"Fee Switch" Pilot Update & Vote

[Leighton]

This post is co-written with @guil-lambert

In July of 2022 a proposal was put forward to pilot turning on the “fee switch” for a small set of Uniswap protocol pools. A temperature check passed and a consensus check also passed voting.

During the process, many people voiced a desire to have more time to research this proposal before voting. In light of that, the vote was postponed until December 1st. This pause led to helpful research from Alastor as well as input from the newly created Uniswap Foundation.

Now that it’s December 2nd, the vote is moving forward! The purpose of this post is to re-iterate key points of the proposal and next steps so the community can be ready!

Key Proposal Points:

The rationale behind the proposal remains the same. For clarity, we are stating a few of the key points here but it’s recommended to read the original post.

• This proposal is solely designed to test the impact of the “fee switch” on protocol usage. The proposal does not dictate how any tokens retained during this period are used (or if they ever will be used). The proposal does not create any expectation the retained tokens will be paid out to UNI token holders. On the contrary, this proposal advocates that if future proposals use accrued tokens, that use be restricted to public good purposes for the community and to grow the Uniswap protocol.

• This proposal is truly designed as a pilot program intended to evaluate the impact of turning on the fee switch on trade execution.

• Turning on the fee switch will not increase fees for people using the protocol to swap. It will simply retain a small portion of what is currently being paid out to liquidity providers.

• To limit the potential impact of the pilot program on liquidity providers, we are proposing to activate the lowest possible settings for the “Fee switch” (1/10) on a limited subset of pairs (ETH-stablecoin “sister pools”) and for a limited time (120 days).

• The metric to measure success of the experiment is the following: If trading execution is not diminished for the pools with the “fee switch” turned on – the experiment is a success.

Proposal Settings:

We thank the Alsator team for the thorough and comprehensive “Uniswap Fee Switch Report” which has been released earlier this month (link to the report: Uniswap Fee Switch - Alastor Report (11.14.2022).pdf - Google Drive). The key takeaways from the report are that the fee switch should be designed to grow both volumes and market share of the Uniswap protocol. We wholeheartedly agree with this goal and we are sure most UNI stakeholders will be guided by similar principles.

However, we chose not to follow their recommendation to not turn on the fee switch for any of the “lower fee tier” pools. Their report argued that doing so could decrease Uniswap’s volume and market share, but at this point we are frankly unsure if anyone knows whether that would be the case.

They also made the observation that “sophisticated” LPs often sit in lower fee tiers, which by association also suggest that the retail-level LPs sit in the higher fee tiers. Hence, only targeting the higher fee tiers could disproportionately impact smaller players. We believe the community would agree that the fee switch should not negatively impact retail users while at the same time favor sophisticated LPs.

In addition, several users raised concerns about turning on the fee switch for the ETH-Dai 5bps pool. As per the Alastor report, the ETH-Dai pairs collectively have the least amount of liquidity amongst the ETH-stablecoin pairs, and we agree that targeting the ETH-Dai 5bps fee tier, which has the most volume, could negatively impact liquidity providers at all levels.

We thus decided to follow @alanalevin suggestion to use the ETH-USDT-0.05% as a possible alternative to the ETH-Dai-0.05% pool.

Therefore, we propose to set Fee parameter to 1/10 for the following pairs:

• ETH-USDT-0.05%
• DAI-ETH-0.3%
• USDC-ETH-1%

All accrued protocol fees will remain “uncollected” inside each pool smart contract until governance agrees on best use for funds via a vote.

Proposal Success Metrics:

The metric to measure success of the experiment is the following: If trading execution is not diminished for the pools with the “fee switch” turned on – the experiment is a success. Note that this criteria explicitly does not consider the impact of the fee switch on liquidity provider returns.

More intangibly, we hope this proposal helps catalyze robust discussion and work around Uniswap, the world’s most successful decentralized finance protocol. We hope this work is broad and creative, showing the world what new models of coordination are possible.

Timeline:

We anticipate the on-chain vote going live in the next 14 days. We are currently doing technical diligence to ensure the proposal itself is formatted correctly. We will post this code when it is ready for additional review by the community

Disclosures

The authors of this post hold no UNI tokens nor have any price exposure to UNI via alternative methods (i.e. margin, leveraged trading, etc.).

[Anonz]

@Leighton

trading execution is not diminished

Please define how you will evaluate the following “ trading execution is not diminished”.

