What Drives Liquidity on Decentralized Exchanges? Evidence from the Uniswap Protocol

Brian Zhu, Dingyue Liu, Xin Wan, Gordon Liao, Ciamac C. Moallemi, Brad BachuarXiv 2024 · DOI · arXiv

This paper provides a data-driven deep dive into the mechanics of liquidity on Decentralized Exchanges (DEXs), specifically focusing on Uniswap v3. By introducing a clever new metric, the authors decompose market depth into two distinct drivers—capital volume and capital concentration—revealing how factors like gas prices, volatility, and competition from "private" market makers shape the DeFi landscape.

TL;DR

The researchers developed a "counterfactual spread" metric to separate the effects of total capital (TVL) from how efficiently that capital is concentrated around the current price. They found that while volatility and fees affect both, factors like gas prices and DEX competition primarily impact liquidity by changing how tightly LPs (Liquidity Providers) cluster their positions.

What Problem Does This Paper Address?

In traditional finance, liquidity is well-understood through limit order books. In DeFi, liquidity is provided via Automated Market Makers (AMMs). Uniswap v3 introduced "concentrated liquidity," allowing LPs to provide capital only within specific price ranges. This makes measuring "market depth" (how much you can trade without moving the price) complicated.

Is a pool deep because it has a lot of money (Total Value Locked, or TVL), or because LPs are aggressively concentrating their money near the current price? Furthermore, how do external forces—like high gas fees or "aggregators" that route trades to private market makers—affect the willingness of LPs to provide on-chain liquidity? This paper aims to untangle these variables.

Methods

The authors employ a regression-based approach to predict market depth (effective spread) at time $t+1$ based on factors at time $t$. Their most significant methodological contribution is the $v2$ counterfactual spread ($Cv2S$).

  1. The Metric: They calculate what the spread would be if the pool’s current TVL were deployed in a standard Uniswap v2-style "constant product" pool (where liquidity is spread infinitely thin).
  2. The Decomposition: By comparing the actual Uniswap v3 spread to this counterfactual v2 spread, they can isolate concentration efficiency. If the ratio is low, LPs are concentrating capital effectively; if it's high, liquidity is "stale" or spread too wide.
  3. The Factors: They test five primary drivers: gas prices, token returns, volatility, fee revenue, and "markout" (a proxy for adverse selection/informed trading).

Datasets

The study uses a robust dataset spanning over three years (May 2021 to July 2024):

  • Uniswap v3 Data: Publicly available data from the top four blockchains by volume: Ethereum Mainnet (L1), Arbitrum (L2), Optimism (L2), and Polygon (L2).
  • Pool Selection: They focused on the top 100 pools by volume, specifically selecting token pairs that appeared across all four chains (e.g., WETH-USDC, WBTC-WETH) to ensure comparability.
  • External Data: Token prices from Centralized Exchanges (CEXs) for "clean" return/volatility metrics and Dune Analytics for gas prices and DEX aggregator volumes.

Key Insights and Ideas

  • The "Two Channels" of Liquidity: The paper proves that liquidity isn't a monolith. Some factors make LPs pull money out of the pool (TVL channel), while others just make them widen their price ranges (concentration channel).
  • Gas Prices as a Barrier: On Ethereum L1, high gas prices don't necessarily make LPs withdraw money, but they do stop them from rebalancing. This leads to "stale" liquidity that isn't concentrated at the current price, effectively increasing the spread for traders.
  • The "Informed Trader" Tax: Using "markout" (how much price moves against an LP after a trade), they show that when LPs face informed traders, they widen their ranges to protect themselves. This reduces concentration and hurts market depth.
  • Aggregator Optimism: A surprising finding is that "internalization"—where DEX aggregators fill orders using private inventory instead of the public pool—decreases the pool's TVL but does not significantly hurt the overall market depth. This suggests that public pools and private market makers can coexist healthily.

Key Results

  • Volatility and Fees: These are the "all-rounders." Higher volatility increases spreads through both lower TVL and lower concentration. Higher fee revenue does the opposite, improving liquidity through both channels.
  • Predictive Power: The models show that returns and gas prices affect liquidity almost exclusively through the concentration channel.
  • L1 vs. L2: Spreads are generally lower and less dispersed on Ethereum L1 compared to L2s, despite L1's higher gas costs, likely due to the higher total capital (TVL) present on the mainnet.
  • Competition: A higher market share for competing DEXs negatively impacts a pool's liquidity, primarily by discouraging LPs from concentrating their positions.

Limitations and Future Work

The authors acknowledge that while their framework is applicable to various AMM designs (v2, v3, v4), the empirical results are specific to the Uniswap v3 environment.

Future research could explore:

  • Dynamic Ecosystems: How liquidity shifts as more complex "hooks" (custom logic) are implemented in Uniswap v4.
  • L2 Evolution: As Layer 2 networks mature and gas costs drop further, will the "concentration channel" become less sensitive to gas prices?
  • Aggregator Dynamics: More granular studies on how different types of order flow auctions (like Dutch auctions vs. batch auctions) impact the incentives for on-chain LPs.

Summary updated

Abstract

We study liquidity on decentralized exchanges (DEXs), identifying factors at the platform, blockchain, token pair, and liquidity pool levels with predictive power for market depth metrics. We introduce the v2 counterfactual spread metric, a novel criterion which assesses the degree of liquidity concentration in pools using the ``concentrated liquidity'' mechanism, allowing us to decompose the effect of a factor on market depth into two channels: total value locked (TVL) and concentration. We further explore how external liquidity from competing DEXs and private inventory on DEX aggregators influence market depth. We find that (i) gas prices, returns, and a DEX's share of trading volume affect liquidity through concentration, (ii) internalization of order flow by private market makers affects TVL but not the overall market depth, and (iii) volatility, fee revenue, and markout affect liquidity through both channels.
BibTeX
@misc{zhu2024what,
  author = {Brian Zhu and Dingyue Liu and Xin Wan and Gordon Liao and Ciamac C. Moallemi and Brad Bachu},
  title = {{What Drives Liquidity on Decentralized Exchanges? Evidence from the Uniswap Protocol}},
  year = {2024},
  eprint = {2410.19107},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2410.19107},
}
Figure