Given the recent volatility on stablecoins, we wanted to give the community a heads-up that Gauntlet is in touch with community contributors and participants. Action could be required both from on-chain governance and Guardian via emergency because of bad debt risk. We will keep the community informed of recommendations.
Very important to have the community informed and updated.
The Chaos team is also analyzing and preparing a course of action. We look forward to working together to provide the community with clear recommendations.
Gauntlet Risk Analysis
For proactivity, given the unique situation, we provide Gauntlet’s recommendation below.
We recommend that all V2 and V3 markets should be paused temporarily.
Main scenarios that Gauntlet analyzed are the amount of insolvencies that could occur if:
- The price of USDC stabilizes,
- The price of USDC recovers, or
- The price of USDC drops significantly lower.
V3 emode assumes correlation of stablecoin assets, but at this time, those correlations have diverged. The risk has increased given that the liquidation bonus is only 1% for USDC on emode. To account for these assumptions that no longer remain true, we recommend pausing the markets.
At current prices, insolvencies are ~550k. These can change depending on the price trajectory and further depegs.
On-chain liquidity has also declined meaningfully, which is impeding liquidations. For example, this account was not liquidated when it was at a health factor below 1. Any liquidations on this account going forward will only prevent the account from recovering and provide no benefit to the protocol.
Below is the total value of the emode pools for Aave V3 pools:
Is there any good reason for the asymmetry across chains here? Are market makers just busier on other chains?
[Edit an hour later - looks like things have adjusted now]
From the technical side, we are actively coordinating with the risk providers regarding a course of action, especially on everything that requires the usage of functions under the control of the emergency admin roles on Aave v2 and v3 instances.
As an initial step, and given the Aave Guardian has the emergency capability to freeze (only stopping new deposits and borrowings) and set LTV to 0 on stablecoins on Aave v3 Avalanche, we think it could be a good measure, with the only risk of stopping refilling of collateral on volatile assets. This would be just temporary, as the freezing could be reverted relatively fast.
Why is this unique to Avalanche? Is stopping new deposits and borrows not enough there?
Setting LTV to 0 would only impact ability to take additional borrows and wouldn’t impact liquidation thresholds, correct?
Setting LTV to 0 definitely helps everywhere, but on the Avalanche v3 Pool, given that cross-chain infrastructure doesn’t cover Avalanche, the Aave Guardian can act immediately.
Setting LTV to 0 in practise discounts the “borrowing power” of the asset, without affecting the HF of any user position
While Gauntlet’s recommendation is to pause, as an interim next step the community can proactively freeze stablecoins and set LTV as zero.
Once the other risk provider of the community (@ChaosLabs) confirms these steps should only be positive from a risk perspective, we will coordinate with the Aave Guardian on Avalanche to execute this protective action.
For the sake of complete transparency, the payload to be used (freezing and setting LTV to 0) can be found here https://snowtrace.io/address/0x85fb6dff5f59695d77cc61898e60d040165b1923#code#F14#L1
We support the proposal of freezing all stablecoins on Avalanche by setting LTV to 0. In addition, we recommend pausing stablecoin markets across all Aave deployments as a temporary phase until USDC, DAI, and other stablecoin peg situations clarify.
At this point, we don’t recommend fully pausing supply and borrowing, mostly in order to allow existing borrowers to deposit collateral to maintain their positions and avoid liquidations.
We are monitoring liquidations and bad debt / insolvent positions on Aave through the Chaos Labs Collateral at Risk Dashboard. Most liquidations, bad debt, and under-collateralized positions are around stablecoins. The main risk vector we are taking into account is a massive in-flow of risky stablecoins into the protocol increasing bad debt. Therefore, our initial recommendation is to pause stablecoin markets on Aave across all deployments, because of the contagion effect as we have seen with DAI/USDC.
We are flagging the $9.5M in current under-collateralized positions which together hold over $800K of bad debt, mostly on Avalanche stablecoins through E-Mode. These positions are slowly clearing out through liquidations.
source: Chaos Labs Risk Explorer
Following the confirmation by both risk providers of the community, we can confirm the Aave Guardian has executed the freezing of USDC, USDT, DAI, FRAX, and MAI on Aave v3 Avalanche, together with setting their LTV to 0.
To insist: this doesn’t affect anyhow the current Health Factor of active positions, only disables providing more liquidity or borrowing on these assets, together with borrowing more against them
Aave Companies supports Aave DAO freezing USDC, USDT, DAI, FRAX, and MAI on Aave V3 Avalanche. This action prevents new risk from entering the V3 Avalanche market while users can continue to manage their positions in other assets as normal.
Contributors and Guardians have taken quick action to limit risk. We appreciate the Aave DAO’s quick action, conservative risk management and robust process to reduce risk, and we encourage continued analysis and stress testing across all Aave deployments.
Isnt this exactly what the price sentinal is supposed to handle?
To pause markets so drastically so soon after v3 launches when v3 was supposed to address what caused mass freeze/pausing after the curve squeeze sets a very bad precendent and harms the reputation of Aave v3 before it has had a chance to demonstrate what it is capable of.
