. One critical feature of such a network is that it is operated not by a single authority such as a central bank but by some number of entities operating independently of each other, all governed by the rules of the network. This maintenance of the network by numerous participants is important for various reasons, in particular to ensure the integrity of the network.

Those who own some of the scarce coins or tokens that exist on a cryptocurrency network have an interest in ensuring the network is properly maintained. If all the computers are turned off, the cryptocurrency can’t be used, and neither can the network be used for other purposes. But most cryptocurrencies are designed to further encourage participation in the maintenance of the network. A common approach to the problem of network maintenance involves the creation of new tokens.

Maintenance of a cryptocurrency network involves registering and publishing transactions made on the network so that there is a trusted record of, for example, how many tokens are held in each account. The typical approach is to publish this record in a ledger that is periodically updated to include recent transactions. Copies of this ledger are stored on many different computers, giving rise to the term “distributed ledger,” and computers that maintain an up-to-date copy of the ledger are often referred to as nodes. Recent transactions are batched together and appended to the ledger in a block, and this chain of appended blocks of transactions provides the common term for this kind of ledger, the “blockchain.” Participation in network maintenance is often called validation because a core task is to confirm the legitimacy of transactions before they become a permanent addition to the blockchain. Those who perform this network maintenance are called validators.

The new tokens that provide an incentive for the network’s maintenance are commonly included in one of the transactions in a new block and thus are often described as block rewards. In this report, block rewards are synonymous with reward tokens; they are the tokens created ex nihilo according to the rules of the cryptocurrency software to encourage network maintenance. These new reward tokens increase the total number of tokens in existence. For that reason, these tokens are sometimes described as inflationary block rewards. Because the term “inflation” has various connotations, not all of which fit well with the technical sense of token inflation, this report primarily uses the terms “token creation” and “token creation rate” in connection with block rewards. This report argues that those tokens are created by people who maintain the network, but following common industry usage, we can also say those tokens are “issued” or “created” by the network’s computer code and are “received” by those who participate in validation. “Token issuance rate” therefore has the same meaning as this report’s preferred locution of a “token creation rate.”

B. Rationing Validation Rights

Block rewards provide an incentive to help maintain a network, but an equally important consideration is how to allocate or ration network maintenance rights among those who have an incentive to help with that maintenance. This is where we encounter the concepts proof of stake and proof of work, which describe different approaches to rationing maintenance responsibilities in a way that preserves the integrity of the network.

The art and science of allocating network maintenance responsibilities is largely beyond the scope of this report, but it is important to understand the basic principle. For the moment, it is enough to note that some method is required to choose who gets to maintain the network, and that method must prevent any single entity, or small number of entities, from taking over the process. If too much responsibility is concentrated in too few validators, the cryptocurrency can be compromised.

A related challenge is ensuring that validators, however they are selected to help maintain the network, do their job correctly. Again, the art and science of this requirement is largely outside the scope of this report. For now, it is enough to note that the creation of block rewards hinges on doing validation correctly.

The complexity and significance of these twin considerations — rationing validation rights and keeping validators honest — is important to mention because, as it turns out, the effort involved in actually validating transactions is rather minimal. It’s done in the blink of an eye, by software that typically can run on normal computers. Indeed, it could be done much more efficiently by a single entity.

C. Bitcoin and Proof of Work

Bitcoin, the original cryptocurrency, is an example of what is now called a proof-of-work protocol. In Bitcoin, the rights to publish new blocks of transactions to the Bitcoin blockchain and to create the associated rewards are allocated according to participants’ “work” on the network. This work is typically measured by the computing power that a Bitcoin validator — known as a miner — has dedicated to the network. For each particular block, the right to publish it to the Bitcoin blockchain is “proved” by publicly sharing the computationally expensive solution to a cryptographic puzzle. It is important to understand that the effort involved in collecting valid transactions and preparing them to be published is trivial compared with the work required to entitle the miner to publish that block in the first place. Proof of work refers to this method of allocating and verifying block publication rights, not to the act of publishing valid transactions to the Bitcoin blockchain.

Bitcoin mining has become a specialized and capital-intensive business, so today most who own bitcoin do not mine it. Mining requires specialized computers fed by electricity. Miners compete with each other for the bitcoins that one miner or another brings into existence about every 10 minutes. At recent bitcoin prices of around $8,000 each, about $14 million worth of bitcoins are created each day. Although reliable statistics are scarce, this effort probably accounts for about 0.2 percent of the world’s electricity consumption.¹¹

Bitcoin miners need not own bitcoin. They can immediately sell any block rewards they receive without affecting their ability to mine more blocks in the future. And because bitcoin mining was specifically designed to require costly computing power, miners are also likely to liquidate their block rewards to finance their mining operations. If a cryptocurrency validator immediately liquidates all tokens, the puzzles concerning the tokens’ taxation are greatly diminished. We will return to Bitcoin and proof of work generally in Part 2, Section VI, once the argument regarding taxation is fully developed.

D. Tezos and Proof of Stake

Proof-of-stake validation is different from proof-of-work validation, although within the genre, there are innumerable methods for allocating validation opportunities according to validators’ stake in the network.¹² The Tezos cryptocurrency is an illuminating example, in part because Tezos is designed to encourage near-universal participation in the maintenance of the Tezos network. Tokens on the Tezos network are called tez, and virtually every owner of tez can participate in the process, and can do so easily. Most Tezos tokens do indeed participate in validation; therefore, the question of block reward taxation affects the owners of most tez in existence.¹³ Tezos also has a simple token creation policy — new tokens are created at a more or less consistent rate — so it’s easy to create simple models that illustrate the economics and taxation of Tezos block rewards.¹⁴

Unlike Bitcoin and other proof-of-work cryptocurrencies, validation in Tezos requires ongoing ownership of a stake in the network. If you liquidate your tez — to pay taxes, or for any other reason — you can’t participate. Also, taxation of proof-of-stake block rewards can introduce significant administrative and compliance burdens. Leaving aside whether the taxation of block rewards is equitable, with Tezos, tax compliance obligations could be conspicuously high compared with taxpayers’ level of involvement with the cryptocurrency. For example, a casual Bitcoin user can hold his bitcoin indefinitely without triggering a taxable event. In contrast, as explained later, a casual Tezos user holding even a modest amount of tez would have to account for about 125 taxable events each year if block rewards are included in gross income when they are received.¹⁵

1. Tezos basics.

At the time of this writing, there are about 806 million tez in existence. Approximately 115,000 new tez are created each day.¹⁶ Each minute or so a validator, also known in Tezos as a baker, adds a new block of transactions to the Tezos blockchain.¹⁷ Each new block also introduces a maximum of 80 new tokens to the network. Anyone with an internet connection, some programming savvy, and the required minimum of 8,000 tez can become a validator and compete for a proportionate share of these reward tokens. Validation opportunities are allocated at random according to the validator’s stake, or share, in the Tezos network. Over time, a validator with 16,000 tez will receive twice the opportunities that a validator with the minimum 8,000 tez will receive.

