Researchers affiliated with Canada’s central bank identified weak points in DeFi lending protocols and reported on the potential they saw for mitigating them.
The Bank of Canada has released a working paper that examines lending protocols in decentralized finance with regard to sources of instability and their relation to crypto asset prices. Its findings point to potential ways to optimize DeFi lending platforms, or possibly the practical limits of decentralization.
The authors of the paper, titled “On the Fragility of DeFi Lending” and released Feb. 22, acknowledge the inclusiveness DeFi offers and the advantages of smart contract protocols over the use of human discretion — but they also identify the systemic weaknesses of DeFi. Information asymmetry, a key issue for regulators, is highlighted, with the twist that in DeFi, the asymmetry favors the borrower:
“The collateral composition of a lending pool is not readily observable, implying that borrowers are better informed about collateral quality than lenders are.”
This is because borrowers are at least aware of the quality of the assets they used as loan collateral. Moreover, “Only tokenized assets can be pledged as collateral, and such assets tend to exhibit very high price volatilities.” Price and liquidity produce a feedback loop, the paper argues, saying that the price of an asset affects borrowing volume, which, in turn, affects asset price.
In addition, smart contracts’ lack of human input can have undesired effects. Traditional loan contracts can be modified by loan officers in response to current information. However, smart contracts are inflexible because terms are preprogrammed and “can only be contingent on a small set of quantifiable, real-time data,” and even minor changes to the contract can require a lengthy discussion process.
“As a result, DeFi lending typically involves linear, non-recourse debt contracts that feature over-collateralization as the only risk control.”
Efficiency, complexity and flexibility are thus reduced in comparison with traditional finance, and “self-fulfilling sentiment-driven cycles” of pricing arise. The authors used advanced mathematics to examine a number of propositions for achieving market equilibrium in those circumstances.
A flexible optimal debt limit was found to provide equilibrium. However, “simple linear haircut rules” typically designed into smart contracts cannot implement a flexible limit. It would be hard to create protocols with that feature, and they would be highly dependent on the choice of oracles. Alternatively to that challenge, “DeFi lending could abandon complete decentralization and re-introduce human intervention to provide real-time risk management.”
Thus, the authors conclude, the DeFi trilemma of decentralization, simplicity and stability remains unconquered.