(Note this model has been updated. Details of the new model is in https://medium.com/harmony-one/harmonys-new-tokenomics-bcdac0db60d7)
Token economy of public blockchains is as important as the blockchain protocol itself. A good token economic model should support the security and longterm sustainability of the network. At the same time, it should also benefit all the participants of the network, including the validators, token holders and the users, thus growing the network over time.
We publish here Harmony’s token economic model, including issuance rates, fee market, validator yield and target staking ratio.
Harmony’s economic model features dynamic issuance around a target range of staking percentage to offer validators a sustainable long-term yield while reducing the cost paid for securing the network. Below, we detail the choices around the parameters which underpin our economic model.
Staking Target Percentage
We believe a target stake ratio in the range of 35% is optimal for the following reasons:
- An attacker would have to purchase a controlling stake on top of the amount currently staked which would be expensive given that purchasing these tokens would in turn increase token price. The cost of an attack will be high enough to deter any attempts, especially with slashing in place.
- A lower target offers more ONE tokens to be used for DeFi and Web3 applications as collateral or other uses within these projects.
- Most importantly, Harmony will not overpay for security as a high staking target percentage would require a high token issuance rate to keep a reasonable return for validators.
Quantitatively speaking, this target is more of a range than it is an exact value. It’s impossible to know exactly what yield the market, as a whole, requires as compensation for the financial and technical risks you’re exposed to as a validator. The monetary policy mechanism is designed to increase issuance as the percent of supply staked decreases, and decrease issuance when supply staked increases.
A granular analysis is provided at our public spreadsheet harmony.one/charts.
We visualize this dynamic below based on the circulating supply starting in 2022, as that’s when a majority of the total supply gets unlocked. A monthly breakdown of the supply unlock schedule is viewable at harmony.one/unlock.
Comparison: Other projects like Cosmos and Polkadot have chosen staking percent targets of 66% and 50%, respectively. We believe the incremental value of targeting a higher stake percentage to bolster security does not justify the added cost in issuance. Cosmos did not explicitly explain why they chose 66%, but it may be related to the fact that BFT systems need 66% to reach consensus. Its issuance rate will initially be 7%, but it will adjust based on how long the stake rate is above or below 66%. Issuance will gradually increase towards 20% if the stake rate stays below the target, and gradually decrease towards 7% if the stake stays above the target. The stake rate adjusts up to 13% per year. Polkadot chose its 50% target based on a breakdown of a 3:2:1 distribution for staking, parachain deposits and liquidity, respectively.
The networks of Dash, Decred, EOS, and Tezos have staking participation rates of ~38%, 51%, 57%, and 72%, respectively. These participation rates vary widely and are generally due to differing issuance rates.
Network Issuance and Inflation Trends
Our dynamic model’s 35% target staking participation rate results in network issuance of 3.3% in 2022 which trends down to 1.7% by 2050. Assuming the network is at this target rate, yield for validators will range from 10.5% to 4.1% over these same periods, respectively. We believe this issuance is low enough to ensure security on the network and is not an extremely high issuance rate. We also believe the yields for validators are attractive given this issuance schedule and target staking percentage, as discussed in the section Gross and Net Yield To Validators. Below we present how the issuance rates vary based on stake, and how they all naturally decline over time.
This structure is designed to gradually approach zero issuance as the growth and adoption of the network leads to increased gas fees that will gradually reduce the dependence on issuance for sustainability. At a supply staked level of 35%, the starting yield is 10.6% in 2022. This is based on a per shard block reward of 18 ONE tokens. For every 1% increase (or decrease) in stake, the block reward decreases (or increases) by .4 ONE tokens. Although issuance shifts linearly, the fact that it shifts inversely to supply staked allows the yield to change exponentially. This creates strong downside defense from having too low of a stake as yields become very attractive at lower levels.
Comparison: We believe an issuance rate in the 1% to 3% range will allow Harmony to be comfortably sustainable in the long run using our target stake rate. This will enable the protocol to be secure while boasting a considerably lower issuance rate range relative to many other known networks. Our model is below Bitcoin’s current issuance rate of 3.71%, below Ethereum’s current issuance rate at ~4.5%, and well below protocols like zCash whose 42% annual issuance rate could potentially undermine the value of its token.