All other tez holders may delegate their validation rights to validators, thus adding to the stake that determines each validator’s opportunities to create blocks and the associated reward tokens. Validators who seek delegations set their own terms and might, for example, retain 10 percent of a delegator’s rewards as their fee. Block rewards accrue with the creation of each block. Deposits from the baker’s own balance of tez are required for each act of validation for security reasons and to help limit the size and market dominance of individual bakers. Deposits and rewards are frozen for between 15 and 17 days; block rewards are released into each validator’s full control (and can then be relayed to delegators) in batches every three days.

Some readers may find the following in-depth explanation of Tezos block rewards helpful. But note that this report also argues that the precise mechanics of network maintenance and the creation of reward tokens are not critical to the question of taxation. Different cryptocurrencies operate differently, and taxation policy ultimately shouldn’t turn on the details of particular protocols. Moreover, cryptocurrency protocols can change, and Tezos in particular is designed to evolve according to a protocol amendment process that is closely tied to its method of delegation and validation.

2. Validation in Tezos.

In contrast to Bitcoin’s mining, in Tezos the validation process is known as baking, and Tezos validators are sometimes referred to as bakers. A new block of transactions can be published to the Tezos blockchain (or baked) once 60 seconds have elapsed since the publication of the previous block. Essentially, Tezos distributes opportunities to perform validation by lottery, with a validator’s stake in the network determining the number of lottery tickets and therefore the probabilistic frequency of opportunities to perform the tasks that yield block rewards.

Baking is done by maintaining a public copy of the Tezos blockchain and running the Tezos software on an internet-connected computer, so that the software can communicate with other bakers’ software. A baker’s stake is measured in rolls, with each roll consisting of 8,000 tez. A minimum of one roll is required to be a baker. Each block in the Tezos blockchain is numbered sequentially, and its number is also known as the block’s height. Periodically, the network takes a snapshot of all bakers’ rolls, and then uses a random number seed drawn from recent blockchain data to assign future baking rights to specific rolls. A baker with 100 rolls is 100 times more likely to receive a particular validation right, or slot, than is a baker with one roll.

Tezos validation consists of two principal operations: baking (or creating) blocks, and endorsing blocks. The baker assigned to create a particular block does so by signing a block of new transactions, having 512 tez frozen as a required deposit, and sharing the signed block with others on the network. The signature requires the baker’s private cryptographic key as an input and serves as proof of his stake in the network entitling him to create the block.

When the 512 tez are frozen as a deposit, the baker is also credited with the block reward: 16 newly minted tez, plus any transaction fees collected from transactions recorded in the new block.¹⁸ The block reward, along with the deposit, remains frozen for a set period. While frozen, deposits contribute to the baker’s stake but cannot be spent or transferred.

If the baker with the right to create a block fails to do so within the allotted one-minute window — say, if the baker isn’t connected to the internet, or is running an incompatible version of the Tezos software — the baker next in line, as previously determined by the lottery, then has a new 60-second window to sign a block and create the reward. If baker No. 2 also fails, baking rights again shift to the next in line, and so on. The network remains operational as long as new blocks are created and successfully added to the blockchain. If all bakers turn off their computers, the blockchain will halt.

The second operation, endorsing, serves to verify that a just-signed block is a valid addition to the blockchain. Endorsement rights are distributed by the same type of lottery. Up to 32 endorsements can be made per block, each requiring a deposit of 64 tez and each resulting in a block reward of at most two tez.

Missing an opportunity to bake or endorse a block carries its own penalty: no block reward. Other failures are more grave. The blockchain can be compromised if validators use their baking rights to sign or endorse more than one block at the same height.¹⁹ The forfeitable deposits are required to ensure that block creators and endorsers do their jobs correctly. A baker who catches another baker’s attempt to sign more than one block of the same height and submits proof to the blockchain is rewarded with a portion of the forfeited rewards.

Block rewards and deposits are unfrozen on a defined schedule, about once every three days. More precisely, Tezos operates in cycles of 4,096 blocks, with each cycle lasting a minimum of 4,096 minutes, or about two days and 20 hours. Block rewards and deposits remain frozen for the duration of the cycle during which they were frozen, plus an additional five cycles. During this time, they remain subject to forfeiture. Tokens frozen six cycles earlier are unfrozen into the full custody of the baker at the conclusion of the cycle.

Each new block contains up to 80 new tez: 16 for the block’s baker, and up to two tez each for up to 32 endorsers. The minimum block time is 60 seconds, so up to 4,800 new tez are created each hour, and up to 115,200 each day.

3. Delegation in Tezos.

The second way for tez holders to participate in validation is known as delegation. Tez that are delegated to a baker are counted in the baker’s balance of tez, and thus contribute to the number of rolls, for the purpose of obtaining validation rights in the lottery.²⁰ Delegation is an operation recorded on the Tezos blockchain, and tez holders can delegate their tez to a chosen baker with a few clicks. Delegators continue to own and control their tez and can remove the delegation, or redelegate their tez to a different baker, at any time. When tez are sent to another account, the delegation is automatically severed, and the recipient account fully controls their use.

Typically, a baker agrees to forward the block rewards attributable to a delegation of tez after deducting a portion of those rewards as a fee. The portion of block rewards withheld is set by the baker. A typical fee is 5 percent to 20 percent.

Because delegation doesn’t transfer ownership or control of tokens, bakers must themselves hold enough tez to surrender the deposits required to bake and endorse blocks. In Tezos, the risk of forfeiting deposits is borne by the baker. Deposit requirements also impose an upper limit on the amount of delegation a baker can accommodate. Over-delegated bakers will run out of tokens for deposits and miss blocks, losing out on potential rewards. The required ratio of a baker’s own tokens to those that can be accepted through delegation is a function of the size of deposits (512 to bake a block; 64 to endorse one) and how long they remain frozen. Under the current holding period of five full cycles, about 8.25 percent of a baker’s total baking balance must be held by the baker.

Deposits enforce honest validation but also have another important function. Validation is not costly in terms of computer resources; baking with 50 rolls costs nothing more than baking with five rolls. Requiring bakers to control one-twelfth of their baking balance constrains bakers’ size and helps ensure the diversity of validators required to maintain the network’s integrity.

Bakers can advertise their services on the basis of fees, performance (such as history of missed baking opportunities), and other factors.²¹ For example, Tezos bakers participate directly in Tezos governance — that is, bakers may propose, and then vote on, changes to the Tezos software. Delegated tez contribute to a baker’s rolls, and a baker holds votes equal to his number of rolls. Delegations therefore contribute to both baking rights and voting rights. Delegators may have an interest in how their baker will vote on amendments to the Tezos software and will choose their baker accordingly.