[breeze]

Please explain, how I (as a liquidity provider) will benefit from earning 10-25% less fees?

[guil-lambert]

Excellent question!

We expect some LPs will remove liquidity from the fees switched pools to another pool. This is totally understandable and expected to happen. But how would that liquidity affect the execution price?

In other words, if liquidity is moved from the 30bps pool to an identical price range in the 5bps pool, would that increase the slippage for the average trade? The swap router could alleviate some of the inefficiencies by re-directing trades to the “most liquid” pool, but I’d say the actual impact of that liquidity re-deployment is largely unknown.

“Trade execution” could thus be monitored by tracking the (slippage+fees)/tradeSize for all ETH-stablecoin trades that went through the SwapRouter before and after the fee switch. This would basically track whether the “effective” liquidity of the whole protocol went up/down or stayed the same. We’re open to other ideas as well.

At the end of the day, we first and foremost want to focus on studying the impact of the fee switch on trade execution for swappers, and not on its impact on LP revenue.

[PatrickOD]

Has the Uniswap Foundation or any other parties provided any analysis on tax or regulatory considerations for this pilot?

I know this was a concern brought up by a16z and others a few months ago and seemed to be one of the main reasons for those not in favor. Just wondering what progress has been made.

[devinwalsh]

Hi @PatrickOD, we posted this brief earlier today on legal and regulatory considerations: Uniswap Protocol Fee Switch Considerations — Uniswap Foundation

[PatrickOD]

ty! and sry did not see that linked by OP… That post is helpful in understanding the thought process that went into the proposal and how it addresses those concerns. I realize there’s been very little guidance on either tax or regs, at least in the US, so there’s unlikely to be definitive answers in the short term. I think that leaving the fees “uncollected” is a very sensible way to reduce some of those risks while still assessing impact on trading activity.

[RasterlyRock]

This seems like it would make sense for the definition of “trading execution.” Need something formulaic that can be measured (and verified) before and after the fee switch is turned on. Would make sense to measure the 120 day period leading up to the fee switch change (+120 days following the experiment) and compare those results to the 120 day period where it is turned on.

[RayFi]

with the collapse of CEX, it’s great timing and opportunity for Uni to move forward, to stand on the center of DeFi.
A good incentive to Uni token will bring much more awareness and attentention from blockchain community, CEX user will move to Uni more willingly, we can expect a trading volumn increase significently persentage wise.
And benifit the LP with more trading fees eventually.

[msilb7]

Hey (personal opinions), super interested to see this experiment go live. Love that the approach is to test this small and evaluate.

Proposal Success Metrics:
The metric to measure success of the experiment is the following: If trading execution is not diminished for the pools with the “fee switch” turned on – the experiment is a success. Note that this criteria explicitly does not consider the impact of the fee switch on liquidity provider returns.

Throwing out some other ideas to evaluate for the eventual analysis:

Is this good/bad/non-consequential for the Uniswap product?

• LP Flows: Compare net inflows-outflows for “fee switch” pools vs ~equivalent non-“fee switch pools” & other DEX’s pools (i.e. does this cause LPs to leave? With an attempt to control for price movement & external factors)
• Volume Share: Is Uniswap’s DEX market share of volume for these pools higher/lower/unchanged? Do we see any patterns of volume moving to alternative pairs (i.e. other stables, other fee tiers)?

Does this matter for token holders?

• UNI holder composition: Do we see an increased UNI holder base, do tokens get more concentrated or distributed? (Could measure this with the first forum post as the “start date” - also thinking of this like user acquisition/retention/churn)
• UNI holder loyalty: Do we see greater/unchanged Uniswap usage by UNI holders? (i.e. raw volume, LPing, share of holders’ total DEX volume - Thinking of this like how Prime made consumers more loyal to Amazon)

I’m sure this is being considered, but it would be great to see another Grants Competition to analyze this!