I think precaution is correct here. And it only affects one market for now. Better safe than sorry. Would aave be way bigger and had a way bigger treasury I would say it’s not necessary but Aave is still young and needs to grow till it’s mature enough to handle every situation or can repay bad debt fully.
Quick update from the Chaos war room - we are conducting an ongoing analysis of the situation throughout the day, focused on assessing an alternative path forward to pausing all V2 and V3 markets.
Pausing all markets is clearly a robust risk-mitigating step regarding a momentary/temporary depeg of stablecoins. However, it may lead to bad debt from the inability to liquidate other (non-stablecoin) assets in a case of a broader market downturn. With SVB’s collapse and USDC’s exposure coming to the forefront of markets on Friday, we may see a spike in volatility across markets globally.
To this end, we’re conducting a holder and position analysis across V2 and V3 deployments and simulating different price drop trajectories across various asset classes. We aim to quantify potential loss for these scenarios, which will help provide a holistic recommendation on the path forward. We will share our conclusions after completing this research and further analysis.
Update: Gauntlet is in touch with community contributors. Under certain conditions, Gauntlet proposes pausing all stablecoins on Aave V3 markets that have emode (Arbitrum, Optimism, Avalanche, and Polygon). Most importantly Avalanche and Polygon should be paused if those situations occur.
We support freezing USDC, USDT, DAI, FRAX, and MAI on Aave V3 Avalanche and set the LTV to 0. This will minimize additional risks in case of further de-pegging, not only USDC.
We believed that as of now, we should not fully pause all the supply/borrow markets. We should do a risk assessment across all V2 and V3 deployments.
Thanks, @BristolBlockchain. To clarify - freezing those assets only mitigates risk against new positions, not existing ones, which are already sizable in e-mode markets.
Thank you for your clarification @Pauljlei. But we cannot control the existing position right? i.e. it’s based on the position’s owner to pay out the debt. Can you give your reasoning?
USDC Pause pros and cons
Our recommendation for all deployments currently is not to pause. With the peg restoring, we believe there is a higher risk of incurring bad debt due to non-stable (WBTC, WETH) volatility compared to the threat of a USDC/DAI depeg to 0.85 prices and below.
The inflection point of large bad debt is in the following scenarios:
Unpaused: USDC/DAI reaches 0.85 prices, causing over 2 million bad debt (and a sharp bad debt accumulation thereafter)
Paused Markets: Nonstables (BTC, WETH, etc.) price drops over 40%, causing over $2 million in bad debt (and sharp bad debt accumulation thereafter).
When assessing the situation, we’re addressing the following scenarios:
- USDC regains its peg.
- USDC keeps trading around the current discount level.
These scenarios may include a momentary depeg much more significant than the current level.
We are not accounting for a scenario in which USDC completely loses its peg for 2 major reasons:
- We don’t identify any strong evidence to support that
- This would encourage decreasing USDC and DAI exposure as fast as possible, even at the cost of losses. Given that we do not see this as a likely scenario. The damage here can be greater than the reward.
Pausing Assets - Pros and Cons
We have evaluated the possibility of pausing or freezing assets to mitigate protocol risk. The main risks are listed by order of significance, 1 being the riskiest and 5 being the least risky.
For AAVE V3 Deployments:
V → No new risk introduced, subject to market movements
X → New Risk exposure due to stable depeg or new reserve configuration
|Disabling stable collateral (LTV=0), Governance Vote Required||Freezing Stables, Governance Vote Required||Pausing Stables||Pausing All|
|1. Toxic In-Flow Risk - New Positions (net new USDC dumped into Aave)||V||V||V||V|
|2. Toxic In-Flow Risk - Existing Positions||X||X||V||V|
|3.a. Bad Debt from Liquidations of stablecoins||X||X||V||V|
|3.b. Bad Debt from other assets||V||V||V||X (liquidations cannot be made as a function of market movements)|
|4. Leverage Increase by Collateral Withdrawals||X||X||V||V|
|5. Leverage Increases by increasing borrows||X||X||X||V|
For AAVE V2 deployments:
|Disabling stable collateral (LTV=0)||Freezing Stables||Pausing All|
|1. Toxic In-Flow Risk - New Positions (net new USDC dumped into Aave)||V||V||V|
|2. Toxic In-Flow Risk - Existing Positions||X||X||V|
|3.a. Bad Debt from Liquidations of stablecoins||X||X||V|
|3.b. Bad Debt from other assets||V||V|
|4. Leverage Increase by Collateral Withdrawals||X||X||V|
|5. Leverage Increases by increasing borrows||X||X||V|
Markets Not Paused -Stablecoin Depeg (worst case - liquidation of entire collateral)
Markets Paused (Assuming re-peg) - Loss due to disabling Liquidations Incurred During Volatile Asset Price Crash
- Don’t pause currently; continue monitoring and updating as a function of market changes.
- Freeze stable markets, set LTV to zero
- Don’t pause currently; continue monitoring and updating as a function of market changes.
- Freeze Stables set LTV to zero.
It is important to note that pausing can be introduced via Admin functionality anytime. However, freezing markets requires a governance process and can take several days. Where possible (like Avalanche), we recommend freezing immediately.