Work tokens as a breakthrough in financial instrument design
At AmaZix advisory, one of our specialties lies in the study of token economics. Over our 2+ years in the industry, our team has broken down and analyzed well over 300 token economic models as well as playing a key role in designing and refining the token economic models of many of our 130+ partners.
Our experiences lead us to believe that token economics is still one of the most underrated, misunderstood and important areas in crypto. Unlike the internet and other technologies that came before it, blockchain is not merely a technical innovation but also an economic one, creating transparent, trust-minimized programmatic incentive systems that allow disparate entities to cooperate to achieve common goals while acting only in their own self-interest. Token economics concerns the study of the various token models projects use to achieve these goals; how they work, what tradeoffs they make, and how these cryptoassets, which in many cases represent entirely novel financial instruments, actually accrue value.
I’ve written extensively about token economics: why they’re important, the single biggest problem with them (part 1, part 2 — I lied, there are two single biggest problems) as well as an analysis of Ethereum’s token economics. In my most recent piece, I classified all known token economic models into a taxonomy, including key features and potential valuation methodologies for each one.
In this piece, I’ll spend a bit more time on “Work Tokens” which I believe are the most interesting and crypto-native token economic models out there, representing a new kind of financial instrument. I will begin by defining work tokens and discussing their features at a high-level. I will then delve into specific ways they can be used, with particular focus on the discount token, an underrated category of work token with characteristics which I believe make it one of the most interesting and widely applicable token models out there.
Work tokens — Innovations in financial instrument design
Financial instruments such as bonds, equities, derivatives, etc. are all “passive” instruments, in that ownership of the instrument, achieved through an investment of capital, is all that’s required to reap the legally enforceable benefits of that instrument (ownership of underlying assets, capital appreciation, voting, cashflow distributions). These instruments form part of the foundation of our capitalist system in which owners of capital rent out this capital to others, receiving a rent or yield in return. Under this system, workers (those who do not own capital) are a separate class, instead renting out their time to others in exchange for a salary.
Work tokens change this dynamic by creating “active” financial instruments which require both an investment of capital in order to own the instrument (thus unlocking its “capital” or “speculative” value) as well as an investment of “work” (defined as an additional, non-capital resource such as time, computing power, storage, transcoding, governance, etc) to unlock the “utility” or “cashflow” value and generate cashflows from the instrument. This means that work tokens, unlike traditional financial instruments, benefit workers (or more precisely, worker-capitalists) over capitalists as while capitalists benefit only from the capital value of the token, worker-capitalists benefit from both the speculative and utility (cashflow) value.
Crucially, the type of “work” these tokens enable their holders to provide corresponds to the resource or utility provisioned by the broader cryptonetwork in question. Effectively, work tokens incentivise the creation of digital, decentralized “co-operatives” with aligned incentives in the sense of networks that are entirely owned and financed by their “worker capitalist” owners. Indeed, work tokens enable these networks to finance themselves while simultaneously bootstrapping their supply and/or demand side ecosystems and communities, or to bootstrap their ecosystem while simultaneously financing themselves. This virtuous circle occurs because capitalist investors (providers of capital who purchase the cryptoasset) are incentivised to also provide work to the network in order to benefit from the cryptoasset’s cashflow value and maximise their ROI whereas workers (providers of non-capital resource to the network) are incentivised to purchase the cryptoasset and become capital providers in order to maximise the value of their work (and benefit from capital appreciation).
Thus, while traditional financial instruments mean companies must first raise money from capitalists in order to then hire workers to build out their supply side and marketers/salespeople to build out their demand side (thus increasing the value of the capitalist’s investment), cryptonetworks, through properly designed work tokens, enable all 3 to occur simultaneously, aligning incentives between all stakeholders and vastly accelerating the speed at which these networks propagate and grow.
What is a work token
A cryptonetwork can be said to be using a work token in the case that it pays out rewards to network participants who fulfil both the following conditions: a) hold or stake the native token and b) through holding/staking the native token, are entitled to provide/receive one or more types of potentially valuable work/utility (non-capital resource) to/from the network.
(a) is crucial since it disqualifies both BTC and ETH 1.0 from being work tokens, as in both cases rewards are paid out based on work but no token ownership is required. On the other hand, Steem, Augur and Synthetix are all examples of work tokens by this definition. For (b), it’s important to note that one token can entitle the user to provide/receive more than one type of utility/work as in Steem for instance tokens entitle users to both vote for block producers and curate/create quality content.