This overview might make the system of delegation and validation appear simpler than it is in practice. Recall that both proof of stake and proof of work refer to methods of allocating, or rationing, opportunities to maintain the cryptocurrency network in a way that ensures the security and the nonstop operation of the network. For our purposes, the critical piece is ensuring an adequate diversity of validators. Different proof-of-stake cryptocurrencies approach this challenge in different ways. In Tezos, anyone with 8,000 tez can be a validator. In some other systems, there is a fixed number of validators. Token holders can compete to become one of these validators by receiving the support of other token holders, similar to delegation in Tezos. Validators thus “elected” then share reward tokens with their supporters according to the supporters’ stakes.²²

E. Proof of Stake and the Language of Finance

It is helpful to pause here to address the terminology used in proof-of-stake cryptocurrency. Participating in Tezos validation results in gaining new tez — specifically, a share of the daily infusion of 115,000 new tokens into the network. But from this true statement, it is only a short step to ambiguous, misleading, and even erroneous characterizations of the new tokens received as block rewards. This is in part because the existing language of finance and tax is not perfectly suited to the economics of cryptocurrency, and in part because the economics of cryptocurrency remain novel and elusive. Confusion abounds, even among those familiar with cryptocurrency mechanics. Finance terms with a superficial bearing on the form and function of cryptocurrency incentive structures can outrun their usefulness and lead to confusion. In turn, inapt terminology can introduce further confusion about taxation.

To wit: Participants in Tezos validation get new tez. This much is true. From this, it follows that validation yields more tez, so it is easy to say that validation has a “yield.”²³ From here, one might reasonably conclude that token ownership yields a “return.”²⁴ In light of a validator’s or delegator’s ownership of valuable tez — the prerequisite for validation that lies at the heart of proof-of-stake technology — it is again a short step to describe this return as a “return on investment.”²⁵ This return can be expressed as an annualized percentage of the participant’s holdings of tez, so it is another small step to describe this return on investment as “interest.”²⁶ The return is received in tez, not dollars, enabling an analogy to corporate distributions that would render these rewards payment-in-kind “dividends.”²⁷ And because participation in validation through delegation need not involve considerable effort, expense, or even oversight, the stream of rewards might even be labeled “passive income.”²⁸

Given the novelty of cryptocurrency, these descriptions of proof-of-stake block rewards are understandable. But they can also be misleading, and not only because there are meaningful differences between, for example, “interest” and “dividends” in finance and tax. If understood by their colloquial meanings, these descriptions tend to overstate the degree to which validators or delegators are made better off.

Here is a more accurate account of Tezos block rewards: An owner of tez owns a stake in the Tezos network. A stake is fundamentally a percentage share, not merely a quantity. A stake can be expressed as a percentage of the Tezos network, in which the denominator is the total tez outstanding. The sum of all stakes is always 100 percent. Because the denominator is always changing — 115,000 tez are created each day — a stake is more conveniently, if less accurately, expressed as a quantity of tez. When referring to a stake in terms of a quantity of tez, the ever-changing denominator — total tez outstanding — is left implicit.

New tez are created and distributed as block rewards. On the one hand, these new tez dilute every tez holder’s stake. Nonparticipants receive no block rewards, so the new tez dilute their stake proportionate to the rate the new tokens are created. Participants in validation, on the other hand, offset the effects of dilution as they receive newly created tez. In this way, participants in validation may indeed increase their stake in the network. But this improvement in a validator’s position is not captured by the terms “yield,” “interest,” “dividends,” or even “income” as they were used earlier or in the cited examples drawn from the media.

Suppose that yield is defined as the annual percentage increase in one’s tokens as a result of participating in validation. Defined this way, a yield of 20 percent is not necessarily better than a yield of 10 percent. New tokens do not determine the gains from validation; increased stake does. And new tokens do not, taken alone, determine the increase in stake.

Tezos offers real-world numbers to illustrate. As noted, at the time of this writing, there are about 806 million tokens in the Tezos network.²⁹ With approximately 115,000 new tez created each day, in one year there will be about 42 million more tez. The current token creation rate is therefore around 5.21 percent.³⁰ At present, about 572 million tez participate in validation.³¹ In other words, 572 million tez, or 71 percent of the total, are held by bakers or have been delegated to bakers and therefore contribute to the number of rolls that determine validation rights.³²

The daily infusion of 115,000 tez is shared among validation participants. Because of the luck of the lottery and other factors such as missed validation opportunities, the distribution of these reward tokens will not be perfectly uniform. But on average, bakers will accumulate new tokens at the rate of 7.34 percent.³³

The validator will end the year with 7.34 percent more tez, but meanwhile, the supply of tez will have increased by 5.21 percent. The validator’s stake, therefore, will have increased by just 2.02 percent.³⁴ The importance of this distinction — between an increase in tokens and an increase in stake — is best illustrated through the question of taxation.

F. Options for the Taxation of Block Rewards

There are three basic approaches to the taxation of block rewards, which this report addresses in the following sections:

1. include in gross income the FMV of the block rewards as of the date they are received by the taxpayer;

2. include in income the taxpayer’s actual new wealth accrued over the course of the year; or

3. defer taxation until the sale or exchange of the taxpayer’s tokens.

III. Option 1: Gross Income When Received

Option 1 has the advantage of clarity and of following the well-worn example of property received for services in many other contexts. This is the option announced by the IRS in Notice 2014-21:

Q-8: Does a taxpayer who “mines” virtual currency (for example, uses computer resources to validate Bitcoin transactions and maintain the public Bitcoin transaction ledger) realize gross income upon receipt of the virtual currency resulting from those activities?

A-8: Yes, when a taxpayer successfully “mines” virtual currency, the fair market value of the virtual currency as of the date of receipt is includible in gross income.³⁵

The guidance refers to “mining” and the example of Bitcoin; when the notice was issued in early 2014, proof-of-stake cryptocurrencies were in their infancy.³⁶ Although the notice does not provide the reasoning behind its conclusion, the language fits with the idea that block rewards are a form of compensation offered in exchange for the taxpayer’s “activities” maintaining the network. More precisely, on the assumption that cryptocurrency is property and not real currency for tax purposes,³⁷ the reference is to barter. This approach has intuitive appeal. Some transactions result in income for tax purposes even if no money changes hands. If someone washes your car in exchange for a bushel of wheat, that wheat is your gross income, measured at the wheat’s FMV at the relevant time.³⁸ It would seem to follow that if a person helps secure a cryptocurrency network to obtain some tokens, those tokens must also be gross income, measured at the tokens’ FMV when they are received. The analogy of a corporation making distributions of its own stock also seems apt, and generally those distributions are gross income at the time received.

Despite their intuitive appeal, these options have problems. The report returns to the issues of corporate distributions and barter compensation in Section V. For now, our focus is on the economic consequences of viewing block rewards as taxable compensation for validation services or as some other form of gross income.

A. Dilution and Redistribution

It is true that newly minted tokens have economic value, and we can further assume that they also have a reasonably determinable FMV at any given time. The problem is that block rewards convey value to validators through the combined effects of dilution and redistribution. Increasing the token supply means that new tokens dilute the value of all tokens. Each new token reduces the stake in the network represented by each existing token, and in turn each existing token loses value because it represents a smaller fraction of total network value, defined as the value of a token multiplied by the total number of tokens. Counting the new tokens as taxable gross income can overstate the recipient’s new wealth if it ignores the new tokens’ dilution effect on all tokens. Moreover, block rewards are a zero-sum game, in the sense that the receipt of new tokens increases validators’ wealth only insofar as it redistributes network value away from non-validators.

