Enable USDT as collateral on Aave v3 on Avalanche

Thank you for this proposal, @luigidemeo88 and @Ava_Labs. Below, we provide Gauntlet’s analysis.

Gauntlet’s risk assessments focus on market risk. Notably, the main risks that USDT poses to Aave (if it is enabled as collateral) is not market risk but rather other risks that are highly difficult to quantify (e.g., proof of reserves, counterparty risk, etc.). The community must consider their risk preferences when assessing the risk/reward tradeoffs of enabling USDT as collateral outside of isolation mode. To clarify - USDT is currently enabled as collateral on V3 Avalanche but in isolation mode.

To help inform the community of the risk/reward tradeoffs, here are several potential risks should USDT be enabled as collateral:

  • If there is a depeg of USDT, this can cause significant liquidation cascades depending on factors including the size of USDT usage on Aave and user collateralization ratios. These liquidation cascades can lead to insolvency that are losses for the protocol.
  • The reason why the UST depeg did not meaningfully impact Aave is precisely because UST was not enabled as collateral. Because UST was not a collateral asset on Aave, there were no liquidation cascades impacting Aave.
  • In times of depeg uncertainty, there may be strong demand to borrow USDT in order to short it. This presents a risk to Aave because high utilization of USDT can interfere with atomic liquidations of positions that supply USDT as collateral. Again, the quantified risk depends on many factors, including 1) the size of USDT collateral usage and user collateralization ratios and 2) if the users have supplied other assets as collateral. We can imagine a scenario where users supply USDT as collateral and the borrow asset is a volatile asset, so if atomic liquidations are interfered with, then the position can go directly to insolvency based on price movements (borrow amount in $ > USDT in $). As a result of high utilization preventing liquidations, this means that even if there is not a realized depeg event, there is still a risk to Aave as long as there is a perceived chance of a depeg that traders capitalize on by borrowing USDT. Although a high borrow interest rate may help prevent this, Gauntlet has observed that traders may continue to borrow the asset, and the borrow interest rate must exceed their perceived risk-adjusted return to be effective. We note that borrow utilization of USDT on V3 Avalanche is currently high at 93%, showing that indeed high utilization can be a possibility. Supply and borrow caps may also be a solution, but they are not guaranteed to prevent 100% utilization and need to be dynamically adjusted (which may be too slow given governance constraints).
  • Isolation mode significantly reduces the risk profile of USDT. First, the contagion is more limited because users who supply USDT as collateral cannot supply other assets as collateral in the same position. This has very nuanced risk impacts. In the scenario where an asset is not isolated, then a USDT depeg can have a meaningfully higher chance of liquidation cascade because of the interactions with other assets. This is because liquidators may not seize the USDT collateral (because they want no exposure to USDT) and instead seize other collateral assets and sell them, which can instigate liquidation cascades of those assets. In addition, we can limit the total exposure by setting the debt ceiling. At the same time, being in isolation mode may meaningfully reduce USDT’s usefulness, given that users can only borrow stables in isolation mode.
  • We note, however, that if USDT depegs, the market would have broader impacts.

None of this is to say that Gauntlet has quantified the risk of depeg. Gauntlet cannot quantify the likelihood that USDT depegs as it depends on factors outside market risk. We simply provide more context to the community on the risks that USDT as collateral would pose to Aave under the scenarios outlined above.

5 Likes