[research]

I just want to point out a few questions that I have, I hope you guys can answer them. We have analyzed the Uniswap largest pools, and most of the liquidity provided is done by a few actors, some of which are providing liquidity although they are out of range =( see:

161944 (USDC/ETH 0.3%) 9M
368120 (USDC/ETH 0.3%)10M
314029 (USDC/ETH 0.3%) 6M
259042 (USDC/ETH 0.3%) 12M
More than 20% of the pool is filled by 4 LP

360541 (USDC/ETH 1%)10M
50% of the pool is filled by one LP - Maybe good to ask him if ok - But I guess as you guys chose that pool you probably have already discussed with them. If not, that would be a mistake.

etc…

As a large Uniswap LP ([RFC] The Optimism-Uniswap Protocol Liquidity Mining Program - #20 by research), we are delighted if you put a 1/10 fee in the protocol to foster innovation, increase utility and VOLUME for the platform, we believe this is absolutely amazing, please put the fee switch on, for every pool we LP, please! We will vote for this!

My main question is “how to actively but organically increase volume on Uniswap?” Short term LM incentives are not a long-term viable solution IMO.

BMV - Bring More Volume - Why do people trade on DEX? Who trades on DEX? How do people trade on DEX? Isn’t the DEX volume mostly automated? Isn’t DEX volume mostly related to Arbitrage and MEV opportunities? Is there a competition between CEX and DEX or are they complementary and one needs the other for it to work properly ? Do CEX use DEXes? If not, why not?

I think the discussion should be centered around this rather than knowing if LP will stay or leave because of the fee switch. If fees increase 50% thanks to Fee Switch, why would LP care about a 10% tax on top of this. If volume decreases drastically because of Fee Switch, then LP might need to derisk and exit LPing even if the tax was 0.5%.

Hope that little grain of salt helps =)

Looking forward to discuss.

[AbdullahUmar]

We (Blockchain @ Michigan) are in favor of moving forward with the fee switch pilot. The community must understand that this implementation is merely a 120 day test on a select few pools with the lowest % fee collection possible. The tradeoff of not going forward with experimentation is stifling governance/revenue model innovation.

As for success metrics, we should make those more clear to remove any confusion. @guil-lambert’s comment on tracking the ETH-stable trades via the SwapRouter before and after fee switch would be a good addition.

[Jack_Longarzo]

Hi Uniswap community. I’d like to share my thoughts on this proposal as I think it is a very important moment for Uniswap. I come here as someone who wants to see DeFi succeed and sees Uniswap as integral to this. The research by Alastor and the Uniswap Foundation has been excellent and should be used to continue to drive discussions on the fee switch, but I think now is the wrong time to execute this proposal.

First, I feel for the Uniswap community who may be hurting due to price action of the UNI token. The crypto markets are down and the Uni token is as well. It is natural to want to pursue changes when things seem like they are not going well. In this situation, however, I think that inaction is the best action.

There are two core reasons why I think this is not a good idea. First, I think a fee switch would be counterproductive to the objectives identified by Alastor. I fully agree that TVL, market share, and trading volume growth should be the Uniswap Protocol’s top priorities in this phase of its lifetime. While we may identify that the fee switch does not or has minimal negative impact on these metrics for certain pools, taxing LPs certainly cannot help here. I see only downside across the core KPIs.

The other reason is that this proposal could have very serious and unpredictable negative externalities for the Uniswap community. The Uniswap Foundation’s brief discusses the severe lack of clarity for DeFi in the U.S. It is very possible that this proposal could negatively impact community members like Uniswap Labs, the Uniswap Foundation, and even individuals in the community. I don’t think the proposed exploration is justifiable amidst these uncertainties.

The community treasury is large and there is no lack of funding for growth initiatives. There is no need to risk stifling growth or any other consequences right now.

Today UNI is the single most valuable DeFi governance token by market capitalization. This isn’t because of the prospect of immediate cash flows. It is because the token marks ownership in a decentralized community for the single most revolutionary and useful protocol in all of DeFi. The Uniswap vision is far bigger than a fee switch. It should not be jeopardized or hindered at this point in the protocol’s lifetime for the exploration of unnecessary fees.

Disclosure
I do not hold any UNI nor do I have any exposure to UNI. This is not financial or legal advice. I am only providing my thoughts on this unprecedented initiative and will support the Uniswap community in its decision.

[GFXlabs]

At GFX, we’ve been working towards getting the Fee Switch turned on for more than a year and, over the last several months, have been slowly gathering opinions from delegates and UNI holders. Our general observation has been that most UNI holders are interested in seeing the protocol monetized. The differences in opinion tend to come from when and how the protocol should be monetized. Some believe it should be put off until Uniswap gains greater market share, while others believe that monetizing the protocol today could rekindle interest in the protocol, governance, and UNI. Over the last six months, there has been a jump in interest in getting the switch activated, and now how has become the main question.