These two critical features of cryptocurrency economics are explored in depth later, but the core insight concerning the effect of dilution is intuitively simple. Suppose that a cryptocurrency were to add two zeroes to the balance of every account in its ledger. One could say that new tokens have been created, and indeed token holders would have 100 times more tokens than before. But it would seem odd to say these new tokens constitute income of any sort.

But this is close to what cryptocurrencies actually do. The computer code behind a cryptocurrency can be programmed any which way: It can enable the creation of many new tokens to encourage network maintenance, or it can enable the creation of very few new tokens. And if new tokens are counted as gross income at their FMV when they are received, the cryptocurrency that creates many new tokens will create more immediately taxable gross income than the cryptocurrency that, all other differences aside, creates few new tokens.

This difference can be measured. A cryptocurrency that increases its token supply by 50 percent over the course of one year will, during that year, “create” gross income under the IRS’s definition equal to 41.66 percent of that network’s value.³⁹ An otherwise identical cryptocurrency that increases its token supply by 10 percent over the course of one year will “create” gross income equal to just 9.45 percent of that network’s value. This curious fact — that fixing a variable in a cryptocurrency’s code base might determine the total taxable income resulting from efforts to maintain that network — should indeed call into question the notion that reward tokens should be income when received.

From the standpoint of cryptocurrency design, the token creation rate is not arbitrary. One token creation policy might be better than another. In short, the higher the creation rate, the more serious the consequences of choosing whether to participate in validation. A higher token creation rate always increases the penalty for not validating. But as we’ll see, a higher creation rate does not necessarily increase the gain to be had by validating. So the token creation rate is not arbitrary but, taken alone, neither is it determinative of the new wealth enjoyed by reward tokens’ recipients. Ultimately, a cryptocurrency’s token creation rate is not a reliable proxy for the taxpayers’ new wealth that Treasury intends to tax.

Of course, gross income is not the same as taxable income. Gross income can be offset by deductions such as business expenses, and losses when they are realized can offset gains. Accordingly, the proof of inequitable taxation is not quite as simple as the observations above might suggest. The full argument is developed later, but as we’ll see, the inequity does indeed persist when we delve deeper into the implications of counting block rewards as immediate gross income.

First, it’s worth noting that Notice 2014-21 mentions the use of “computer resources” to describe the taxpayer’s “activities” that maintain a cryptocurrency network.⁴⁰ With Bitcoin, those resources — principally electricity and computer hardware — are indeed a significant business expense. Moreover, if the value of rewards rises — because of either the creation of more bitcoins or an increase in the value of each bitcoin — total expenditures on power and hardware can be expected to rise as well. Compared with Bitcoin, the computer resources required to maintain the Tezos network are decidedly minimal, and in any case their costs are not in proportion to the rewards from baking. The cost of computer activities in baking — which can be as low as running a spare laptop connected to the family Wi-Fi — can safely be ignored in the analysis of taxation policy. For delegators, this cost approaches zero: Tez holders perform the delegation with a few clicks, using the same digital wallet used to receive, store, and send tez. In the proof-of-stake examples that follow, we also ignore computer costs. Of course, these costs might nonetheless be deductible business expenses.

Second and more importantly, the inequity persists because if block rewards create gross income, the dilution effects of the new tokens are treated asymmetrically under the tax code. New tokens assume the form of (immediately taxable) gains, while the offsetting losses caused by dilution remain trapped as paper losses until the diluted tokens are liquidated. In the meantime, taxpayers must account for, and pay taxes on, income that overstates the extent to which block rewards increased the taxpayers’ wealth.

Administrative costs are also relevant here. Potentially, every Tezos token holder would be required to account for, and report, annual income. Adding insult to injury, in every case the gross income thus reported will overstate the extent to which the tez holder actually benefited from participating in Tezos validation. This remains true regardless of the change in value of the Tezos network, and of individual tokens, resulting from effects other than dilution. This is important because in practice, the volatility of a cryptocurrency may well exceed the effect of dilution, helping to mask that effect.

Our first hypothetical below illustrates the dilution effect. It demonstrates that block rewards, if treated under Notice 2014-21, are not a reliable proxy for the increased wealth that the law seeks to tax. This example also provides the relationship between a token creation rate and the gross income the IRS guidance would deem realized by validators’ receipt of block rewards. Next, we turn to a slightly more nuanced example that illustrates the combined effect of dilution and redistribution. This second example shows that even when block rewards improve a taxpayer’s financial position, this gain cannot accurately be measured in terms of the tokens received.

B. Standard Oil Coin and the Effect of Dilution

1. Standard Oil Coin.

The following is an example of a simple proof-of-stake protocol that allocates validation responsibilities according to stake in the network. Our hypothetical cryptocurrency, Standard Oil Coin, consists of 500,000 tokens distributed among a number of owners. Every owner of these tokens participates in validation, and the cryptocurrency’s protocol provides that new tokens will be created in the form of block rewards. The protocol provides for the token supply to increase by 250,000 over the first year, so that at the end of year one, there are 750,000 tokens in circulation.

In this simple system, validation responsibilities and block rewards are distributed according to each owner’s share of the network. One coin owner, Myrtle, initially owns 2,200 tokens, so when the token supply has increased by 50 percent at year-end, she owns a total of 3,300 tokens.

Should Myrtle’s 1,100 new tokens be gross income? The example should evoke the intuition that perhaps they should not. Myrtle has 1,100 new tokens, but her stake in the network remains constant at 0.44 percent. To start, she owned 2,200/500,000 of the network; at year-end, she owns 3,300/750,000.

To borrow from the nearly 100-year-old Supreme Court decision that inspires this hypothetical:

The proportional interest of each [stake]holder remains the same. The only change is in the evidence which represents that interest, the new [tokens] and the original [tokens] together representing the same proportional interest that the original [tokens] represented before the issue of the new ones.⁴¹

In this well-known case, Myrtle Macomber owned 2,200 shares in Standard Oil Co. of California. All shareholders received a 50 percent pro rata stock dividend, which brought Macomber’s holding of Standard Oil shares to 3,300. Congress’s intent to tax those stock dividends was not in doubt. Macomber paid the tax under protest and filed suit.

Ruling for Macomber, the Supreme Court reasoned that regardless of Congress’s intentions, its taxation power could not reach Macomber’s stock dividend unless those shares were in fact income under the recently adopted 16th Amendment authorizing the federal income tax. And on this question, the Court held that Macomber’s stock dividend was not income, because her new stock certificates and the company’s transfer of corporate funds from a surplus account to a capital account did not change her proportionate share of Standard Oil ownership.

As with the doubling of one’s shares in a 2-for-1 stock split, since Macomber, pro rata stock dividends have been understood to create no taxable income. The consensus today is that Macomber is no longer much, if any, constraint on Congress’s power to tax stock dividends as it wishes.⁴² We can concede as much: We are concerned with whether Myrtle’s new tokens should be considered income, not with whether Macomber might limit Congress’s authority to tax new cryptocurrency tokens as it wishes. We will return to this case and the example of stock dividends later in this report,⁴³ but regardless of Macomber’s status, it is invoked here just to illustrate a commonsense principle: Not every instance of printing new, larger numbers in a ledger should be understood to create income.