Polkadot and Cosmos will each have considerably higher issuance rates as seen in the charts above. Eth 2.0 will have a lower issuance rate than Harmony, but we believe their issuance rate is still lower than desirable, evidenced by the profitability issues validators will likely experience early on (or they’ll have an insignificant portion of the supply staked).
Yield Adjustment Based On Staking Percentage
For every 1% that the staking participation rate is in the wrong direction of the target (35%), block rewards will adjust by 0.4 ONE tokens per shard. On the lower end, this creates an environment where even a small dip below the target stake rate would significantly increase yields and encourage validators to step in and capture these returns. The benefit of a lower target stake rate is that providing these increased yields does not come at the cost of significant issuance.
On the high end, this slope allows the protocol to naturally approach and reach 0% issuance once 80% of the network is staked. Ideally this would naturally occur as a thriving protocol creates a thriving fee market. This in turn encourages more of the network to be staked, which then improves security.
Comparison: Ethereum 2.0’s issuance is purely based on the quantity of Eth staked, rather than percent of supply. Issuance increases as stake increases, but at a slower rate, so yield goes down.
Polkadot’s monetary policy is structured to maximize issuance at their target stake rate of 50%, and then have issuance drop off when stake deviates in either direction. We don’t believe the slight benefit of not having issuance increase when staked supply goes down offsets the issues that come with it. For one, it creates a less consistent rate of change in yield when stake deviates from its target. Also, the precipitous decline in yield as soon as the stake goes above the target rate creates unnecessary yield instability.
Cosmos’ mechanism isn’t based on the degree of variability from 66%, but rather the amount of time spent below or above the range. This annual change is limited to 13% points of issuance per year, with a cap at 20% and floor at 7%. Although this structure is directionally similar to Harmony’s model (more stake less issuance, less stake more issuance), the adjustment mechanics vary considerably. Since the issuance rate adjusts purely based on whether total stake is above or below 66%, it can vary significantly even if stake is unchanged for an extended period at 65%. Incorporating this type of instability didn’t seem appropriate.
Harmony’s gas fee market will be a standard first-price auction similar to that of Ethereum. However, we are also carefully considering the benefits of implementing a fee model as described in EIP 1559. Such a fee model can make fees more predictable for users and prevent overpaying, as well as provide a token burning mechanism to offset issuance. In the future, Harmony will implement a storage fee model. The details of this model will be published when it’s ready.
Comparison: A first-price auction is the standard fee market design across many of the most successful protocols including Bitcoin and Ethereum. The advantages of this approach is its simplicity and proven track record.
Linear Slashing for Double Signing
A minimum slashing rate of 2% is exacted for double signing. This increases linearly with the number of double signers of the same block (e.g. 66% slashed if 2/3 validators double sign). Of the amount slashed, half is burned and half goes to the reporter. The latter is used to incentivize users to police malicious behavior on the network.
Comparison: We believe having slashing mechanisms for double signing behavior is critical to curtail malicious behavior on the network. Other networks such as Cosmos have slashing mechanisms, and ETH 2.0 plans to as well. For Cosmos, a 5% penalty is given, the unbonded validator can’t rebond for 21 days and the validator is jailed forever. On Tezos, the security deposit of 512 XTZ (per block) for baking or 64 XTZ for endorsements is slashed. On ETH 2.0, a portion of the validators deposit equal to three times the portion of validators that were penalized around the same time as them is used.
Voting Power Leak for Going Offline
Harmony features penalties for validators who go offline as this impacts the security of the network. If blocks are missed for three consecutive hours then ¼ of the validator’s voting power is leaked, continuing until all voting power is leaked after 12 hours. As such, the validator loses voting power over time if they are offline. When a validator loses all voting power after 12 hours of being offline, they will will lose at minimum the rewards they could have earned in those 12 hours.
Comparison: Other protocols feature similar penalties such as EOS, where if blocks are not produced for 24 hours, the block producer is removed from consensus and Tezos where if a baker misses the last 5 cycles (15 days) the baker is jailed for 1 cycle.
