Harmony Token Economics

#1

Draft v0.1 - Kenny White

Introduction

Substantial research has been done regarding the crypto-economics for proof of stake (PoS). This research centers around creating a high-performant digital ledger that no one controls yet everyone trusts. Although important, this research is incomplete. It tells us nothing about the ideal token economics model for PoS public blockchains. Although decentralized, blockchains are not charities: writing the software requires a lot of resources to build and maintain. Look no further than the multi-million dollar R&D budgets of leading generation 3 blockchain projects. To be sustainable, these blockchains need a token model that both rewards the development team and their investors without compromising the integrity of the decentralized blockchain. Furthermore, this system should not create an unstable macroeconomy that disincentives validators and users from participating in the network.

What’s wrong with the current model

The current token economics model for PoS generally works as follows:

  • The development team sells their platform’s native tokens to investors and keeps some for themselves
  • Validators use these tokens for staking to the consensus protocol
  • Users pay validators transaction fees in the native token
  • Validators, in addition to user-paid fees, are incentivized by block rewards paid in the newly created native tokens
  • As more people use the platform, demand for the native token increases, thus increasing prices for the token and rewarding investors and the development team

The advantage of this approach is that it allows the project to raise capital, rewards investors, gives users a native medium of exchange and provides economic incentives to validators. Unfortunately, this model has some serious drawbacks, despite the aforementioned attributes.

The first drawback is that this model creates price volatility in the platform’s native token. This volatility is a direct and inevitable consequence of these currencies’ rigid supply schedule. If the supply of tokens does respond dynamically to demand, price will be volatile. Users will not want to hold and pay fees in a volatile asset, which is a major UX problem. Additionally, validators do not want to be paid in a volatile asset. Both parties will be better off if the platform’s medium of exchange is stable.

The second drawback is less obvious and more subtle: the native token is serving double-duty. It needs to function simultaneously as a medium of exchange to pay fees and it needs to function as a validator/governance token. This is bad economic design. In software development, you don’t want a single component in the architecture to handle two completely different tasks. Using a single token as both stake in a consensus protocol and as a medium of exchange violates this principle.

We should not have one token trying to fulfill two, non-complimentary roles. A single token model would be like Apple forcing customers to pay for their iPhone using Apple stock. The more people buy iPhones, the more demand there is to buy Apple stock, Apple stock appreciates and investors make money. Obviously, this model is non-sensical. It is far better for Apple to sell iPhones for money, and pay investors dividends out of their company profits.

The third problem is rooted primarily in pragmatism. Investors and the development team are taking a big risk. They need to be rewarded commensurately. Typically, investors and the development team want to keep as much equity in a company for themselves as they can. No one blames Mark Zuckerburg or Larry Page for retaining a controlling interest in their companies. We all would, if we could.

The project’s investors and development team should be allowed to keep as much ownership of the company as market conditions will permit. Unfortunately, this objective is in direct conflict with the need for decentralization, namely at the validator and governance level. If the project’s team and investors kept 51% of all tokens, then the ledger will be effectively centralized and thus rendered irrelevant.

To summarize the three problems of the current PoS token models:

  • it causes volatility for users and validators
  • it conflates multiple economic roles for the same token
  • it forces founders and investors to either dilute their economic reward or centralize control of the platform

New proposal

An ideal token economic system will have a stable medium of exchange for validators and users, use separate tokens for separate roles, and grant founders/investors the ability to profit from the platform without centralizing control. With these objectives in mind, here is what I propose:

  • Use two separate tokens - one for validators to stake, one for users to pay fees
  • Fees are paid in stablecoins, block rewards are earned in more validator tokens
  • Validator tokens have different tiers - all have equal weight in terms of stake, but some earn more fees

Allowing users to pay fees in stablecoins solves the volatility problem. Now users do not need to hold a separate volatile asset and neither do validators.

Separating validator tokens from stablecoins also rationalizes the economic design of the system. In every for-profit industry (except crypto), customers pay service providers with money. This money covers the costs of operation, and whatever is leftover goes to the owners/investors of the business. With this new token model, being a validator is just like running a business. Or perhaps more accurately, like driving for Uber. Being a user of the platform is just like being a customer of a traditional business.

Crucially, validator tokens become far easier to value. There is no clear way to value single-token blockchains. In theory, the value of a token like Ethereum can be inferred from the Equation of Exchange (MV=PQ). The problem with this approach is that velocity (V), is poorly understood and extremely difficult to predict. Moreover, the Equation of Exchange was never meant to be applied in this context in the first place, so any valuation framework built upon it is destined to be flawed.