The how (implementation) generally consists of the flowing questions (ignoring v2 to simplify this):

1. Which pools should protocol fees be turned on for?
2. What should the protocol fee be set to?
3. How will the protocol set the fees?
4. How will the protocol claim the fees?
5. How will the protocol manage the positions it will accrue?
6. What will the protocol do with its revenue?

However, those questions only address the technical implementation of the Fee Switch. There are several legal questions we’ve come across that voters would like to see addressed, such as the following:

• Is Uniswap responsible for paying income taxes (and other potentially relevant taxes)?
• Do UNI voters need to come up with their own process to assess whether assets in the protocol are or aren’t securities similar to traditional exchanges?
• What might happen if Uniswap generates fees from a pool that contains an asset designated as a security?

We think there are certain ways to activate the Fee Switch, which could maximize the value of the protocol while minimizing legal risks.

The metric to measure success of the experiment is the following: If trading execution is not diminished for the pools with the “fee switch” turned on – the experiment is a success.

We’ll be voting against the proposal if it progresses to a formal governance vote because we believe the primary metric of success is unlikely to be met with the proposed implementation. Additionally, the proposal needs to address the above key questions. The below explains why we believe success is unlikely; we can make a follow-up post regarding why the proposal doesn’t sufficiently address the above question if it needs to be clarified.

While some may say that the proposal is merely an “experiment,” an unsuccessful implementation will likely hinder future efforts to monetize the protocol and reflect poorly on the state of Uniswap.

Question: Why do we think the proposal will not meet its stated success metric?

If passed, the proposal will set a protocol fee of 1/10 of the fee tier of the pool for the designated pools: ETH-USDT 5bp, DAI-ETH 30bp, & USDC-ETH 100bp. For example, the USDC-ETH 100bp pool applies a 100bp fee to trades and distributes the full fee to the LPs. With the fee applied, the fee on swaps remains the same, but the fee to LPs drops to 90bps.

Necessary context:

The protocol has four fee tiers: 1bp, 5bp, 30bp, and 100bp. Each asset pair can only have these four fee tiers; no additional pool can exist. For example, there is one ETH-USDC 1bp pool, one ETH-USDC 5bp pool, one ETH-USDC 30bp pool, and one ETH-USDC 100bp pool. If someone were to try to make a second ETH-USDC 30bp, the factory contract would prohibit it. If someone tried to make the inverse pair like USDC-ETH 30bp, the factory contract would also prohibit it.

To help normalize this information, its best to view them in a familiar format:

Tiers	Taker	Maker
100bp	1.00%	-1.00%
30bp	0.30%	-0.30%
5bp	0.05%	-0.05%
1bp	0.01%	-0.01%

Takers are people swapping with the pool, whereas the Makers are the LPs in the pool. Makers are currently receiving a 100% rebate for liquidity provided.

Answer: By introducing a 1/10 Protocol Fee on select pools, the protocol is reducing the rebate the LP will earn. Further, by only introducing the Protocol Fee to select pools, active LPs will likely move to one of the other three fee tiers where the rebate remains 100% or will move to a like-kind pair.

For example, if we were an active LP in the ETH-USDT 5bp pool and saw a rebate reduction of 10%, but the ETH-USDC 5bp still offered a 100% rebate, we’d simply move to that pool instead.

Potential alternatives
Below are our brief thoughts on implementing a Protocol Fee for the purposes of analyzing changes to LP behavior and swap execution. The list is from the most optimal implementation to the least optimal implementation to active Protocol Fees.

1. Set a single fee for all Uniswap deployments: leaving LPs the fewest alternatives
2. Set a single fee for a single deployment: LPs could move to another deployment
3. Set a single fee for a pair, and it’s like kind pairs: the LPs could move to other assets
4. Set a single fee for a select few pairs: the LPs could move to like-kind pools

Call to action
If you’re a UNI token holder and supportive of a thoughtful proposal to turn on the fee switch, please reach out.
governance@gfxlabs.io

[RayFi]

The immediately answer from LPs will say “no”, since this switch will cut off immediately.

But are they really gonna to do so? Taking economically, A rational LPs will only do it if they can find a better alternatives, which I consider would be very difficult.

It could be another protocol, or another Uni pool but which haven’t selected for this trail.