To clarify the lesson of Standard Oil Coin and Macomber, let’s assume that block rewards are indeed gross income when received. Taxes are paid in U.S. dollars, so it helps to view the matter in dollar terms. Let’s assume that Myrtle, and all token holders, purchased the initial supply of tokens for $100 each, so that the network of 500,000 tokens is valued at $50 million.

Of course, we can’t assume that tokens will retain their $100 value over the course of the year. Otherwise the total network value would increase by 50 percent simply by the infusion of 50 percent more tokens, and our hypothetical cryptocurrency would be a rather remarkable wealth-printing machine. We assume, therefore, that network value remains constant at $50 million throughout the year.

Note, however, that the argument does not depend on an unchanging network value. The value of Standard Oil Coin tokens can change for any reason, but the effects explained below remain proportionate to the token holders’ stakes in the network. In this and future examples, including the real-world case of Tezos, keeping the valuation of the network constant is helpful for illustration purposes but is not required.

On day 1, Myrtle owns 2,200 tokens worth $100 each, or $220,000. Block rewards are distributed consistently over time as new blocks are added to the blockchain. The first token created as a block reward will therefore have a market value just shy of $100 when it is created.⁴⁴ Over the course of the year, the value of each token Myrtle receives — again, at the time she receives it, so that the total network value remains constant at $50 million — will decline. At the end of the year, the final token that Myrtle receives as a block reward will be worth about $67.⁴⁵

Over the full year, the FMV of tokens when they are received will average about $83.⁴⁶ Myrtle will have received 1,100 new tokens at an average market value of $83. If block rewards are gross income when they are received, Myrtle will need to report $91,300 in gross income.

Of course, Myrtle isn’t actually better off by $91,300. Myrtle’s tokens have increased in number from 2,200 to 3,300, but at year-end she continues to own 0.44 percent of the tokens, and her stake in the Standard Oil Coin network is still worth $220,000.

The discrepancy arises because Myrtle’s nominal gain of 1,100 tokens valued at $91,300 is offset by the decline in value of all her tokens. Myrtle’s initial 2,200 tokens are now worth about $67 each, or $146,700. And while the new tokens add up to $91,300 in gross income when valued on the date they were received, these new tokens also decline in value after their receipt as the dilution continues over the course of the year. At the end of the year, Myrtle’s 1,100 tokens in block rewards are also worth $67 each, for a total of $73,300. Myrtle ends the year with 3,300 tokens, each worth $67, for a total of $220,000.

2. Modeling total gross income under Notice 2014-21.

The Standard Oil Coin hypothetical provides us with the relationship between a token creation rate and the total amount of gross income that, applying Notice 2014-21, would be created as a result of the new tokens.

Let T be the token creation rate of a cryptocurrency for the ensuing year, expressed as a fraction of the token supply at the start of the year in question. The stake in the network represented by one token declines over the course of the year, and in turn the market value of tokens will also change. The FMV of new tokens when they are created can be expressed as a function of the average token. If the first new token is worth 1, the last new token will be worth 1/(T + 1). The average token will be worth (1 + (1/(T + 1))/2). Multiplying the token creation rate by this average token value gives us T(1 + (1/(T + 1))/2), which is the total value of all the new tokens created over the year, with their values calculated when the new tokens are created and received by validators as block rewards. The new tokens’ total value is expressed as a fraction of the total value of the cryptocurrency network.

So if we apply Notice 2014-21 to a cryptocurrency with a linear token creation rate, we can establish the total gross income created in the course of one year, expressed as a fraction of the total network value, as GI = T((1 + (1/(T + 1)))/2).

Applied to Standard Oil Coin with its 50 percent token creation rate, the gross income will be 41.6 percent of the $50 million network value, or $20,833,000. The IRS would expect to see $21 million in reported gross income — each year — even though no token holder is any wealthier than before.⁴⁷

3. Modeling the dilution of all tokens.

The token creation rate also provides us with the rate at which all tokens are diluted given an influx of new tokens at rate T. As noted, the final token created as a block reward will have a value of 1/(T + 1), because at the end of the year, every token will have a value of 1/(T + 1). The dilution, or D, of tokens is given by the equation D = (1 - (1/(T + 1)).

If T = 1, over the course of one year, the initial token supply is diluted by a factor of 0.5, or by 50 percent. If T = 0.5, the initial supply is diluted by one-third; for T = 0.1, dilution is 0.0909, or 9.09 percent.

C. Hardship Coin and the Effect of Redistribution

1. Hardship Coin.

The earlier hypothetical assumed that block rewards were distributed pro rata among all token holders, so that every token holder ended the year with the same percentage stake in the network. Those assumptions, inspired by the facts in Macomber, illustrate the limiting case in which the creation of new tokens never results in any taxpayer’s increased wealth due to validation even as the total gross income under Notice 2014-21 remains a function of the token creation rate.

More realistically, we can imagine a cryptocurrency in which not all token holders participate in validation. Building on the dilution effect illustrated by Standard Oil Coin, this example illustrates the redistribution effect of block rewards.

In this example, Myrtle does indeed improve her situation as a result of participating in validation. Critically, Myrtle’s gain is purely a redistribution of stake in the network, and her gain depends on how many validators share in that redistribution. Myrtle’s gain from validation, in other words, comes at the expense of non-validators, and the extent of that gain depends on how many other token holders are also validators. But, as in the previous example, Myrtle’s block rewards continue to overstate her new wealth if they are valued when they are received.

Hardship Coin follows the same assumptions as before except that participation in validation is 50 percent. That is, validation rights and therefore block rewards are distributed among the owners of 250,000 of the initial supply of 500,000 Hardship tokens. With 50 percent participation and keeping the creation of new tokens at 250,000 over the first year, all the new tokens created as block rewards will go to validators, who will double their total holdings to 500,000 tokens. Non-validators, meanwhile, will continue to hold a total of 250,000 tokens.

We again assume that Myrtle participates in validation, and therefore her initial 2,200 tokens increase in number to 4,400 at year-end. Unlike her result with Standard Oil Coin, here Myrtle does indeed end the year holding new wealth caused by her participation in validation. Myrtle begins the year with 2,200 tokens and a 0.44 percent ownership stake in the network and ends the year with 4,400 tokens and a 0.587 percent stake.⁴⁸ Her stake has increased by 33 percent.⁴⁹

To understand the implications for taxation, it helps to again view the example in dollar terms. We assume the value of the network stays constant at $50 million, so at year-end, each of the 750,000 tokens in the network is worth $67. (Again, this assumption is not required. If the value of the network triples over the course of the year, the effects of dilution and redistribution remain the same, just a bit more difficult to capture mathematically.) Myrtle’s initial $220,000 of network value will increase to $293,300.⁵⁰ Her true gain is $73,300, or 33 percent.

