Analyzing the solutions to the DeFi cascading liquidation problem from three dimensions

This article is an original piece by Chain Catcher, authored by Gu Yu.

The chain reaction of liquidations from DeFi lending protocols is one of the most significant systemic risks in the DeFi world. Due to excessive liquidation volumes and insufficient market liquidity, every time there is a drastic change in market conditions, it triggers further turmoil in the market, leading not only to additional losses for collateral borrowers but also exacerbating market volatility.

So, how should we alleviate this major systemic risk? Currently, the DeFi industry has seen a plethora of practices addressing this pain point. Chain Catcher will summarize these explorations in this article, primarily analyzing them from three dimensions: optimization of liquidation mechanisms, improving collateral replenishment efficiency, and restructuring lending logic.

1. Optimization of Liquidation Mechanisms

The liquidation mechanism and its algorithms are the core capabilities of lending protocols, directly determining whether users' collateral will be liquidated and the specific price at which liquidation occurs, ensuring that the platform and lenders do not incur losses.

In terms of liquidation mechanisms, due to several significant market events the industry has experienced in the past, the liquidation mechanisms of mainstream DeFi lending protocols have become relatively mature and stable. However, as seen in the massive liquidation event of over $100 million in XVS from the largest BSC lending protocol, Venus, in May this year, many emerging DeFi lending protocols still lack maturity in this regard.

Specifically, the liquidation mechanism can be roughly analyzed from three aspects: selection of collateral assets, collateral ratio settings, and price feeding mechanisms.

Regarding collateral assets, lending protocols generally need to select mainstream assets that have high liquidity, are relatively dispersed, and are not easily manipulated as collateral. Otherwise, it may cause significant abnormal losses to users borrowing related assets on the platform. The abnormal large liquidation events that occurred in BSC lending protocols in January and May this year were due to including CAN and XVS, two highly controlled assets, as collateral.

In terms of collateral ratios, lending protocols need to set relatively high collateral ratios to ensure that when liquidation occurs, the collateral assets can be liquidated at appropriate prices in the market, preventing situations where market depth is insufficient for liquidation, leading to platform losses.

For example, MakerDAO requires a minimum collateral ratio of 150% for ETH, while Compound requires a minimum collateral ratio of 133%. The minimum collateral ratios for other assets are often even higher. One of the main reasons for the massive XVS liquidation event at Venus in May was that the collateral ratio for XVS was set at only 125%, reflecting the platform's poor risk awareness.

Regarding price feeding mechanisms, many DeFi lending protocols use external asset prices as references for liquidation, such as Chainlink, UMA, and Coinbase. However, a single source with insufficient depth for price feeds is easily manipulated, leading to abnormal liquidation situations. Therefore, stable and diverse price sources are crucial for lending protocols.

On November 26, 2020, Compound experienced an abnormal liquidation volume exceeding $100 million, primarily due to an anomaly in the price feed source for DAI from Coinbase, which spiked to $1.34 in a short time, causing many positions borrowing DAI to fall below the required collateral ratio and subsequently be liquidated.

Currently, multiple emerging DeFi lending protocols are attempting to create a competitive space by leveraging second-tier assets that leading lending protocols are unwilling to support and lower collateral ratios. Although this can meet the needs of some users, it remains an unstable factor for the overall ecosystem of the project.

Another relatively unique optimization of the liquidation mechanism is Liquity's tiered liquidation model, which is also a good idea for alleviating the chain liquidation problem. Although this lending protocol has a low collateral ratio of 110% for ETH, which may increase the probability of liquidation, even if liquidation occurs, it will not create immediate selling pressure on ETH because the collateral will be tiered and proportionally allocated to the LQTY liquidity pool, allowing stakers to choose to sell or hold.

2. Improving Collateral Replenishment Efficiency

The root cause of collateral liquidation fundamentally lies in users' insufficient collateral. However, many users actually have the willingness and ability to replenish collateral. If most users can receive timely notifications about insufficient collateral ratios and replenish their collateral, the occurrence of chain liquidations under extreme market conditions would significantly decrease.

Therefore, many projects are practicing to improve collateral replenishment efficiency, with the main attempt being to provide automated liquidation protection functions, which automatically help users replenish collateral when their collateral ratios are insufficient, thus avoiding liquidation, while users do not need to constantly monitor market prices and can free up more energy.

For example, the DeFi aggregation platform DeFi Saver allows users to manage their debt positions on Maker, Aave, and Compound through the platform, while providing automated functions to help users automatically manage their collateralized debt positions. It allows users to automatically repay when their collateral ratio falls below a configured minimum value, using the assets currently deposited as collateral to pay off their debts, and automatically increase the loan amount when the ratio rises above a configured maximum value, allowing users to borrow more to obtain additional collateral assets.

Another DeFi aggregation platform, Instadapp, has a similar approach, launching Instadapp Actions to help users achieve automated refinancing of their debts, while also allowing users to autonomously move their debt positions between MakerDAO, Aave, and Compound based on the best liquidation prices.