In contrast, validator tokens in the proposed two-token blockchain are quite easy to value. Validator tokens’ value is derived from the income they can generate from fees. The more people use the blockchain, the more fees people pay and the more valuable validator tokens become. To estimate their value, an analyst can apply the standard discounted cashflow analysis to arrive at the tokens’ net present value. This is the same approach to valuing any asset with a stream of future income. First, make a forecast of future fee growth each year and then apply the appropriate discount rate. For example, if each validator token is forecast to earn $1 in fees every year, and the discount rate is 20%, then each token is valued at $5. (Value = Income/discount rate).

Being easy to value is useful in and of itself, but more importantly it is indicative of sensible economic design. If an asset or financial product is difficult to value, that means it is overly complex. PoS blockchains are complicated enough already. The complexity and instability deriving from poor economic design is best avoided.

Different tiers of validator tokens allow investors and the development team to reap a disproportionate amount of the platform’s value without centralizing the network. Each validator token, regardless of its tier, carries the same weight in the consensus protocol. However, some of them will earn a higher share of fees. To illustrate this concept, consider the following two-tier example:

Tier 1 tokens: collect 5x the fees

Tier 2 tokens: collect 1/5x the fees

There are 10 validator tokens in total, one Tier 1 and nine Tier 2. In the consensus protocol, the Tier 1 token will have 1 vote in aggregate, the Tier 2 tokens will have 9 votes in aggregate. The Tier 1 token will earn 5/14th of the fees. The Tier 2 tokens will earn 9/14th of the fees in aggregate.

In this example, owners of the Tier 1 token will only control 10% of the network but will collect about 36% of the fees. If the development team would like to capture a higher percentage of the value, then they can increase the fee differential as high as they would like. Additionally, there can be more than two tiers. Indeed, there can be as many tiers as needed. Such a model has precedent in the non-crypto world. Shares in companies have different tiers as well. In many companies, there are shares that have either more votes or collect a higher percentage of dividends. Functionally, this is equivalent to having different tiers of validator tokens.

Concluding thoughts

Current token models for PoS blockchains suffer from poor economic design. They force users and validators to transact with an unstable asset that is very difficult to value. Furthermore, they force investors and developers to make a trade off between their economic reward and the integrity of the decentralized network. A two token model where users pay fees in stablecoins and validators stake a specialized token rationalizes the economic model and solves the stability problem. Additionally, creating different tiers of validator tokens, akin to different types of shares in a company, allow the founding team to capture a disproportionate amount of the network’s value without compromising the network.

There are still some unanswered questions regarding this new token model. First, the regulatory implications are unclear. The focus of this paper is economics, not speculation regarding legal ramifications. Nonetheless, it is worth noting a few points. On the one hand, validator tokens in the two-token model seem highly likely to be classified as securities; whereas, single-token systems may have the option of one day becoming a “utility token”. On the other hand, the medium of exchange, a stablecoin, is probably not a security. Thus, users and validators are free to transact without violating security regulations.

With a single-token model, it is unclear when the network token transitions from being a security to a utility token. Until that transition is complete, it may be highly problematic for users to buy and transact with these tokens. Many users will not be accredited investors (a prerequisite to buying securities) and selling/trading these tokens may have tax implications. This analysis is by no means intended as legal counsel.

A second unanswered question is the technical implementation of a tiered validator token model. The economics are quite clear and, as mentioned previously, follow a precedent set in traditional stock markets. However, at a software level, designing such a system may not be so straightforward. If the consensus protocol in practice functions as a winner-take-all lottery weighted by stake, then the fees earned per token will necessarily be proportional to the token’s weight in the consensus mechanism. Thus, having different tiers of validator tokens without centralizing control will require a workaround in order to be feasible. If the consensus protocol functions more like proportional voting where the winning coalition splits the transaction fees/block rewards, then this problem can be avoided. Nonetheless, more research would be required to see how different tiers of validator tokens might affect the game-theoretic incentives of network participants. A simple model in theory may in fact be prohibitively complicated in practice.

The third unanswered question revolves around stablecoins. Implementing a stable cryptocurrency is not easy. The multitude of failed stablecoin projects are a testament to just how difficult it can be. Basecoin raised over $100 million and yet failed to launch. The only thing harder than building a stablecoin is perhaps building a scalable blockchain. Trying to create a new stablecoin in tandem with building a gen 3 blockchain adds a lot of execution risk. Fortunately, there are already several viable stablecoins already in circulation, namely: TrueUSD, USD Coin and Paxos. Bringing one of these stablecoins onto the platform seems like a viable alternative.

However, this leads to yet another conundrum. If only one stablecoin is picked as the default currency, then the platform’s monetary system is now in the hands of an unaccountable third party. Moreover, the aforementioned stablecoins work by holding fiat dollars in reserve, 1:1, to backup the value of the cryptocurrency. With just one stablecoin provider, the otherwise decentralized platform will depend on a single, centralized bank account.