In the most general terms, a work token is thus comprised of two elements:
(1) a reward pool to be distributed to workers
(2) a consensus mechanism defining what constitutes valuable work (i.e. the conditions upon which the reward is to be paid out to workers), where the conditions necessarily entail both a capital contribution and a non-capital contribution
How the reward pool is financed is a question of fiscal policy and somewhat tangential to the design of the work token. Currently, most of the time the reward pool is paid for by inflation (i.e. subsidized by holders) but it could also be paid through transaction fees (i.e. subsidized by transactors/users), via a percentage of block reward (i.e. subsidized by miners) or in any number of different ways depending on how the network is designed. Importantly, this fiscal policy can also change over time (BTC block reward halvening shifts financing burden from holders to users), programmatically in response to preset parameters (see Hasu’s recent proposal) or based on the decisions of stakeholders through off-chain or on-chain governance.
Nevertheless, yearly reducing inflation represents an elegant way of initially funding the reward because:
a) higher rewards should be paid out to early workers, since they are taking more risk (investing work into an immature system) and reflecting the intuition that the marginal value of each additional unit of work decreases over time and
b) over time, value is reallocated from those doing nothing (passive holders) to those contributing work to the system. Those that hold tokens without providing work feel the full effect of inflation on their reduced relative token holdings compared to those holding tokens and providing work.
Consensus mechanism (type of work provided)
The specific type of work that particular work tokens enable their holders to provide comprises the resource provisioned by the cryptonetwork in question. Importantly, the work provided/resource provisioned can be extremely diverse and is limited only by the creativity of the cryptoeconomic designer and the ability for that resource to be provided digitally and trustlessly. Some examples include:
DASH — Staking the DASH token entitles its holders to provide work, in the form of transaction processing and governance, to the network. Holders are rewarded through inflation (subsidized by holders who don’t operate masternodes). The resource provisioned by the nework is secure, censorship-resistant transactions and governance of the network’s treasury.
Steem — Staking the STEEM token entitles its holders to provide work, in the form of accurate content creation and curation services, to the network. They are rewarded through a percentage of inflation (subsidized by passive holders). The resource provided by the network is censorship-resistant, high-quality content without the need for a centralized curator.
Synthetix — Staking the SYNX token entitles holders to emit a debt to the network, providing collateral and liquidity for synthetic assets created on the network. Rewards are paid out as inflation (subsidized by holders who don’t stake). The resources provisioned are liquidity and a global, trustless counterparty for the trading of derivatives.
Kleros — Staking the PNK token entitles holders to provide judgment on disputes. Rewards are paid out of inflation. The resource provisioned is trustless, automated and digital dispute resolution.
FOAM — Staking the FOAM token entitles holders to serve as “location anchors” and register points of interest on a map. Rewards are paid out of inflation. The resource provisioned is decentralized mapping services to compete with GPS (which possesses many known issues in addition to being a virtual Google monopoly) as well as a decentralized location history.
Importantly, in all the above cases, passive tokenholders who are not providing work to the network feel the full dilutive effect of inflation on their reduced relative token holdings compared to tokenholders who also provide work. This encourages investors to become workers, bootstrapping the resource provided by the network.
It is also worth noting that in all the above cases the work token is being used to bootstrap the supply side of a network, building up miners/voters in the case of DASH, content creators/curators on Steem, liquidity providers on Synthetix, arbitrators on Kleros and location providers on FOAM. However, a work token can also be used to bootstrap the demand side of a network and one interesting way to do this is through a discount or cashback token.
Demand-side bootstrapping — perpetual discount tokens
At the highest level, a discount token grants the holder/user discounts on transactions performed using another cryptocurrency or fiat. While at first glance the idea of a discount token may seem somewhat underwhelming as we relate it to a gift card or a coupon, there are several fundamental differences between them which make discount tokens both distinct from and far more interesting than gift cards/coupons. As the Sweetbridge whitepaper tells us:
“In brief, discount tokens are digital assets that give their holders a limited right to receive discounts on purchases of products or services from an organization — a company, a coop, or a blockchain network.
