Restructure fees and align incentives

[banteg]

Authors

@banteg @fubuloubu @milkyklim @lehnberg @tracheopteryx

Summary

With Vaults v2 around the corner, it is a good time to analyze Yearn’s protocol-wide incentives and ensure they are aligned to produce the best possible outcome. This proposal reforms Yearn’s fee structure to keep it roughly at the same level, but charged differently, and distributed in a way meant to keep YFI stakers, strategists, and contributors better incentivized over the long term.

Abstract

The Vault withdrawal fee is removed, and replaced by a 2% management fee. The performance fee is raised to 20%. This results in roughly the same total amount of fees charged.

Protocol earnings are used to market buy YFI through the YFI vault, which now becomes the centerpiece treasury vault where all earnings converge.

Note: The change only affects Vaults v2.

Motivation

Yearn’s competitive edge lies in innovation. We want to attract the brightest minds to collaborate on strategies for us, and to do that we need to provide the best incentives to our contributors. Building a profitable and safe strategy is difficult, the strategist compensation needs to make it worthwhile to do so.

The withdrawal fee that was previously in place misaligns incentives as you are only charged for the service when you no longer want to use it. More money leaving the project would lead to more fees, when it should be the opposite.

Replacing it with a management fee on funds deposited in the system optimizes instead for retaining funds for as long as possible. The vault is providing a valuable service that people are willing to pay for, continuously. This also benefits composability with other protocols and opens Vaults up for more experimentation and integration as a “yield lego”.

Specification

Overview

The new model aligns incentives throughout the protocol and adds buying pressure on YFI.

Rationale

The challenge with withdrawal fees

Tl;dr: a withdrawal fee is not aligned with actual protocol usage.

If Yearn does everything right, and there’s an influx of users joining vaults, depositing funds, and earning great yields, there would be no fees as users would leave their money in the system for as long as possible without withdrawing. At that point, perverse incentives would build to FUD the system in order to trigger withdrawals and earn fees. [1]

Comparing fee schedules

The old fee schedule is used as a benchmark for the new proposal – the new fee structure should not earn less than the model it is replacing. At the same time, it should align incentives better. We want Yearn to attact the best talent and pay developers the best rates.

To help the comparison, we’ve used the oldest and the largest Vault (yUSD), which has been live since 2020-07-30 and backtested the models against it:

Old model
5% performance fee: 466,822 (17% of total)
0.5% withdrawal fee: 2,243,078 (83% of total)
Total fees: 2,709,901

Suggested model
20% performance fee: 1,867,288 (73% of total)
2% management fee: 699,237 (27% of total)
Total fees: 2,566,525 (95% of target)

Sankey diagrams

Old model

Suggested model

Example Vault

10,000,000 AUM, 10% APY, after 1 year

Technical Specification

Previous fee model

• 5% performance fee
  • 90% (4.5% of total) treasury vault
  • 10% (0.5% of total) strategist reward
• 0 to 0.5% withdrawal fee
  • 0% when pulling free funds from the Vault
  • 0.5% when pulling funds from the Strategy

Proposed new fee model

• 0% withdrawal fee
• 2% annualized management fee to the treasury (accrued per block, collected on each harvest)
• 20% performance fee
  • Half (10%) to the treasury
  • Half (10%) to the strategist reward

The treasury would now be composed of yYFI in the YFI Vault.

Fees earned (the management fee and half of the performance fee) are split by the YFI Vault between stakers and the contributor reward pool:

• 80% to YFI stakers
• 20% contributor reward pool

Changes to operations budget

Currently yearns operations budget is 181,000 yUSD per month for 6 months as defined in YIP-41. Should this proposal pass the operations budget (listed as “contributors” in the diagrams above) would be replaced by YFI according to the above: 0.4% via management fee (20% of 2%) + 2% via performance fee (20% of 10%).

