Funding Liquidity Risk in Decentralized Lending: An Empirical Investigation from a Financial Intermediation Perspective

Decentralised Lending (DeFi lending) is a new concept in finance. Based on blockchain technology and smart contracts, the innovative design of DeFi lending allows pseudonymous participants to lend and borrow money on a large scale with less intervention from financial intermediaries. Within the framework of financial intermediation theory, DeFi lending is an evolving landscape with the potential to reshape financial services, though it remains a frontier market with numerous challenges. DeFi lending could potentially perform functions similar to those of traditional financial intermediaries like banks in liquidity transformation; however, it also faces a significant problem known as funding liquidity risk. This risk within one DeFi protocol can lead to systemic liquidity risk contagion, potentially affecting the stability of the broader DeFi ecosystem. Unlike traditional banks, which are established centralised financial institutions (CeFi), DeFi operates in an unregulated market and must address illiquidity issues without government bailouts or deposit insurance. Furthermore, it is considered a potential source of financial instability due to its increasing integration with traditional financial products. Compared to CeFi, DeFi features a distinct financial network with greater interconnectedness, characterised by its composability. Composability refers to the ability to create a complex financial system using various components built on crypto assets, akin to ‘Money Lego.’ This feature can enhance interoperability and liquidity transformation while potentially increasing risk contagion. Existing literature indicates that DeFi lending is an emerging and understudied area, with most research to date being conceptual or based on aggregate data. This thesis investigates the dynamics of funding liquidity risk contagion in DeFi lending by using high-frequency, transaction-level blockchain data at 5-minute and 1-hour intervals. This research aims to explore issues related to funding liquidity risk in DeFi lending by examining two main research questions. First, it investigates the level of contagion of this risk within the distinct financial network of DeFi lending and examines the external factors that drive this contagion. The study examines Aave, Compound, and Venus protocols, which collectively account for approximately 70% of the total value locked in the DeFi lending market. Second, as the source of risk contagion often originates from one lending protocol and spreads outward, the study delves deeper into DeFi lending protocols by examining the determinants that affect funding liquidity risk within those protocols to better understand the source of this risk. The study focuses more on internal factors, which can be adjusted through governance and protocol design, to assess how they influence funding liquidity risk. Understanding these issues will deepen our comprehension of the potential challenges and opportunities in DeFi lending, thereby offering valuable insights to improve market efficiency and stability. The findings reveal that the average level of funding liquidity risk contagion in DeFi lending is relatively low compared to CeFi, but it varies over time based on market conditions. High spillover risks tend to occur during bullish or bubble markets, while they are significantly lower in bear markets. Furthermore, crypto policy uncertainty have a substantial impact on the level of risk contagion. The thesis also finds that current algorithmic interest rate models are ineffective as self-stabilisation mechanisms in major pools such as Wrapped Bitcoin (WBTC) and Wrapped Ethereum (WETH). Additionally, lower deposit concentration in these pools may exacerbate, rather than mitigate, funding liquidity risks.