If the goal is to tax Myrtle’s new wealth, this figure of $73,300, reflecting a 33 percent increase in her stake, is the figure we should be looking for. But if block rewards are gross income at the market value of those rewards on the date they are received, Myrtle will not show just $73,300 of gross income. Because the network value remains constant at $50 million, the value of Myrtle’s first reward token will be $100 when it is received. At year-end, the value of Myrtle’s last reward token will be $67, and the average of value of her 2,200 reward tokens when received will be $83.

This results in gross income of $183,300.⁵¹ As with Standard Oil Coin, the discrepancy arises because Myrtle’s nominal gain of 2,200 tokens worth $183,300 at the time of their creation is offset by the real decrease in the value of all her tokens caused by the 50 percent increase in the token supply. That decrease in value amounts to $110,000.⁵²

Hardship Coin is named for the plight of the taxpayer in the 1921 case Phellis.⁵³ In a dividend resulting from a reorganization, Mr. Phellis received an amount equal to almost 90 percent of his invested capital as shares in the new corporation, and his original shares showed an equivalent decline in value. Treasury taxed the new shares as income. On the unusual facts of the case, the unfairness — or at least the misfortune — was hard to ignore. Ruling against the taxpayer, Justice Mahlon Pitney allowed that “the possibility of occasional instances of apparent hardship in the incidence of the tax may be conceded.”⁵⁴ With Hardship Coin the hardship would be neither occasional nor merely apparent or possible but rather persistent and genuine.

2. Hardship Coin and unrealized losses.

A central argument of this report is that block rewards should not be included in gross income at their value when received in the manner presented by Notice 2014-21. As just seen, defining gross income in this way overstates the gain received by validators in that tax year and results in overtaxation. As noted, for sake of simplicity, we are ignoring computer resource costs involved in proof-of-stake validation. These and other costs might be deductible from gross income as business expenses, but for proof-of-stake validation, these costs are not critical to the taxation analysis.

More consequentially, we are also assuming that validators’ losses in the form of diluted tokens remain unrealized and therefore cannot offset those same validators’ gains in the same tax period the block rewards are received. This assumption is warranted under current law, and Notice 2014-21 confirms that cryptocurrency tokens are property subject to gain or loss treatment according to their value when they are sold or exchanged.⁵⁵

Returning to the example of Myrtle and Hardship Coin, under Notice 2014-21, Myrtle will record gross income of $183,300 even though her wealth has increased by just $73,300. The difference is explained by the decrease in value of all Myrtle’s tokens as the result of dilution. If, at the end of the tax year, Myrtle liquidates her 4,400 tokens, she will raise $293,300. Her basis in these tokens will be $403,300, resulting in a loss of $110,000. Accordingly, Myrtle presumably could offset her $183,300 of gross income with a loss of $110,000 on all her tokens, resulting in an adjusted income figure of $73,300.

We acknowledge, therefore, that the overtaxation of block rewards under Notice 2014-21 is a function of the tax code’s asymmetric treatment of the effects of token dilution: New tokens are gross income today, while any losses because of dilution are unrealized until tokens are liquidated. In principle, higher taxable income this year might be offset eventually, when tokens are ultimately liquidated. Moreover, in principle, this asymmetry could be avoided by liquidating cryptocurrency holdings to realize losses as well as gains, and then repurchasing the entire stake, establishing a new basis in the token holdings.

These are not plausible rejoinders to the fact of overtaxation. If tokens are property — and especially if tokens are capital assets in the hands of most taxpayers⁵⁶ — they should be subject to equitable taxation when the taxpayer chooses to sell them. We have now seen how treating reward tokens as income leads to overtaxation because of the differing treatment of immediate gains and unrealized losses, but there are other approaches to the dilution effect, such as treating tokens as depreciating assets or by mandating annual mark-to-market treatment. While ultimately infeasible and undesirable, those approaches are addressed in Section IV. First, we will model the lessons of Hardship Coin and then apply these models to the real-world example of Tezos.

D. Modeling Token Holders’ Gains and Losses

1. Modeling the redistribution effect.

The redistribution effect shown in the Hardship Coin example can be modeled, revealing the fraction of network value that block rewards redistribute from non-validators to validators. Redistribution, or R, is a function of two variables. One is the validation participation rate, or V, which is the fraction of the initial supply of tokens that participate in validation. The second variable is T, introduced above: the token creation rate.

As noted, validators’ gains equal non-validators’ losses, and non-validators’ losses are equal to the decrease in non-validators’ stake caused by dilution. In other words, R is equal to the decrease in stake and therefore the decrease in value of non-validators’ tokens. As demonstrated earlier, the equation 1/(T + 1) gives us the value of each token at year-end, and therefore (1 - (1/(T + 1))) gives us the decrease in a token’s value over the course of the year expressed as a fraction of the network value. Non-validators hold (1 - V) of the initial supply of tokens. Accordingly, R = (1 - V)(1 - (1/(T + 1))) gives us the fraction of total network value that is redistributed from non-validators to validators.

In Hardship Coin, in which T = 0.5 and V = 0.5, redistribution is therefore 0.1666 of the network value. As suggested earlier, if the goal is to tax new wealth represented by the receipt of block rewards, this figure R, or 0.1666 of the value of the network, is the figure we should be looking for as the total new wealth of reward token recipients caused by the creation of new Hardship tokens in the course of a year.

2. Modeling token holders’ actual gains and losses.

The variables T and V also provide us with the change in wealth experienced by token holders.

a. Validators’ gain: VG = R/V = ((1 - V)(1 - (1/(T + 1))))/V.

We can define validators’ gain, or VG, as the fraction by which a validator’s stake increases in a year because of participation in validation. Because R is distributed among validators, each validator’s gain is given by R/V. Because R = (1 - V)(1 - (1/(T + 1))), we can also write this equation as VG = ((1 - V)(1 - (1/(T + 1))))/V. This equation represents token holders’ actual gains from participating in validation.

b. Non-validators’ loss: NL = (1 - (1/(T + 1))).

Unlike validators’ gains, the decrease in stake experienced by non-validators can be established without reference to the validation participation rate. Losses are determined solely by the token creation rate, which determines the rate at which dilution reduces the stake, and therefore the value, of all tokens. We have established that dilution, or D, is given by the equation D = (1 - (1/(T + 1))). Non-validators’ loss is given by the same equation: NL = (1 - (1/(T + 1))).

3. Modeling validators’ gross income under Notice 2014-21.

With the addition of the variable V, we can also establish the gross income that each validator will report if guided by Notice 2014-21. We previously established that the total gross income is given by GI = T((1 + (1/(T + 1)))/2), which for a T of 0.5 yields gross income of 0.4166 of the network value. This is shared among validation participants at the rate of GI/V. If the participation rate is 50 percent, validation participants will receive gross income of 0.4166/0.5, or 0.833 of the validator’s initial stake. This is the figure we saw for Myrtle, whose starting point of 2,200 Hardship tokens worth $220,000 resulted in gross income of $183,300, a “gain” of 83.3 percent.⁵⁷

4. Comparing gross income under Notice 2014-21 with actual gains.

Figure 1 depicts the economic gain from participating in the maintenance of the Hardship Coin cryptocurrency network. Using the models just developed, two measures of gain are presented. The first follows Notice 2014-21 and treats block rewards as gross income at their market value when received. The second presents the actual gain from block rewards after accounting for the dilution and redistribution effects of the new tokens.

