Paper: arXiv 2212.05632

Authors: Yufan Zhang, Zichao Chen, Yutong Sun, Yulin Liu, Luyao Zhang

Abstract

Decentralized finance (DeFi) is known for its unique mechanism design, which applies smart contracts to facilitate peer-to-peer transactions. The decentralized bank is a typical DeFi application. Ideally, a decentralized bank should be decentralized in the transaction. However, many recent studies have found that decentralized banks have not achieved a significant degree of decentralization. This research conducts a comparative study among mainstream decentralized banks. We apply core-periphery network features analysis using the transaction data from four decentralized banks, Liquity, Aave, MakerDao, and Compound. We extract six features and compare the banks’ levels of decentralization cross-sectionally. According to the analysis results, we find that: 1) MakerDao and Compound are more decentralized in the transactions than Aave and Liquity. 2) Although decentralized banking transactions are supposed to be decentralized, the data show that four banks have primary external transaction core addresses such as Huobi, Coinbase, and Binance, etc. We also discuss four design features that might affect network decentralization. Our research contributes to the literature at the interface of decentralized finance, financial technology (Fintech), and social network analysis and inspires future protocol designs to live up to the promise of decentralized finance for a truly peer-to-peer transaction network.

Complexity vs Empirical Score

  • Math Complexity: 3.5/10
  • Empirical Rigor: 6.5/10
  • Quadrant: Street Traders — practical and empirical, lighter on theory

Why this score: The paper applies established network analysis tools (core-periphery modeling) to empirical blockchain transaction data across four major DeFi protocols. While mathematically straightforward, its cross-sectional empirical findings on the centralized reality of decentralized lending protocols offer useful practical and structural insights.

Research Flowchart

  flowchart TD
    A[Research Goal: Evaluate Decentralization Levels in Decentralized Banks] --> B[Data Collection: Transaction Data from Liquity, Aave, MakerDao, Compound]
    B --> C[Methodology: Extract 6 Core-Periphery Network Features]
    C --> D[Computational Process: Cross-Sectional Social Network Analysis]
    D --> E[Finding 1: MakerDao & Compound are more decentralized than Aave & Liquity]
    D --> F[Finding 2: All 4 banks rely on centralized exchange cores like Binance & Coinbase]
    E --> G[Outcome: Protocol Design Insights & Fintech Literature Contribution]
    F --> G