Protocol Fee Sharing and the Future of Uniswap

Atis E
9 min readFeb 27, 2024

--

In the coming months, the Uniswap DAO appears poised to vote on protocol fee distribution to UNI stakers. This article shares personal takes and speculations on the future impact of this vote.

Disclaimer #1: The author is an independent researcher, not associated with the Labs or the Foundation.

Disclaimer #2: The discussion focuses on the technical and cryptoeconomic aspects of the new model, without reflecting any legal considerations.

Summary of the UNI staking proposal

A proposal that aims to turn UNI into a fee-sharing token is currently being discussed on the Uniswap’s governance forum. The main ideas are to:

Upgrade Uniswap Protocol Governance to enable the permissionless and programmatic collection of protocol fees

Distribute any protocol fees pro-rata to UNI token holders who have staked and delegated their votes

Allow for governance to continue to control core parameters: which pools which are charged a fee, and the magnitude of the fee

This initial proposal lies the groundwork of another, future proposal, which would actually introduce protocol fees on some Uniswap v3 pools:

Assuming a successful on-chain vote, the community will then have the option to turn on fees.

As a Uniswap delegate, I’m broadly in favor of the proposed fee distribution to UNI stakers from Uniswap v3 pools. I think it would be an interesting experiment to run. The thorough preparation behind this proposal clearly shows the Foundation’s competence and capability.

At the moment there are still uncertainties, such as the specific pools where fees will be implemented and the exact protocol tax rate. The fee switch must be tailored to each pool, meaning fees will only be activated for a select group of Uniswap v3 pools. Economically, it doesn’t make sense to apply fees to pools with low volume or short lifespans, especially when considering the DAO governance efforts required for each fee switch decision. It’s probable that the pools selected will largely coincide with those where Uniswap Labs’ 0.15% frontend fee is already in place. Based on previous votes, it’s anticipated that the protocol fee rate will be set between 10% and 20% of the total swap fees.

Currently, there are no publicly available plans to implement fees on v2, v4, or UniswapX. These versions can be addressed in future discussions, with the V3 fee switch experiment serving as a crucial reference point.

The future of v3

Uniswap v3 is currently the most dominant iteration of the platform and, by some measures, the largest DEX in existence. However, I’d argue that Uniswap v3 is already on the decline for several reasons, outlined below. This implies that experimenting with the fee switch on v3 is relatively low-risk action.

V4 is coming and will largely replace v3

Uniswap’s development strategy has some analogies with the Intel’s former tick-tock model, where architectural advancements (“tocks”) are followed by optimizations (“ticks”). In this context, Uniswap v1 and v2 represent one tick-tock cycle, with v3 and the forthcoming v4 constituting another.

Uniswap v1 was a significant advancement over the state of the art at the time. V2 was architecturally similar, but with new features such as ERC20/ERC20 swaps and price oracles, and with other enhancements to the overall design. Despite the competition from v3, v2 remains robust, underscoring its enduring appeal.

V3 introduced another architectural leap with concentrated liquidity. However, it does have some major drawbacks, being not only more complex to use, increasing the market risk of the LPs, but also being much more affected by MEV leakage to arbitragers, particularly in lower-fee-tier pools for volatile assets.

I predict that Uniswap v3’s design will not sustain in the long run, with the tried-and-trusted v2 model likely outlasting it. Once launched, v4 is expected to quickly overshadow v3, driven by several improvements:

  • Gas optimizations: v4 introduces various optimizations, including a singleton design reducing ERC20 transfer costs and transient storage mechanisms significantly lowering gas expenses.
  • LP-friendly innovations: Features designed to recapture lost value from arbitrage (LVR) through auctions and dynamic fees promise a more favorable environment for LPs.
  • Support for v2 features: V4 will reintegrate full-range liquidity positions, accommodate fee-on-transfer tokens, and allow for LP donations within the liquidity pool, among other enhancements.

In essence, v4 is poised to address v3’s shortcomings while incorporating the strengths of its predecessors.