As discussed, gains from participating in network maintenance — that is, gains from acting as a validator — are a function of the token creation rate and the validation participation rate. We assume a linear, consistent token creation rate, and in the Hardship Coin example, the token creation rate is 0.5 — that is, over the next year, the number of Hardship tokens created as block rewards will increase the token supply by 50 percent. The validation participation rate reflects the percentage of the initial supply of tokens that participate in the validation process. The owners of these tokens — the validators — receive block rewards, while non-validators do not. To keep the numbers simple, we assume that validators’ tokens participate in validation 100 percent of the time and that non-validators’ tokens don’t participate at all. Statements of a validator’s gain are given as a percentage of the validator’s initial stake in the Hardship Coin network.

Validation participation rate	10%	50%	90%
Validator’s gross income under Notice 2014-21a	417%	83.3%	46.3
Validator’s actual gain accounting for dilutionb	300%	33.3%	3.7%
Overstatement of actual gain given by Notice 2014-21	39%	250%	1,250%
Non-validator’s lossc	33.3%	33.3%	33.3%
aA validator’s gross income under Notice 2014-21 is given by (T((1 + (1/(T + 1)))/2))/V. See supra Section III.D.3. bA validator’s actual gain is given by ((1 - V)(1 - (1/(T + 1))))/V. See supra Section III.D.2.a. cA non-validator’s loss from dilution is given by (1 - (1/(T + 1))). See supra Section III.D.2.b.

E. Dilution and Redistribution in Tezos

We now have the tools to analyze the consequences of Notice 2014-21 for a real-world proof-of-stake cryptocurrency such as Tezos. The facts of Macomber, involving a 50 percent pro rata stock dividend, led us to use a seemingly dramatic 50 percent token creation rate for our hypothetical Standard Oil Coin and Hardship Coin cryptocurrencies. Tezos demonstrates that treating block rewards as immediate gross income results in meaningful overstatement of income even for lower, arguably more realistic token creation rates.⁵⁸

As of this writing, about 806 million native tokens exist on the Tezos network.⁵⁹ Of those, 572 million participate in validation, and approximately 42 million new reward tokens will be created in the next year. At the moment, therefore, Tezos has a token creation rate, or T, of 0.0521, and a validation participation rate, or V, of 0.71.

Figure 2 depicts the gains from participating in the maintenance of the Tezos network given the current token creation rate of 5.21 percent.

Validation participation rate	71%
Validator’s gross income under Notice 2014-21a	7.156%
Validator’s actual gainb	2.02%
Overstatement of actual gain given by Notice 2014-21	354%
Non-validator’s lossc	4.952%
aA validator’s gross income under Notice 2014-21 is given by (T((1 + (1/(T + 1)))/2))/V. See supra Section III.D.3. With a 5.21 percent token creation rate, Tezos token holders who delegate their tez to a baker who retains 10 percent of block rewards as a fee will show gross income of 6.44 percent of their stake. bA validator’s actual gain is given by ((1 - V)(1 - (1/(T + 1))))/V. See supra Section III.D.2.a. With a 5.21 percent token creation rate, delegators subject to a baker’s fee of 10 percent will increase their stake by 1.82 percent. cA non-validator’s loss from dilution is given by NL = (1 - (1/(T + 1)). See supra Section III.D.2.b.

Figure 2 presents the results as a percentage of tez holders’ stake, or equivalently as the annualized rate of change. They are independent of any increase or decrease in the total value of the Tezos network. That is, if the value of the Tezos network triples over the course of the year — or perhaps declines by two-thirds — validators will still see their stake in the network increase by 2.02 percent, and non-validators will still see their stake decline by 4.952 percent.

With this in mind, we can still clarify the comparison by assigning a dollar value to the Tezos network. At the time of this writing, the market value of Tezos tokens has ranged between $0.34 and $1.84 over the past year.⁶⁰ For simplicity, we can set the network value based on an initial token value of $1, for a network value of $806 million. Figure 3 shows that the total of validators’ actual gains declines as participation increases, while total gross income under Notice 2014-21 remains unchanged.

Validation participation rate	71%
Total annual gross income under Notice 2014-21 regardless of participation ratea	$40,950,000
Total of validators’ actual gains, equal to redistribution from non-validatorsb	$11,575,000
Overstatement of actual gain given by Notice 2014-21	354%
aUnder Notice 2014-21, the total annual gross income from the creation of new reward tokens is a function of the token creation rate regardless of the validation rate, and is given by (T((1 + (1/(T + 1)))/2))*(network value). See supra Section III.D.3. With a token creation rate of 5.21 percent, Tezos gross income is therefore 5.08 percent of the network value of $806 million, or $40.95 million. bRedistribution, or total gains to validators, is given by (1 - V)(1 - (1/(T + 1)))*(network value). See supra Section III.B.1. With a token creation rate of 5.21 percent, validators’ total gains decrease as participation in validation increases, as shown by the slope of the curve. With a validation participation rate of 71 percent, total gains are 1.436 percent of the network’s value of $806 million, or $11,575,000.

F. Burn Coin: A Deflationary Cryptocurrency

Block rewards overstate proof-of-stake validators’ income if validators’ losses caused by dilution are not captured in the tax period those losses are incurred. In Section IV this report considers other methods to acknowledge the effects of dilution in order to make annual taxation equitable. And arguably, taxing block rewards as immediate income is required under the tax code, even if it is inequitable, inefficient, and administratively burdensome. The report takes up that issue in Section V.

All else equal, of course, self-interested taxpayers would also prefer not to need to annually report gains as income in the first place, even if those gains are equitably taxed. So it is helpful to demonstrate that because computer code is flexible, a cryptocurrency could be designed to encourage its maintenance without resorting to the creation of new tokens. This possibility preaches caution for any approach to taxation that would elevate the form of cryptocurrency incentives (for example, new tokens) over their function (that is, encouraging participation in validation through redistribution of stake in the network).

A cryptocurrency can be programmed to create new tokens, but it can also be engineered to destroy, or burn, tokens. As in our previous hypothetical cryptocurrencies, Burn Coin starts with 500,000 coins each purchased for $100. But instead of increasing the token supply with newly created tokens, this cryptocurrency decreases the token supply by withdrawing and burning tokens from accounts that don’t participate in validation.

Suppose that the validation participation rate is again 50 percent and that the supply of tokens shrinks by 125,000 over the first year. At year-end, 375,000 tokens will remain in circulation. At the end of year one, validators continue to hold 250,000 tokens. Non-validators, however, see their account balances reduced by a total of 125,000 tokens, so that these non-validators’ tokens dwindle in number from 250,000 to 125,000. Validators now own two-thirds of the network (250,000 of the remaining 375,000 tokens). We assume the network value holds steady at $50 million, so validators’ tokens are now worth $33,333,333, a gain in stake of 33 percent that, with a constant network valuation, is also a 33 percent gain in wealth. Non-validators’ tokens are worth just $16,666,666, for a 33 percent loss in stake and value.