It is worth noting that the service provided by Instadapp is launched in collaboration with Gelato Network, which is an Ethereum smart contract automation protocol that uses bots to automatically execute smart contract processes. The project has recently partnered with KeeperDAO to launch Instant Underwriter (JITU), which continuously monitors the health of users' positions opened in kCompound. If a position falls below a certain threshold, JITU will immediately take action on behalf of the user and provide additional collateral as a buffer.

In the aspect of improving collateral replenishment efficiency, another attempt is to help users quickly receive notifications about insufficient collateral ratios. Currently, DeFi lending users cannot directly receive notifications about their positions' insufficient collateral ratios, which also limits their timely actions to add collateral and improve position health. The main explorers in this area are EPNS and HAL.

EPNS aims to establish a decentralized notification layer on Ethereum to help DeFi users receive on-chain notifications through wallets, apps, etc., although the mainnet has not yet been launched. Currently, EPNS has reached pilot cooperation with lending protocols/derivatives protocols such as Aave, Alpha Finance, bZx, and UniLend, enabling platform users to subscribe to alerts for liquidation risks, declining loan health, and more.

HAL is less known than the former but was launched in June 2020 and currently supports users in subscribing to notifications from nearly ten leading DeFi protocols such as Aave and Compound, including loan health warning notifications, and supports sending notifications to users' email, Discord, Telegram, and Slack accounts, with over 4,500 users currently.

Some lending projects are also developing such services independently. For example, the BSC lending project Venus allows users to set warning notifications for their position's collateral ratios directly on their account homepage. Once the collateral ratio reaches the user-defined value, a Telegram bot will send notifications to the user's account, currently charging 0.06 BNB per month.

As more DeFi lending protocols integrate message notification services in the front end in the future, users may choose to subscribe to warning notifications when borrowing, which could significantly enhance their efficiency in replenishing collateral while avoiding further liquidation pressure on the market.

3. Restructuring DeFi Lending Logic

Currently, the vast majority of DeFi lending projects adopt an over-collateralization mechanism, which is the basis for liquidating collateral under drastic market fluctuations. However, in the past year, several projects based on credit loans/under-collateralized loans have emerged in the DeFi industry, reconstructing the DeFi lending business at the product logic level, using trust relationships between users or offline legal relationships as the foundation for building lending businesses, thereby reducing the market risks associated with liquidation.

Aave launched a credit authorization loan service last year, allowing deposit users with idle lending capacity to authorize their credit limits to addresses of users they trust. Both parties pre-agree through a legally binding smart contract protocol called Openlaw, allowing the latter to directly withdraw borrowed funds without collateral.

Another lending protocol, Union Finance, has a similar idea, where any user can set a credit limit and deadline for a specific address on the platform, transforming the risk of collateral liquidation into a trust relationship risk among friends.

ARCx's approach is to determine collateral ratios based on on-chain credit scores, which are calculated based on users' borrowing history, asset scale, governance experience, and other behaviors. The higher the score, the lower the collateral ratio can be (minimum 105%), allowing users to obtain the same loan amount with less collateral. Additionally, users can gain higher yield farming opportunities. This product aims to incentivize users to value their credit history through big data behavioral analysis and various benefits, encouraging users to repay under-collateralized loans, thereby reducing the potential scale of market liquidations.

Teller, incubated by a16z, also attempts to provide lending services based on credit, primarily by connecting users' bank accounts and determining loan conditions and amounts based on their historical records. Users with good credit can obtain unsecured loans or under-collateralized loans. If users default, Teller will reduce the user's score with the cooperating credit rating agency in a specific manner.

Truefi is also exploring unsecured loan solutions, mainly targeting institutional borrowers rather than retail users. Any borrower needs to undergo KYC and credit review. Additionally, each loan ultimately requires voting by platform token TRU holders. If a borrower defaults, the platform will liquidate up to 10% of TRU held by voting users and initiate legal action against the defaulter.

Physical collateral lending is another solution. Projects like Centrifuge and NAOS Finance are exploring bringing physical assets into the DeFi lending market and have several cases. The prices of physical assets are often more stable, and the diversification of collateral assets can also reduce the pressure of collateral liquidation on the cryptocurrency market.

According to data from Dune Analytics, the current outstanding loan scale in the DeFi market has reached $28.2 billion, more than three times the growth since the beginning of the year. At the same time, perpetual contract products based on lending logic are also rapidly developing, which means that the potential scale of liquidation in the DeFi market is significantly increasing the pressure on the secondary market, forming an important systemic risk to the cryptocurrency market.

The previous analysis addressed the issue of chain liquidation from three dimensions, optimizing the product logic of the lending market from different angles, and reducing the probability of on-chain liquidations. Notably, several projects with automated liquidation protection functions, such as Instadapp and DeFi Saver, have seen rapid increases in TVL recently. These explorations and data are somewhat helpful in enhancing the health of the DeFi lending market and reducing the volatility of the cryptocurrency market.

As for the dimension of restructuring lending logic, most of these projects are still in the early stages of development, with low user adoption and recognition, and still require further development.

Currently, the issue of chain liquidation is an important topic in the industry that will receive increasing attention, and the overall effectiveness of the aforementioned solutions will be tested in the next major market event.