Intensifying Competition in the DEX Landscape

1. Rising rival AMMs

Uniswap v3 initially capitalized on its innovative concentrated liquidity model and restrictive licensing to dominate the sector. However, these particular advantages are no longer there. V3 core code now has been under the GPL for nearly a year. High-quality competing AMMs are emerging. Uniswap v3 does still have the most lindyness, and the most battle-proof and solid codebase, while some competitors have fallen to hacks. However, it’s inevitable that these benefits are going to wane with the time, too.

2. Evolving trading mechanisms

On-chain trading is moving towards intent-based systems. As a result, we can expect a decline in the proportion of trades that get routed through AMMs. While it’s far too early to declare the AMM model obsolete, the rise of platforms like UniswapX and Cowswap will likely continue and increase in their market share.

3. Emerging DEX-focused chains and rollups

The debate over whether a dedicated Uniswap appchain or rollup is the future of DEXes has compelling arguments on both sides.

  • The case for: As highlighted by Dan Elitzer, in the current model swappers pay more in Ethereum’s network fees than they pay to Uniswap’s LPs in swap fees. This is a shocking inefficiency. Additionally, the trader losses from imperfect execution due to slippage are also larger than the swap fees — at least that was the case in 2021/2022. The DEX trading UX does get significantly better with each year, but if the focus of Ethereum remains to be a settlement layer, these problems are unlikely to be fully solved in the future. One can keep optimizing and complicating the design of DEXes, however, it’s just not possible to fully replicate the CEX trade experience on a chain that has 12 second block times — at least not without sacrificing decentralization or censorship resistance.
  • The case against: no-one want to bridge their assets across chains for marginal improvements in their trading experience. Also, we really don’t need further liquidity fragmentation.

However, the potential for shared sequencing on Ethereum’s mainnet could be a gamechanger. This innovation would enable synchronous composability among rollups, allowing DEX contracts on a Uniswap rollup to access liquidity across all participating rollups seamlessly. Such a development would remove most of the concerns around liquidity fragmentation and user inconvenience.

All together, these factors make the continued dominance of the mainnet’s Uniswap v3 very unlikely. It does not have much to lose from carefully trying out the fee switch.

The LP question: taxation without representation?

Risks of the fee switch

Switching on fees could potentially backfire if it leads to many liquidity providers (LPs) pulling out of Uniswap. Losing LPs would mean less liquidity in the pools, making the pools in turn less appealing for traders. This decrease in swapping activity would lower the earnings for remaining LPs, causing more of them to leave. This could set off a chain reaction, where liquidity keeps dropping, leading to a worst-case scenario of a death spiral.

Fortunately, as of now it appears that DeFi traders aren’t overly sensitive to price changes. Even if some LPs decide to leave after fees are introduced, the overall volume of trades might not fall significantly. This means the LPs who stick around would end up with a higher APR (before the protocol fee is taken into account). There’s a point where these higher earnings from trades would balance out the costs of the protocol fees, leading to a stable equilibrium.

Supporting this view, Gauntlet’s research (currently being updated) indicates that a death spiral is very unlikely, provided that the fees aren’t set too high. Their data-driven approach gives some reassurance that introducing fees, if done carefully, shouldn’t lead to a mass exodus of LPs.

Fair value of the protocol tax rate

10% to 20% tax rate is not uncommon in the real-world economy. However, doesn’t directly translate to Uniswap’s situation for several reasons:

  • Typically, taxes target profits, not gross revenue. In the context of LPs on Uniswap, swap fees represent their gross income, not their net earnings. Research that measure the LP earnings using using either the impermanent loss model or the loss versus rebalancing (LVR) model (suitable for unprotected and hedged LPs, respectively), conclude that the net profit margin for Uniswap v3 LPs is nearly zero. A tax system aiming to levy earnings accurately would, therefore, yield minimal revenue under these circumstances, highlighting a mismatch between traditional tax logic and the dynamics of LP earnings.
  • Taxes serve to fund public services in the traditional economy, such as infrastructure and healthcare. Redirecting tax revenue to UNI token holders doesn’t inherently ensure benefits for the contributors, the LPs, in a manner that aligns with typical taxpayer expectations.