Suppose Myrtle buys 2,200 tokens in the Burn Coin network and participates in validation. At year-end she will still have 2,200 tokens, except her stake will have increased by 33 percent and her tokens will now be worth $293,300.

For validators like Myrtle, the deflationary Burn Coin functions the same as the inflationary Hardship Coin. In both cases, Myrtle will have the same $73,300 gain. But now Myrtle has no new tokens to show for it. Under a tax policy governed by form over function, there’s nothing to point to as a gain subject to immediate taxation.

But Burn Coin isn’t perfect. Let’s suppose that because she received no new tokens, Myrtle at year-end shows no gross income from her increased stake in Burn Coin. She’s happy about this, but there is a potential loser in the bargain. The result is to shift the taxation consequences of the incentive mechanism to a different group. Instead of adding tokens to dilute the value of all tokens, now we are subtracting tokens to concentrate the value of all remaining tokens. The deflationary effects of withdrawing tokens from the network exactly mirror the inflationary effects of creating new tokens under the Hardship Coin example. The result is that Burn Coin validators no longer hold new tokens subject to immediate reporting as income; instead, non-validators have been stripped of the tokens that demonstrate the extent of their loss.

Suppose for a moment that Myrtle buys Burn Coin tokens but doesn’t participate in validation. Over the course of the year, her 2,200 tokens dwindle to 1,100, each now worth $133. Suppose she then liquidates her stake, selling her 1,100 remaining tokens for $146,700. Now she tries to capture a loss of 33 percent, calculated as her $220,000 purchase less the sale proceeds of $146,700. By hypothesis, however, the law is concerned with tokens, not stake. Instead of reporting a loss of $73,300, Myrtle will show the sale of 1,100 tokens, each worth $133, for a gain of 33 percent or $36,600.

Myrtle can capture her true loss if the tax authorities recognize she has sold 100 percent of her stake, or, to the same end, if they allow for an adjustment in the basis of her remaining tokens to account for the burned tokens. Perhaps the loss of tokens in this scenario is so salient that we should expect the victim of token concentration to prevail on the argument that stake, and not nominal tokens, should determine tax liability. This as-yet hypothetical scenario, however, is indeed symmetrical to the case of new tokens, and under the stated assumptions the inflationary Hardship Coin and the deflationary Burn Coin are economically equivalent.⁶¹

G. Haig-Simons Coin: A Coin Without Tokens

With inflationary cryptocurrencies, the challenge has been to illustrate why block rewards, because of the dilution effect, shouldn’t be valued at the time of their creation for taxation purposes. With a deflationary incentive structure, the challenge is to illustrate that burned tokens should be recognized as a factor in calculating taxpayers’ gains or losses when the tokens are sold or exchanged.

Both challenges can be stated as the problem of the denominator: When viewed in isolation from the changing number of tokens extant in a network, tokens become unmoored from the stake in the network they ultimately represent.

The lessons of Hardship Coin and Burn Coin suggest a final hypothetical. Haig-Simons Coin avoids the denominator problem by incorporating the denominator into its ledger entries. In this cryptocurrency, individual account balances do not show tokens, the total of which would provide a denominator for the calculation of holders’ proportional stakes.

Instead, ledger entries in Haig-Simons Coin show only a number, and the numbers add up to 100. Each stakeholder owns a share of the network, expressed as a number that can also be read as the percentage of network ownership. Named after the influential theory of income taxation that would tax taxpayers’ changes in wealth annually,⁶² the structure of the Haig-Simons Coin lends itself to taxation under its namesake policy but not to taxation under existing law. The hypothetical cryptocurrency also illustrates the zero-sum nature of reward tokens and other methods of encouraging network maintenance.

We start again with a network valued at $50 million. Myrtle pays $220,000 for a 0.44 percent stake. Another taxpayer spends $5 million to purchase a 10 percent stake. Otherwise, the hypothetical unfolds as before: The protocol is engineered to reward validators and penalize non-validators in a manner that mimics the effects of Hardship Coin’s 50 percent token creation rate or Burn Coin’s 25 percent token burn rate. Accordingly, with 50 percent participation, the stakes of Haig-Simons Coin validators and non-validators change at the rate of +33 percent and -33 percent, respectively. If Myrtle participates in validation, she’ll end the year with a 0.587 percent stake. If the 10 percent stakeholder doesn’t participate, his stake will dwindle to 6.666 percent.

To identify income under this arrangement, the tax authorities would have to look beyond labels; there are no tokens to be found. But if the tax authorities were to interpret stake as units of ownership, it would indeed be possible to accurately tax validators’ gains. Myrtle would gain 0.147 units as a result of validation,⁶³ and those new units could be taxed.

Haig-Simons Coin is unwieldy because of the contortions that would be required to reduce it to computer code and also because of its unfamiliar accounting system. It just doesn’t fit with our intuitions about how the world should work, and it should probably remain nothing more than a thought experiment. But it helps illustrate how reward tokens, through the combined effects of dilution and redistribution, are ultimately tools to allocate stake in a cryptocurrency network to those who participate in the network’s maintenance. By incorporating the denominator into the protocol’s statement of stakeholders’ share in the network, Haig-Simons Coin prevents the misreading of nominal tokens as real gains.

But Haig-Simons Coin also illustrates the flip side of the same token, which is that redistribution is a zero-sum game. Using the above assumption of 50 percent participation, for every Myrtle whose stake increases from 0.44 units to 0.587, there is a non-validator whose initial 0.44 units decrease to 0.293 units at the end of the year. If Myrtle shows 0.147 units of income, how should the non-validator’s loss of 0.147 units be treated? Section IV addresses an option for taxing gains through an annual tax on changes in wealth, and it explains why this is not an attractive option for cryptocurrency block rewards.

* * *

Including proof-of-stake block rewards in gross income results in inequitable overtaxation. As shown by Burn Coin and Haig-Simons Coin, this overtaxation is not an ineluctable consequence of encouraging network maintenance through the allocation of cryptocurrency stake to those who help with that maintenance. Rather, the overtaxation is an accident of the method of accounting that cryptocurrencies to date have chosen to achieve this adjustment of token holders’ stakes — namely, the creation of new tokens when before there was nothing. On their surface, these cryptocurrency tokens appear to be like other indicia of value, such as dollar bills or the dollar figures in a ledger representing a bank’s liabilities to its customers. But these metaphors are incomplete. As demonstrated, cryptocurrency tokens behave differently.

If the tax law requires these new tokens to be treated like dollars, the law should be changed. Failure to do so would discourage U.S. taxpayers’ participation in this new technology, at least until and unless cryptocurrencies’ mechanics are altered so that the metaphors used to understand their functioning no longer lead to inequitable taxation.

Congress generally intends taxation to extend close to its constitutional limit of “incomes, from whatever source derived.”⁶⁴ Part 2 of this report explores options for the equitable taxation of cryptocurrency reward tokens and how they might be supported by existing policies and principles.