Decentralization illusion, voting democracy and liquidity risk in decentralized finance

Sun, Xiaotong (2024) Decentralization illusion, voting democracy and liquidity risk in decentralized finance. PhD thesis, University of Glasgow.

Abstract

This thesis is made up of four chapters on Decentralized Finance (DeFi). Simply, a blockchain is a distributed ledger that facilitates and records transactions, and DeFi is blockchain-based decentralized financial systems. The execution of transactions in blockchain and DeFi relies on code instead of trusted third party, allowing any agents to access blockchain and DeFi without limitations of space or time. Therefore, decentralization is considered as the most significant virtue of blockchain and DeFi. This thesis answers two important questions. First, is governance truly decentralized in DeFi? Second, is DeFi immune to liquidity risk? Are DeFi users with centralized power the source of liquidity risk?
In DeFi, the most widely adopted solution to decentralized governance is Decentralized Autonomous Organization (DAO), where all DAO participants have the authority for decision-making within the underlying DeFi application. Conceptually, DAOs embody decentralized governance structure. However, but the centralized distribution of decision-making power in influential DAOs is challenging the DeFi supporters’ beliefs in true decentralization. In chapter 3, I focus on governance centralization in DeFi and choose MakerDAO and its Maker protocol, the most attention-getting DAO and DeFi application, as a case study. Through an analysis of the voting history in MakerDAO governance, I present novel facts about highly centralized governance, such as low voting participation and concentration of voting power. Furthermore, I investigate the impact of centralized governance on market performance of Maker protocol. The empirical results suggest that governance centralization has complicated influences on DeFi, implying that DeFi users face a trade-off between decentralization and DeFi performance.

The decision-makers in DeFi are not separated individuals, and the interactions between decision-makers may exacerbate governance centralization. In chapter 5, I delve deeper into MakerDAO governance, developing a method to detect potential voter coalitions in MakerDAO. By applying clustering algorithms to voting history of MakerDAO governance polls, I identify three distinguished voter coalitions, with one coalition comprising the most voters and contributing to most total votes. Furthermore, I study the dissimilar effects of voter coalitions on the performance of Maker protocol, where both voting share and group cohesion of voter coalitions matter. Surprisingly, the largest coalition (i.e., the one with the most voters) often exert the opposite influence compared to smaller coalitions. Empirical results also indicate that voter coalitions can drive cryptocurrency flows issued by Maker protocol in different ways. This chapter seeks to enrich our understanding of governance centralization in DeFi by considering the dynamics of cooperation and competition of voter coalitions.

Beside participating in DeFi governance, DeFi users have other ways to gain centralized power, introducing potential financial risks. In chapter 5, I focus on liquidity risk and market concentration in Lending Protocols (LPs), resembling banks in DeFi ecosystems. Diverging from traditional financial institutions, LPs operate without a trusted third party, with all borrowing and lending activities automated through code. However, this distinction doesn't render LPs immune to financial risks. Given that LP users can easily initiate a loan and withdraw their deposits, concentrated loans and deposits can be a concern. Utilizing Aave protocol as a case study, I find that liquidity risk is very likely to exist, and both regular users (that repeatedly borrow and deposit cryptocurrencies) and large users (that contribute significant amount of loans and deposits) exert complicated influences on the protocol. Additionally, the study uncovers cross-LP effects between liquidity risk and market concentration in Aave, illustrating interconnections among prominent LPs.

Together, these four chapters offer a comprehensive overview of blockchain and DeFi and also novel insights into centralization in DeFi. Through empirical evidence on centralization in DeFi, the research demonstrates that specific users can serve as sources of centralization. Importantly, I also suggest that DeFi users face a trade-off between centralization and the market performance of DeFi.

• Item Type: Thesis (PhD)
• Qualification Level: Doctoral
• Subjects: H Social Sciences > HG Finance
• Colleges/Schools: College of Social Sciences > Adam Smith Business School
• Supervisor's Name: Stasinakis, Professor Charalampos and Sermpinis, Professor Georgios
• Date of Award: 2024
• Depositing User: Theses Team
• Unique ID: glathesis:2024-84406
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 26 Jun 2024 14:18
• Last Modified: 26 Jun 2024 14:23
• Thesis DOI: 10.5525/gla.thesis.84406
• URI: https://theses.gla.ac.uk/id/eprint/84406