The Risk Team has reviewed this proposal. It seems important to look at the current positioning of YFI within Aave as well as Aave’s risk appetite which are misaligned with the ideas proposed.
YFI current Positioning in Aave
There is $126m YFI on Aave V1 and V2 representing 2.4% of liquidity and 2.2% of TVL with around $40m of borrows. Only variable borrowing is available for YFI, with a 30d average borrow rate of 1.25% on V1 (over 30% utilisation) and 0.80% on V2.
Aave Risk Framework places YFI at a B risk level with the highest risk being volatility at C-. The risk parameters are:
- V1, LTV 40%, LT 65%, LB 15%
- V2, LTV 40%, LT 55%, LB 15% and RF 20%
In parallel Aave’s stablecoin borrowing demand is strong, leading to high 30d average variable borrow rates driven by utilisation above the optimal point:
- USDT—V1 30.53%—V2 19.90%
- sUSD—V1 22.22%----V2 22.61%
- USDC—V1 15.19%—V2 16.24%
- DAI-------V1 7.78%----V2 17.79%
- TUSD----V1 7.76%----V2 35.49%
- BUSD----V1 6.38%----V2 17.38
- GUSD---------------------V2 11.83%
- V2 borrow rate average is around 20% while the borrow rate at optimal utilisation is just 4%; showing a very high market demand for stablecoins.
Now let’s dive in the proposal:
1. Disable YFI as a borrowing asset
Borrowing of assets is the number one source of income for the Aave ecosystem as well as the only ecosystem collector source of income from Aave V2. It’s not only a source of income but also of governance power collection.
Disabling YFI borrowing would result in nearly $30k of weekly income loss. YFI is still mostly used as collateral so it would continue to be leveraged by borrowing against, resulting in an income in the corresponding borrow assets. There is already a very high borrowing demand for stablecoins, this could add $15m of borrowing.
Governance attacks have been a concern for Aave, as described in @Jordan’s brainstorm post.
Furthermore, disabling borrowing would enable aToken voting, hereby increasing the functionalities of aTokens. However, there is currently $26m of borrowed aYFI which would result in double counting.
Overall disabling YFI borrowing has a negative impact on Aave’s revenue and governance power with the loss of YFI’s borrowing market. This would be accompanied by some user experience improvements of aToken voting, yet some assets are already borrowed so more though needs to be put into implementing it.
2. Special Borrowing Terms
It’s surprising to see the proposed parameters are completely out of the range of Aave’s risk appetite. It would be interesting to understand the rationale behind this parameterisation.
Gauntlet Network has already identified YFI as the riskiest asset for Aave, while the full report is imminent, @wfu share a primer on the forum. In a preliminary analysis focused on Aave V1, Gauntlet drilled into the Aave Protocol to understand the market risks of each assets and wether the risk parameters are appropriate. Gauntlet stress test projections show that YFI (used as collateral) would result in the most insolvency value in case of crisis. This is due to a combination of YFI market metrics and Aave’s YFI depositor behavior who are riskier than average. As a result, V2 YFI’s risk parameters were adjusted to be more conservative in line with Gauntlet’s recommendations.
Minimum Deposit 1000 YFI
This proposal minimum deposit corresponds to $35m or 31.5% of YFI currently deposited on Aave. This results in increased concentration risk on YFI with a significant increase in potential liquidation volume resulting in higher slippage which affects liquidators profitability. Combined with larger market price impact during liquidations which may cause cascading liquidations. These two factors reinforce each other through a recursive process threatening the Aave Protocol of systemic failure. YFI has already been identified as the riskiest asset, with higher exposures, the risks increase further.
YFI borrowing limits
The proposed risk parameters for are completely out of the range of both versions of Aave:
- LTV 65% corresponds to YFI’s V1 Liquidation Threshold. This would require to increase the risk adjusted borrowing power of V2 by over 50%
- Liquidation Threshold 85% corresponds to the highest liquidation threshold in Aave, allocated to the safest asset USDC with a volatility risk at A+ while YFI level is C-. This would also require to increase the V2 threshold by over 50%
- Liquidation Bonus 5%, 3 times less than the current bonus. The analysis previously pointed that the increased YFI volume would result in higher slippage for liquidators; reducing the liquidation bonus will results in a further reduction in liquidation incentives; already negatively affected by network congestion. This means liquidators will have less incentives to liquidate YFI collateral positions with higher risks of undercollateralisation.
Fixed borrow rate
The Aave Protocol relies on the interest rate model to manage liquidity risk transferring some of this risk to borrowers. When utilisation goes up, the borrow rates go up to incentivize the repayment of loans and compensate depositors who’s liquidity risk increases as there is less liquidity to withdraw. The stable rate is a solution that offers more stability, whilst still offering risk mitigation possibilities in extreme utilisation cases.
A fixed borrow rate, ignores Aave’s risk management strategy. It’s not clear which assets will be borrowed, but most likely stablecoins which are already under high demand with the highest need to mitigate liquidity risk.
In addition to generating significant additional risk, the requested 3% would also result in a significant discount and revenue loss for depositors, being 17% lower in absolute than the current 30d average borrow rates of V2 stablecoins.
Given the algorithmic nature of the protocol, as it is, this cannot be implemented.
3. Native Credit Delegation to Yearn Vaults
There is indeed some excess credit capacity on Aave with capital that could be allocated for more efficiency. Facilitating credit delegation for Aave users could enable them to optimise their strategies. This can easily be built on top of Aave without the need to change the smart contracts. What are the advantages of modifying the source code?
Yearn is asking to modify Aave’s source to facilitate operations in exchange of revenue share from the Credit Delegation - are there some projections and estimations?
Aave is open to exploring potential interplay with decentralized protocols for collaborations that are beneficial for both. The whitelisting process will be its own governance vote by the $AAVE community.
It is not clear how this proposal might benefit Aave, on the contrary, points 2 and 3, leads to a significant increase in risk without compensating rewards:
- Increased systemic risk - albeit an increase in TVL
- Increased liquidity risk
- Reduced income and influence of Aave (no YFI borrowing, borrow rate at -17% of the protocol’s rate)