[Proposal] Revising delegated vault fees (part 1)

Proposals Vaults
1

Summary

In this proposal I try to justify the required changes to vault performance fee, so that it will be more justly for investors in delegated vaults, also to have more strategies.

Background

Currently vault performance fee is 20% [1, 2], 10 percent for vault and 10 percent for the strategist. For delegated vaults which transfer capital from one vault to another this fee will be charged at least twice for the same capital, once in the originating vault and once in the destination vault (for delegated vaults in more than 2 tiers, the situation will be worse). I think this behavior is not desirable and should be corrected since the fees are being charged for the same yield more than once, although some operations are done for transferring the capital (like depositing some token and borrowing another one [3]).

In the current proposal voting will just be done for vault performance fee, although I will describe my opinion for strategists performance fee, I will defer voting to another proposal after this one passes, since that part may be more controversial.

Motivation

I think current fee structure for delegated vaults is not appropriate because the investor will pay fees multiple times on the same yield (s)he has collected.

The other problem is that by the current fee structure, some delegate strategies specifically in long delegation chains may become impractical because of high fees.

The logic for supporting not taking the fees more than once, is similar to the practice of preventing double taxation of parent companies on the profits of their subsidiaries.

Specification

  1. The performance fee for the vault, could be made configurable for each strategy, and the global vault performance fee removed. By doing so, each strategy will have two performance fees, one that goes to the strategist and the other which goes to the vault. For delegated vaults we can set the vault fee for the strategy to zero in the source strategy and collect it in the destination strategy once (The last strategy doing the actual work).

  2. This second item is not the subject of voting for the current proposal (probably some future one[s]), I have described it here because it is correlated and for completeness.
    The performance fee for the strategist may be a little more complicated (For strategy A delegating the yield generation job to strategy B in another vault, which strategy should get what? Should they get equal portions of 10%? Is any of them more important than the other? What would be the situation if delegation path has more than 2 strategies?). Currently each strategist gets 10 percent of the yield his/her strategy has generated for its vault; Dividing this 10 percent between 2 or more strategists may be challenging (if agreed upon to have a division in further proposal(s), otherwise each strategist may take his/her 10 percent like before).
    The following method for dividing the performance fee between strategists comes to my mind: For each chain of strategies, an investment line will be defined. For example vault A has a strategy A1 which invests in vault B with some strategies including strategy B1 which invests in vault C with a strategy C1 which actually generates the yield. The yields generated in vault C may belong to some number of investment lines one of them being (A1, B1, C1) another one will be (C1) for direct investment in vault C, another line will be (B1 , C1) for investment in vault B. The strategist performance fee belonging to each investment line generated by the last strategy will be divided according to an agreed upon ratio between strategies, for example for (A1, B1, C1) each one may get 3.33% (one third of 10%), and the yield will be passed untouched during the path from C to B to A (nobody will take fees in between, strategists will take their fees from the last strategy in the line).

Disclaimer

I may be beneficiary to this proposal (for what I call improved performance) in future as a strategist by proposing some delegated vaults strategies.

Voting for the proposal

For: The vault performance fee should be moved from vault to strategy and made 0 for delegating strategies.

Against: The performance fee should remain as is.

Poll:

  • For
  • Against

0 voters

2

There are some assumptions here I’d like to address:

  1. Vaults aren’t delegating the funds, the connected Strategies are.

This seems like a subtle assumption, but I promise you it’s important. The authors of the Strategies, which are connected to Vaults your funds are deposited to, made a decision to architect the Strategy to delegate funds in some way. It is not a “Yearn decision” to choose that architecture. The Strategy author could just as easily cut out the “middle man” of depositing into another Vault (increasing the yield rate the strategy earns), but they are taking the shortcut of just leveraging the deposit so they have to write and maintain less code, and get access to a broader and more sustainable yield for the deposits they are sourcing with the Strategy (usually based on a leveraged or borrowed trade).

This takes me to my next point:

  1. Strategies take a lot of effort and maintenance to get working, they don’t come “for free”

As mentioned before, there is a lot of work that the Strategist has to do to build and maintain their Strategy. In compensation for their efforts, the Vault code rewards them 10% of the yield that the Strategy generates with capital from the Vault, which encourages them to build the safest and strongest Strategies possible. Now as these Strategies get larger, single authors tend to form “committees” that share the work and responsibilities, pooling rewards in the process. Yearn is not making the decision to do this, ergo the decision to change reward structure comes at the cost of less incentives to create and maintain Strategies, which negatively impacts your deposit’s yield in the long-term. I’m not saying that the fees can’t be changed, but it is important to understand why the fee works the way it currently does before suggesting a change.