From the simulations this should be adequate though may need tuning in the future. Although details of treasury management are outside the scope of this proposal, we have some ideas for a mechanism to provide stablecoins or yUSD to pay grants which is preferable to payment in YFI, though we will have the ability to provide vesting YFI as compensation additionally. Provisionally, we would target treasury management to 20-25% YFI and the remainder in yUSD yield-generating stablecoin

Discussion

[Proposal] Increase Strategist Rewards

Summary: Strategist rewards are way too low at the moment and it is unlikely that new strategists would even cover the costs of testing. Abstract: Strategists currently receive 0.5% of the profit from their strategy. Imagine a strategy that averaged $1m AUM at 10% APR and lasted for 6 months. The strategist would only earn $250 ($1m * 10% * 0.5% / 2). Which is less than what they would have spent. This may not have been a problem for V1 as AUM and APR were very high per strategy. But going i…

References

1. Dune vault stats

[fubuloubu]

@banteg

Example Vault

10,000,000 AUM, 10% APY, after 1 year

This is a great chart. I feel the need to add a few caveats though:

1. Obviously, this is just a sample. This should be obvious because there is no ticker or units to the numbers.
2. 10% raw APY for all strategy returns is a conservative estimate. We’ve seen higher in the past, and that performance depends on how well the community manages for on-chain conditions, security events, etc.
3. The 10% that strategists earn on overall yield generated is split N ways, where N is the number of Strategies that the Vault has over the year (5-10 or more is to be expected). We should also expect to see Strategists collaborating on more than one strategy over the course of a year (probably 10-20).
4. This is an example of just one Vault. The overall yearn ecosystem could probably manage 10-20 or more of these successfully, all at varying AUM and APY.
5. Stakers and contributor rewards are cumulative over M Vaults, so that should be used a milestone for judging this chart.

[YFI_MOON]

LGTM

[YFI_MOON]

What is this type of chart called? I cant read it

[Ceazor]

If I summarize this as follows, have I errored in any way.

The fees will remains nearly the same as present for users.

The fees will shift from being dominantly sent to the Treasury to be sent dominantly to the devs contributing.

The remaining Treasury fee allocation will instead be sent to the yvYFI. With a much smaller portion set aside for the Treasury

[gilfroy]

This sounds much more sustainable than the current structure.

Wondering why there is still a considerable amount of revenue going to stakers?

[cnumber]

great job guys…this should build a solid foundation to attract talent in the long term…

[mamatrohimat]

I think, the withdrawal fee of 0.5% should be maintained. Let have Management fee 2%, performance fee 20% and withdrawal fee 0.5%

[Dark]

Love the proposal good work! So I take the above to mean that now stakers will generate their yield in yfi and not ycrv, if they are staking in gov? I personally would like to see the number more at 75% stakers 25% contributors.

[Rorschach]

This analysis treats withdrawals as a constant value which is not the case. Withdrawals occurred because vault strategies were were not achieving gains or inactive. This entire premise is based on the notion that a variable input is actually a constant and the users pay a fee for this. What happens if a vault is inactive? Do consumers still pay 2%? That seems unfair.

A 20% performance fee will increase strategist rewards sure, but decrease value in the vault. HIGH management fees in the finance industry are 2%.

This should be implemented, but it should be done so on a vault specific basis based on defined risk parameters of the vault. Is the goal of the vault high risk / high reward strategies? Then yes, this makes sense.

If the goal of the users in the vault just want to make a little extra on some existing capital at low risk, this becomes an active disincentive for people to use the vault and withdrawal fees make more sense.

[captainobvious]

The general proposal is sensible and the direction of aligning incentives is correct in my opinion.
What I would suggest is a simplification of the proportion of output to each party as it makes it easier to budget, plan for and rationalise to external/new audience.