Despite these challenges, there are compelling reasons to consider the protocol fee:

  • The protocol fee resembles a value-added tax (VAT) more than income or capital gains taxes. A key feature of VAT is that it’s paid by the customer, not by the provider. LPs can somewhat force the swappers to pay more, and in this way offset the protocol fee, by migrating their assets to pools with higher fee tiers. However, without coordinated action, this shift is unlikely to happen broadly, leading to continued underpricing of LP’s liquidity in Uniswap v3.
  • The fee distribution model incentivizes UNI token holders to actively participate in governance by requiring token delegation for revenue sharing. Research shows that active delegation improves DAO decentralization. It’s also a large step making the Uniswap ecosystem (in particular, the Foundation and the DAO) self-sustainable, rather than making it to rely on the treasury.

LPs and the DAO

Combining the above, it’s evident that the fee switch proposes to:

  • exchange a portion of Uniswap LPs’ fee income
  • for enhancements within the Uniswap’s ecosystem.

However, it’s not the LPs who are being asked to make this trade — or at least, not primarily the LPs. The current situation is one of taxation without representation. At least, without adequate representation. Delegates who have explicitly prioritized their support for the LPs only have a few million UNI votes.

Data from https://vote.uniswapfoundation.org/?topStakeholders=lp: Uniswap delegates with a statement, filtered out by those who primarily represent Liquidity Providers. In total, only six delegates are in the list, three of which have zero voting power.

Moreover, there are some notable trading companies between the large DAO delegates. There’s a clear potential for conflicts of interest between them and Uniswap LPs, especially retail LPs.

My position is that LPs are the lifeblood of any DEX, and their voting power should reflect that. Hopefully, the upcoming UNI staking and re-delegation will be an opportunity for the Uniswap LPs to capture more votes.

One technical step towards increasing the LPs’ voting power would be through allowing the delegation of UNI tokens in their liquidity positions. This would necessitate some technical changes but is certainly feasible.

Compensating LPs?

Several ideas are floating around on how to link LP support with the fee switch initiative, for example:

  • liquidity mining programs
  • airdrops for “loyal” LPs
  • gas rebates

Even assuming that the LPs have sufficient voting power to pass these ideas (which is uncertain), I find none of the ideas fully persuasive:

  • Targeted liquidity incentives as a means of tax redistribution might be effective. Yet, they risk disrupting the free market and compromising Uniswap’s credibility as a neutral platform. Accurate targeting is challenging and would entail substantial monitoring, research, and management expenses, while poorly aimed incentives would mostly attract mercenary LPs, potentially harming the ecosystem.
  • The criteria for an airdrop could be easily manipulated by large token holders if it involves public DAO voting. Future airdrops could even more likely lead to governance crises, undermine Uniswap’s neutrality, and encourage airdrop farming as a speculative LP tactic. Moreover, airdrops without a KYC requirement would disproportionately benefit either large holders or sybils, while asking for KYC would be a much worse alternative, contradicting the permissionless ethos of Uniswap.
  • Gas rebates would unfairly favor active LPs over those taking lower risks or retail LPs. Their long-term effect would be to nudge the LPs towards de facto worse strategies.

The Uniswap Foundation’s strategy to focus on growing the ecosystem appears more viable than any of these options (with the possible exception of targeted liquidity mining). The ideal scenario is for LPs to obtain a relatively smaller share, but of a significantly larger pie.

Conclusion

  • Encouraging UNI token holders to delegate their tokens is expected to improve the decentralization of the DAO and the sustainability of the ecosystem.
  • At this moment, implementing the fee switch on V3 will provide critical data that will inform decisions for future versions of Uniswap and other protocols within the ecosystem.
  • LPs are the clear short-term losers from the anticipated fee switch, and it’s uncertain whether the long-term expected benefits for the ecosystem will be sufficient to make up for their losses.
  • LPs should strive for better coordination and a stronger voice in the DAO decision-making. This includes rallying support for more conservative approaches to the fee switch: for example, imposing no more than a 10% fee, and only on a small group of pools.

--

--