Unlike gift cards, discount tokens are not invalidated when used (“burned” in blockchain parlance), but remain in possession of the holders. The specific size of the discount that each token realizes for its owner is designed to grow in step with the overall utilization of the network.”
As Julien Genestoux tells us, the discount token initially appears limited as each token’s value seems to be capped by the actual value of the discounted service. However, this is only the case if the discount each token represents is fixed (e.g. a 20% retailer coupon or a 10% student discount card). In a better model, rather than applying the discount to an individual token, the discount can be applied to the entire token supply, linking the discount provided by each token (and consequently its value) to the size of the network itself.
To understand this, we can look at the example presented in the CoinFund blog on discount tokens. Let us imagine Amazon Prime issues discount tokens which holders can activate (stake) to offset 100% of subscription fees. Assume also that Alice owns 10% of the total token supply. If in year one total subscription fees are $1000, then Alice can offset $100 of personal spend based on her membership. Assuming Prime costs $100 a year, Alice would get it for free. If in year two total subscription fees are $10,000, Alice can now offset $1000 of costs. This means she can get her membership for free and is also incentivised to gift (or sell) 9 memberships’ worth to other users.
As we can see, the discount token helps to bootstrap the demand side of the network since in order to benefit from its utility or cashflow value, investors must pay fees (and therefore transact through the network). At the same time, users of the service are incentivised to become investors in order to minimise the fees paid. As we will see later, depending who is paying the fees, a discount token can serve to bootstrap either the supply or demand side of a network.
As Coinfund tells us, we can therefore think of discount tokens as “entitling holders to a perpetual discount on services (in our case transaction fees) but structured in such a way that the discount is mathematically equivalent to a revenue share/royalty, but only if one utilizes platform services”. From the perspective of the token issuer, this is a royalty model but rather than offering rights to a proportion of total cashflows, it represents rights to a proportion of total discount offered. As a result, the absolute amount of discount provided by each token (and consequently its value) grows linearly alongside network adoption, providing holders with an ever increasing discount which they can either use or sell to others.
Crucially, the proposed design is not like Binance’s BNB model in which fees are cheaper when paid in BNB. In that case, the fee is denominated in BNB whereas in our design the fees may be paid in a separate currency, preventing the velocity problem and giving the token an exogenous cashflow, thus enabling it to be valued using a discounted cashflow (more on this later).
The discount token model has several key benefits:
(1) Value capture and valuation framework — In contrast to Medium of Exchange tokens which suffer from the much covered velocity problem in which increased network adoption can actually lead to reduced token value, the value of a discount token will always grow alongside network adoption. In addition, since the discount provided by each token is mathematically equivalent to a cashflow, valuing the token doesn’t require complex and highly speculative equation of exchange based approaches as it can instead be valued using a simple and well-understood discounted cashflow methodology.
This is a unique property of discount tokens even amongst work tokens as most work tokens cannot be valued using a DCF because the cashflows they generate are denominated in terms of the tokens staked (i.e. they are non-exogenous), leading to circularity. Since a discount token provides a discount on fees paid in another currency/FIAT, its cashflows are exogenous can be modelled using a DCF.
(2) Doesn’t hamper UX — Unlike Medium of Exchange tokens and many kinds of work token, the discount token is not necessary to use the system and thus does not hamper UX or adoption. Users who simply want to use the network can do so without dealing with the complexities of purchasing and using tokens, although they always have the option to purchase the discount token and lower their fees.
(3) Benefits users over speculators — As previously mentioned, unlike traditional Medium of Exchange tokens whose value is the same to both users and speculators since its value is spent when utilised and thus selling is identical to spending it, this is not the case for a discount token. This is because a discount token possesses both resale value (similar to the utility token) and discount value, which can only be realised through discounts on actual services. As Alexander Bulkin tells us: “an investor holding discount tokens for passive appreciation is by definition underutilising them, only able to capture their resale value, but not the discount value”.
(4) Network effects — A discount token creates a clear network effect since users of the network are incentivised to become investors/stakeholders in order to maximise the value of their work, while investors are incentivised to become workers in order to maximise the cashflow value of their token.