But Strategist fees aren’t the only costs involved with a Strategy, which brings me to my next point:

  1. Yearn’s protocol operations (e.g. Keep3r harvesting, gas costs, etc.) cost money too

The Strategies need robust revenue flows to keep them running smoothly. Remember, the reason why Strategies are more capital-efficient and earn a higher yield by pooling together in a Vault with other depositors (rather than operating the Strategies yourself) is because the costs of execution are socialized across all the depositors in the Vault. You could think of the other 10% perf fee that the DAO earns as income as revenue to pay for these operations, as well as numerous other things such as recurring grants to pay for full-time work by very competent devs, Strategy reviews, etc.

Now again, we could potentially make a decision to reduce these fees, as the revenue generated by them might be larger than the expenses we have, but that comes at the cost of less surplus revenue stored for when these varying costs get really expensive. One just has to look to the recent past to find examples of when yields were down, but gas prices were up, and Strategies still need to harvested to stay effective. It’s important to plan ahead for these circumstances, which we do.

But Yearn earns revenue not just from 10% perf fee, but additionally from the 2% mgmt fee, which takes me to another point:

  1. We actually already give a discount on management fees for Strategies that delegate funds

This is done as a convenience to Strategy creators. It is very difficult to create Strategies that earn a consistently high yield, and we were constantly running into scenarios where management fees affect the yield of a delegating Strategy, so we introduced the concept of BaseStrategy.delegatedAssets to compensate for this scenario, and reduce the threshold below which a Strategy cannot be profitable. As this type of revenue is what the DAO takes in the long-term to keep itself growing, it was determined that there was enough slack in our revenue models to create a discount at the management fee, especially if it improves the number of Strategies that can be written for these Vaults, which tend to be some of our highest in terms of funds deposited.

And this takes me to my last point:

  1. $5.66b TVL is okay with paying these fees as is

Well, it’s very difficult to query all of the depositors in our Vaults, however we can leverage the fact that these users haven’t withdrawn as evidence that the fees are acceptable. Could we bring in more TVL if we dropped fees? Yes, that probably is the case, because everyone likes paying less to get more. But Yearn’s design doesn’t benefit from accepting more deposits (as it hurts yields), and it makes the job of incentivizing contributors to do the hard work of keeping Yearn running even harder because of the lower fees. As mentioned above, there’s a number of balancing incentives at play we need to keep in mind when suggesting fee changes, and I have yet to see a proposal to do just that take these into consideration.

But fear not, there’s another way to approach the problem that falls out of this proposal:

The TL;DR here is: what if instead of giving away our valuable revenue streams to make our product worse, we drive back some of the revenue the protocol generates to YFI holders that stake and do helpful things for the protocol? You could think of Phase 1 (xYFI) as being a way to get a “discount” on Vault fees by staking, because the revenue that is used to earn the kickback is actually generated by the same fees you are paying as a depositor. Phase 2 (veYFI) is a way to leverage up on that discount, by commiting to stake your YFI for a longer term. Phase 3 (Management Gauges) and 4 (Useful Work) is a way to get YFI stakers to reduce the workload of operating the protocol by taking some risk on themselves, which means the operational overhead of the protocol for contributors gets smaller, which means the amount of our revenue needed to support their activities also gets smaller and more revenue can be kicked back to stakers leading to even more “discounted fees” in a virtuous cycle.

I think this is a rather elegant and positive-sum solution to the problem of our fees being perceived as “too damn high”. This outcome will improve the scalability of our protocol, and make it easier for us to run even more Strategies in the protocol, incentvizing more yield generation via Strategies and making Yearn a very sticky money lego indeed.

I hope this makes sense to you, my apologies for the long-winded response, but I think it’s important that people understand the finely-balanced game that the fee model is, and how it will take a lot of work and a lot of data to improve upon that equilibrium for the benefit of all stakeholders of Yearn, not just favoring some parties at the expense of others.

2 Likes
3

  • Thank you for the reply and participation

  • This proposal originated when I was trying to design a strategy including some form of delegation, since it was not competent with the best vault strategy sometimes, I started to look into fee structure more deeply. Although the strategy did not develop in its original form for some other fundamental problem, the fee structure concern remained with me.

  • My current concern is more about the sustainability. I prefer my efforts go to a place where all corners has got attention as much as possible, so if community decides the current fee structure for delegated vaults is good enough (for example for investors not to leave), I am ok with it; it may also be in my favor as a potential strategist, my doubt is about long term performance.

  • Another point is that, I want to insist the current proposal voting is NOT about strategist fee or management fee, it is just about vault fee, although we may have discussions about them.

  • For a system, from the user point of view we can have two extremes:

  1. We can have justified costs up to the point, user benefits start to drop and they start leaving.
  2. We can start to pressure system itself in the direction of user benefits, until it starts to collapse.
    I think the second one has strong supporters, I do not want any collapse, just mentioning the extremes.