If the first case happen, then UniSwap has a competitor, we need be careful; if it’s the 2nd cases, there are not much to worries, UniSwap is still the best protocol to stay.

One more thing, those smart LPs, they are wise enough to keep some Uni token in pocket if they going to vote yes. An immediate return on token will surpass the LP income at least a medium long period of time.

[stephenwow]

let it do,i agree…just do it

[craiglr]

Hello Uniswap community. Let me start by saying I support this proposal as written. It represents a first step towards learning how to safely charge fees on Uniswap. There is still plenty more work to be done after using intelligence gained from the trial.

This is a pilot. The decision should come down to:

• Whether it will further intelligence by learning something about the protocol?
• Whether this something is worth knowing?
• What are the costs associated with running the pilot?

The focus of this proposal is narrow.

The proposal is solely designed to test the impact of the “fee switch” on protocol usage.

Observing the reaction of liquidity capital (taking broader market conditions into account) will be useful in assessing how bearable fees are on Uniswap v3. I have little confidence that anyone knows how LP’s will react. Uniswap is a marketplace. There are extremely complex dynamics affecting behaviors.

If fees do turn significant capital away from these pools (or spread out over wider ticks, who knows) then it is worth learning that and improving in future tests. The only way to truly find out is in a live simulation. There is also a predefined end to the pilot. Importantly, the default is not to continue indefinitely.

As I see it the costs are the risk that one or both of the following happen:

• Capital permanently leaves these pools. All else equal that widens Uniswap v3 spreads relative to alternatives.
• Legal issues. I’ll leave that to someone who knows what they’re talking about.

The recent average TVL of these three pools together is approximately $70mm TVL combined relative to Uniswap v3 Ethereum’s $2.6bn. This represents 2.7% of TVL compared to an average standard deviation of the change in TVL of around 5%. The benefit of learning the effect on the Uniswap protocol in a controlled, time limited trial is worth risking a small percentage of this TVL.

[Spongky]

yeah At the end of the day, we first and foremost want to focus on studying the impact of the fee switch on trade execution for swappers , and not on its impact on LP revenue.

[Stastny]

We appreciate the initiative from @leighton and @guil-lambert moving this pilot program forward. In general, Alastor supports moving forward with a fee switch pilot program, even if it doesn’t necessarily match our specific recommendations verbatim. A few additional thoughts from the Alastor side of things:

Regarding Pilot Pools - While we stand by our stance that implementing the fee switch on lower fee tier pools creates perverse incentives for LPs relative to the overall goals of the community, we are not against including one of these pools in a pilot to test this thesis. That said, we would recommend using the 0.05% ETH-DAI pool rather than 0.05% ETH-USDT pool as the 0.05% test case. Our rationale:

• ETH-USDT makes up the largest ETH-Stable spot-trading market in the world today (~3x the volume of ETH-USDC across both DEX/CEX competitors), of all ETH pairs Uniswap should be targeting this volume rather than disincentivizing its most efficient liquidity

• Relatedly, Uniswap decentralized market share in ETH-USDT is much lower than that of the other two ETH-Stable pairs being considered (~50% against, versus >90% in both ETH-USDC and ETH-DAI)

• While the ETH-DAI pools do indeed have smaller TVL than ETH-USDT, they are similar when normalized against volume serviced

Ultimately, we believe that it would behoove the community to utilize ETH-USDT as the primary test case to see if market share can truly be improved by utilizing the fee switch as an LP incentivization tool given the market share runway that currently exists.

Regarding Tracking & Measurement - To echo @msilb7, we would strongly encourage defining decentralized volume market share as the primary measurement of success for this pilot. We also would recommend (at a bare minimum) tracking order volume prior to and during the pilot program as well as tracking TVL flow, on both an aggregate and on an individual level in the days and weeks following implementation. It is important that the ramifications of this pilot not simply be considered anecdotally.

[larry]

Larry from Reverie here.

First off, I really like the idea of running experiments with the fee switch. That’s because over the long run, I view protocol fees as the principal way for the protocol to support itself in a sustainable manner.

Having said that, I think we need to understand the tax impact before we turn on the fee switch.

Simply put, turning fees on will generate revenue for the protocol (even if the accrued fees are not paid out to tokenholders). As I see it, the known unknown is “who will cover the potential tax bill on profits generated by the protocol?”

Before we have a better answer to this question, I don’t think it’s wise to turn on the protocol fee switch.