If, instead, multiple stablecoins become the default currency of the platform, there is yet another problem. Validators and especially users will be unwilling to deal with many different tokens for payment. What if the validator only accepts Stablecoin A but the user only has Stablecoin B to pay the transaction fee? In economics, it is rare to see multiple currencies accepted as tender to coexist side-by-side. It is highly inefficient and generally unstable.

There must be a solution to the stablecoin problem, but delving into the complexity of monetary economics is beyond the scope of this research note.

3 Likes

Kenny White - Cowri chief economist
#2

Excellent, excellent ideas, Kenny.

After reading this, I am convinced that stable coins could resolve many of the issues surrounding the fee systems in current blockchains. I particularly like that you point out that this is not relevant just because it makes fees stable and predictable, but also that it clears up regulatory and tax issues associated with paying in a security/commodity (or whatever the native token happens to be).

The second extremely interesting aspect of this proposal is that such a system would make valuation of staking tokens as straightforward as valuing shares of a company with a DCF model. That is as game changing as the advantages to the UX for miners and users.

The open questions in my mind are:

  1. Does the stable coin become a central point of failure for the entire network?
  • Imagine that the stable coin issuer’s bank account is frozen or seized. If the stable coin is broken, the system would come to a screeching halt. You could allow multiple issuers to have stable coins on the Harmony platform to decrease the risk of a complete shutdown, but still this possibility makes me uneasy.

  • Maybe you could have a back up fee system that can be paid in the native token as a fall back to this kind of situation. So if the stable coin mechanisms break, at least the network can continue functioning.

  1. How does one implement this?
  • It seems you would need some concept of this special stable coin as part of the base protocol to build in natively to the transactions and fee systems. How do you handle this extra token, especially since its supply is controlled by an entity outside of the protocol.

  • I think there’s a simple solution here, but some thought would have to be given to designing this. Especially if there are multiple stable coin issuers.

  1. If there are multiple stable coins on the platform, how does the community decide which ones are valid to pay fees in and which ones are not?
  • I imagine there would need to be some kind of governance mechanism to approve or disapprove of stable coins being added or removed from the list of accepted coins. What would this governance process look like? Is there a more elegant way of doing this?
  1. If the staking tokens become a revenue generating asset, are they necessarily a security?
  • In this model, is it a forgone conclusion that the native token is forever in the security bucket? I don’t think this is such a bad outcome, but I am wondering if it really is like a share of a company or not.

Overall, I’m very excited by this idea. It could be a big differentiator for Harmony. The benefits to miners and users are enormous. Let’s explore this further and solidify the concept!

0 Likes

#3

@Zi_timeless and @William, I would be curious to know your thoughts on this.

How about you @satoshi and @nb1?

0 Likes

#4

Good questions, Nick.

Here are my thoughts:

  1. I think blockchain monetary systems will always be a potential point of failure, just like consensus, networking or governance. If you only have one currency, then this currency can be a central point of failure, regardless of whether the currency is a stablecoin or not. How secure would Bitcoin be if the exchange rate collapsed and miner’s turned off their rigs? The nice thing about stablecoins is that by their nature they are interchangeable with each other so you can in theory have many many different stablecoins used in tandem. In a way the monetary system can become more decentralized with stablecoins.

  2. Implementation is always tricky. I understand the economics pretty well but I’m sure there are a lot of technical details to be worked out. I agree that having one designated stablecoin for the platform would create a lot of risk. Why have a decentralized ledger if the currency is controlled by a centralized third party? This risk can be mitigated if you have a system that allows an arbitrary number of stablecoins to be used on the platform.

  3. A centralized list of accepted stablecoins is feasible but comes with a few draw backs. First, not everyone will agree on which stablecoins to use. Perhaps Alice thinks crypto-collateralized stablecoins like Dai are too risky. Perhaps Bob thinks that fiat-backed stablecoins are overly centralized. It will be difficult to get everyone to agree and even more difficult to force people to accept a stablecoin they don’t like. I have an idea for a decentralized alternative but it’s tricky to explain.

  4. Will these modified staking tokens be a security forever? I’m not sure. That’s a question for the lawyers. I can say that from an economics perspective, this model is very similar to owning stock in a toll road or electric grid. But economic intuition isn’t the same thing as US securities regulation.

1 Like

#5

Good points @kwhite! Implementation is our specialty, but what we implement depends on how we design the other mechanisms such as how many stable coins we can support and the governance of the list of accepted stable coins. It would be interesting to hear the ideas you have in mind. Perhaps you can open a new thread on that topic?

0 Likes

#6

Great suggestions! @nickw have we finalised the two token model already? Is there another thread to follow?

2 Likes

#7

Hey Anir!

Glad to see you here! For launch we will use a typical utility token model, in which our token is both used for staking as well as for fees paid on the network. This is simpler and a well proven model.

Thereafter we will consider adopting a different token model that may look something like what @kwhite suggests in this post. It makes a lot of sense.

Nick

0 Likes