  • A view from an investor could be that, (s)he says: “I will look at Yearn as black box which takes my assets, generates yields, takes 20% of it, gives me the rest. Yearn is free to rotate my assets as much as required inside its system.”

  • I think the current fee structure for delegated vaults is not consistent with what is described here, at least I can not find any difference between delegated and normal vaults (seems just an unanswered question). Also the point mentioned in the reply (“The Strategy author could just as easily cut out the middle man of depositing into another Vault …”) shows that it is a probably good assumption to think the share of fees could be the original 20% and not increased if the path for vault generation becomes longer.

  • Another point which comes to my mind (although not to be an issue now and even never): Let’s assume there will be some buffer strategies which if inserted will just pass the funds to the next vault, but take a portion of yield when yield is back propagated. If we fit everything in that 20%, it is an auto-control mechanism for such insertions, but if it is allowed for every insertion to take its 20%, the internal control just relies on the open source and deployment reviews.

  • About the mentioned points

  1. I think the difference between delegated strategies and vault is also apparent from my original post, I prefer the term “delegated vaults” originally used in yearn medium post where ever it does not produce disambiguation.

  2. I also think strategists should be compensated enough for the system to work properly. Although the voting here is not about the strategists’ share of fees, trying to fit everything inside the 20%, will probably act as feedback mechanism which will make some strategies not profitable enough and push the efforts and resources to something more profitable.

  3. I agree that the cost of system should be covered by its revenue. One of the factors affecting the reduced vault fee by this proposal, is the amount of delegation in entire system, which I guess should not be high (Although cover of the costs may be possible independent of vault fees according to the next paragraph).
    Although it is good assumption to think the numbers has probably somehow changed from this report, my understanding of management fee being 2% of TVL mentioned in 5, will easily cover the costs mentioned in page 7 of the report with a good margin.
    Actually my mention of 0 in the voting, is to make it easier to vote. Since my understanding of the correct poll to post is a [“For”, “Against”] one and this may have already made decision making somehow complex (The “part 1” in the title, other parts may come), I have tried to merge the two polls of 1- Making vault fees configurable for each strategy and 2- Making vault fees 0 for delegated strategies, to one poll. We can revise the second one and define acceptable levels for negligible (0) vault fees for delegated strategies in further steps. I think 0 is fine for now according to my previous reasoning.

  4. Although I should study more what has been mentioned here, the choice of changing system parameters for the benefit of some, remembers me of the early days of ethereum which did something similar and created a fork. Although I was not happy with the change and I think nothing similar has happened again in ethereum, it is now successful. I think a system should not create such exceptions.

  5. About the increased deposits and decrease yields, I am not sure if by investment of I, the yield is Y, by investment of 2I, even Y is not possible (2Y may not be possible, but I think Y should be; I think we have also some capacity limitations implemented here, the distribution for delegated vaults maybe something which may need attention). The TVL amount and investors not leaving, may be a sign of not a good competitor existing yet. A benefit of doing the right thing (is this proposal about?), is building trust which in the presence of good competitors will help the system.

  • About the proposed way out of the proposal, I agree, the problem of distribution of what to who, may be hard in lot of cases, but I do not prefer to look at my current proposal as a question about “we have some extra fees here, why should not we spend it for the sake of Yearn’s benefit” instead of thinking it about “whether or not delegated vaults should be made similar to other vaults as much as possible”

  • Update 1: It seems in replying to point 3, I have compared something quarterly (the report) with something yearly (the 2% management fee), although this should be corrected, because of the high margin, it has no effect on the result.

4

I will make a point about terminology: when I read “delegated vaults”, that makes me think of vaults where 100% of the funds are being delegated to another part of Yearn’s protocol, such as an older version of a vault depositing directly into a newer one.

In that circumstance, I do believe the fees charged should be made as close to 0% as they can be, because this is a maintenance action.

I think for purposes of arguement, we should consider any Strategy who delegates a % of its funds to another part of Yearn’s protocol to be a “delegated Strategy” (and not a delegated Vault) for sake of precision of terms (despite what past articles may have said).

5

About the delegated strategy I agree with the definition; a recommendation would be not to generate yield from both delegation and other operations in a single strategy as much as possible, otherwise it will complicate at least fee calculation.

About the vaults, in order to bring closer

  1. What is mentioned in the article (a vault with a single strategy)
  2. What is mentioned in the reply (“… vaults where 100% of the funds are being delegated …”)
  3. The poll in the current proposal

let’s define the new “virtual sub vault” as a portion of a vault which belongs to at most one strategy and assume the “virtual sub vault” is delegated if its associated strategy is a delegated strategy.

6

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.