Other protocols feature slashing if a validator goes offline such as in Cosmos, where if 95% of blocks are missed in the last 10k blocks (16 hours) there is a 0.01% penalty and validator jailed for ten minutes. ETH 2.0 also features partial slashing where everyone gets penalized slightly if a centralized provider other validators are using goes down, such as AWS. ETH 2.0 also aims to enact slashing if a validator is offline for 18 days, the amount of the block reward will be slashed if a block is finalizing, if a block is not finalizing up to 60.8% of the stake is slashed.
Stake Unlock Periods
Harmony is set to have a stake unlock period of 7 days. After sending a withdrawal request, validators and delegators need to wait for 7 days before their stake becomes liquid again. We believe this is a good timeframe given it allows enough time for stakeholders to realize and act upon malicious behavior on the network before someone can withdraw their stake, but not so long as to curtails participation from entities that may wish to withdraw their funds at some point in the future. The unlock period is used primarily to combat a nothing at stake attack.
Comparison: Other networks all feature withdrawal periods with EOS, Tezos, Cosmos and Polkadot currently at 3 days, 15 days, 21 days, and 12 weeks respectively. ETH 2.0 does not have a withdrawal period set in stone yet to our knowledge but will likely feature one.
Gross and Net Yield To Validators
Gross yields will vary based on stake. Since issuance is adjusted (according to stake %) from a fixed ONE token issuance rather than maintained as a fixed rate, the issuance rate will gradually decline.
We believe most validators on Harmony will use a cloud setup as it has the lowest barriers to entry and is faster than provisioning one’s own hardware. Using AWS as an example given it’s the most pervasive cloud provider, between compute, storage and network costs, the average monthly expense will be $50 to run a node per our calculations using AWS’ t3.small instances and Harmony’s own estimates on network usage. Given this cost, current ONE prices, and our model, we provide a range of net cloud yields below.
These yields are based on a stake rate of 35% which, at the start of 2022, would imply a median stake per validator of 2.7m ONE tokens. You can see how these yields gradually decline as time passes. For context, a 35% stake in 2032 would imply 4m ONE tokens per validator.
Net yield will increase, and gradually approach gross yield, as the price of ONE tokens goes up. However, profitability is not overly conditional on the price of ONE tokens being high. Below we provide a table showing net yields at various token prices for cloud based stakers, assuming there are 3m tokens staked.
Comparison: Starting with ETH 2.0 as a comparable, we provide the expected gross yield for validators based on their newly adjusted proposal. The percent of total supply staked is based on an expected total supply of 11.5B ONE in the beginning of 2022. The yield is much lower for ETH validators and more dependent on network fees.
Cosmos’ yield will vary according to the amount of time the total stake has spent above or below their 66% target. Cosmos will have slightly higher yields than Harmony, but it comes at the cost of significantly higher issuance. Polkadot will also have a higher validator yield than Harmony, at the expense of issuance.
This report is done in collaboration with Delphi Digital, a research and consulting firm focused on the digital asset market.
Delphi Digital receives no compensation from the companies, entities, or protocols they cover for their subscription-based research products. This report was prepared for members of the Harmony team as a paid consultation and does not represent an endorsement nor a recommendation by Delphi Digital and is not to be construed as investment advice of any kind; this report is strictly informational. Do not trade or invest in any tokens, companies or entities based solely upon this information. Any investment involves substantial risks, including, but not limited to, pricing volatility, inadequate liquidity, and the potential complete loss of principal. Investors should conduct independent due diligence, with assistance from professional financial, legal and tax experts, on topics discussed in this document and develop a stand-alone judgment of the relevant markets prior to making any investment decision.
The information contained in this document may include, or incorporate by reference, forward-looking statements, which would include any statements that are not statements of historical fact. No representations or warranties are made as to the accuracy of such forward-looking statements. Any projections, forecasts and estimates contained in this document are necessarily speculative in nature and are based upon certain assumptions. These forward-looking statements may turn out to be wrong and can be affected by inaccurate assumptions or by known or unknown risks, uncertainties and other factors, most of which are beyond control. It can be expected that some, if not all, such forward-looking assumptions will not materialize or will vary significantly from actual results.