The current output allocation I see is:
36.38% Strategists
50.90% Stakers
12.72% Contributors

Suggested simplification of output proportion and rationale
I suggest a reward split of 60/30/10 for Stakers/Strategists/Contributors with the following rationale (noting all three groups are needed)

Stakers - 60% are the ultimate bearers of risk, opportunity cost and price risk. The dividend rewards are what gives YFI its value and are an incentive to ensure sound governance decisions are made and risk is assessed correctly. Without their vote and mindpower, nothing happens and the buck stops with them. Given price is the ultimate marketing tool, a lions share of 60% rewards ultimately makes YFI the attractive DAO to work for and contribute to. Importantly, both strategists and contributors will likely become stakers (if they don’t, we have misaligned incentives in my view, it should be desirable specially for these insiders who know the value proposition). Applying the lions share of 60% to stakers better aligns long term incentives and encourages strategists and contributors to add value for stakers (which includes themselves) and avoids’double counting and over allocation to individuals.

Strategists - 30% These experts create the product and should be rewarded for their time, skill and expertise.Without them, there is no product and they must expend funds deploying test code etc and upskilling in a competitive market. The proposed reward of 36.38% is slightly high in my opinion and the 6.38% would be better positioned as buy support for YFI (stakers) which is more than likely going to be where strategists lock some portion of their their income anyway. Reducing to 30% accounts for this phenomena and avoid too high an allocation of governance rights to individuals (despite their skill and important contribution).

Contributors - 10% This pool I assume provides funds to incentivise less technical, non-dev and community marketing, contribution. This is necessary and the current fee structure is 12.72%so not far different from the proposed 10%. likewise, the 2.72% reduction and allocation to stakers accounts for contributors who will also likely become YFI stakers.

Open to comments and discussions of this simplifying suggestion.

[banteg]

Quite disagree with your definitions of the ecosystem actors.

• Stakers bear zero risk in the current protocol, they just eat free lunch. Compare that to MakerDAO where MKR holders actually act as a backstop and risk dilution. There is no equivalent thing in YFI.
• Strategists. I don’t see how allocating votes to the most capable big brain contributors can be seen as a bad thing.
• Contributors pool is Yearn’s operations budget meant for maintaining the protocol and personnel, including devs.

[ramaruro]

I don’t agree with what you suggest (keep 0.5% withdrawal fee). The 2/20% better represents the incentives for performance and strategists, while not expecting people to leave (withdrawal fee).

[athleticyogi]

I’m for this - how do I sign up

[captainobvious]

Stakers certainly bear risk as their votes are what passes proposals such as this one to make the whole system work. It makes YFI a desirable asset to hold and thus a desireable protocol to work for. All folks who understand Yearn fall into this bucket so it is essentially a catch all for supporters.

Strategists earning 30% remains a high allocation and when combined with the incentive to make staking profitable and desirable, is additional motivation to stake their income (or some portion of it). The idea is one should flow into the other. It is excellent to reward these people, I simply suggest we balance the value across income + staking rewards

Contributors, perhaps this needs to be better defined in the proposal to make it clear what distinguishes devs from strategists (are they often one and the same?)

[gilfroy]

Right. Stakers bear 0 risk. So why give them so much money?

I’m interested in seeing products. The more rewards you give stakers, the more you attract people not really interested in the protocol.

Was there a clear point of diminishing returns when deciding what % to allocated to contributors and strategists?

[captainobvious]

If price falls, who wears that risk? Bad decisions affect price and protocol longevity. If stakers are indeed believed to bear no risk, why allocate to them at all? This is the incentive for people to buy and lock up capital in YFI (and not a competitor) and thus more capital invested makes for more attentive and less apathetic governance decisions.

[ramaruro]

Stakers bear the risk of investing in the protocol. Whoever bought at 20-30 or even 40K, has experienced this first hand… The YFI token can easily go back to 3K-5K if there is no no clear path to monetization for the token holder/staker.
I think that even though stakers are perceived as “free lunchers”, many if not most, invested a ton of money to participate in the YFI community/protocol. Anybody that got their original YFI tokens for free or at very low cost basis doesn’t see it this way, but anybody that got in at a higher cost basis as right now is painfully aware of the cost of being a “staker”.

[banteg]

I’m not opposed to actually introducing risk to holding YFI and it acting as an insurance backstop, for example. Then it’ll be better justified.