This creates a real community around the project with its members incentivised not just to provide as much work as possible/purchase as many tokens as possible, but also to ensure the general growth and success of the network in order to maximise the capital value of their investment. This also serves as a competitive moat since the most active workers on the network (those paying the most fees) will also be those incentivised to own most tokens, increasing switching costs and making it more difficult for competitors to build up their own networks.
(5) Securities regulations — While I’m not a lawyer and this is definitely not legal advice, in general, discount tokens (like many work tokens) do not qualify as a security since they are an active rather than a passive instrument; thus failing condition 3 of the Howey test (“from the effort of others”) as the cashflow is contingent on the holder providing some work to the network.
Discount token case study
To illustrate the discount token in action we may look at our recent token economics client Cudo Ventures. Cudo is seeking to create the “AirBnB for computing”: a marketplace enabling owners of computing resources to rent out their spare capacity to those seeking computation. The economics of monetising spare capacity (specifically, the elimination of fixed/sunk costs means that cloud operators can produce at their variable cost) means that Cudo is able to offer significantly cheaper computation than incumbents such as AWS, Google Cloud and Azure.
Cudo came to us to design a token model that captured value and aligned incentive between its key stakeholders: suppliers of computing power, consumers of computing power and stakeholders. The discount token was a perfect fit for this use case, enabling workers (suppliers/consumers of computing power — anyone who pays fees to the network) to maximise their earnings and become stakeholders in the network while granting stakeholders exposure to the growth of the network and fees paid by the workers. Cudo’s planned token can therefore be thought of as a “Compute Discount Token” which can be staked in order to receive discounts on buying and selling computing power.
In general terms, we can think of Cudo’s discount token as possessing the following parameters:
[Cr] = Cudo’s fee revenues in a given period.
[R] = Discount Rate. A percentage of revenues that Cudo is offering as a maximum available discount. This can be a fixed value (we will assume 50%) or a variable determined by a formula.
[TS] = Token Supply. The number of tokens that participate in the distribution of discounts. This can be a sum total of all the tokens staked for discounts by users or the total circulating supply of tokens.
[DP] = Discount Pool = R*Cr. The portion of fee revenue contributed to discounts in a given period of time (every week, month, etc).
[DpT] = Discount per Token. The value of discount that workers can enjoy for each token they hold.
[MiF] = Fees paid by an individual miner “i”.
[TMi] = Token supply of an individual miner “i” staked for discounts in the period.
[MiD] = Discount enjoyed by an individual miner “i”.
We can thus derive the following:
- [DP]=R * Cr
- [MiD]=min(DpT*TMi ; MiF)
We can see illustrate this more clearly with an example. Let us assume £100M in Cudo Revenue, a discount rate of 50% and 50M in locked token supply. In this case, if a “worker” (defined as either a supplier or consumer of computing power) pays £100 of fees on a given month, each token held would allow her to discount £1 of fees paid and holding 100 tokens would allow her to completely offset the fees paid.
If, on the other hand, she held 110 tokens, the worker would still only qualify for £100 of discount since the [MiD] formula ensures the cashback received is capped at the total fees paid. In this case, she can either: (1) consume/supply additional computing power in order to take advantage of her unused discount tokens (2) use the tokens for some other purpose (Cudo has partnered with and integrated its token into various other platforms) (3) keep the extra tokens as an investment into Cudo (4) sell the tokens onto the market.
We can therefore see that unless every worker allocates the exact amount of tokens necessary to fully offset their fees, a portion of the discount pool will always remain unused. The project must then decide how to use this unused discount pool, although in general these should always be returned to users in some way, for instance by using it as an ecosystem development fund or placing it in treasury and allowing tokenholders to vote on how to deploy it.
Work tokens represent a new kind of financial instrument, creating digital cooperatives owned and operated by a community of stakeholders with aligned incentives and enabling projects to finance themselves while simultaneously bootstrapping their network. Within work tokens, discount tokens are a particularly interesting type of token which possess several crucial benefits, among them the fact that they can be used to bootstrap the demand side of a network, are easily implementable on any network that charges fees and can actually be valued using a DCF.
Thanks very much to Michal Bacia for helping to pioneer the discount token model with Sweetbridge and for working alongside AmaZix to help advance our understanding of it. Thank you also to my colleagues Kai Sedgwick, Luke Saunders from AmaZix and Anil Lulla and Medio DeMarco from Delphi Digital for reading and providing valuable feedback on this.