[ARC] Aave Ethereum V3 market initial onboarded assets

Aave Ethereum V3 market initial onboarded assets3


There are currently two different proposals to select the onboarded assets on Aave Ethereum V3 markets on deployments.

while both proposals have their own merits, the aave governance is invited to select their preferred option, as both proposals are in the forum since long enough to be considered mature for snapshot vote phase, a vote will be published today with a start 24h in the future.

The vote will be between the following options:

  1. Chaos Labs Proposal - full details on the methodology and proposal can be found here
Symbol Isolation Mode Borrowable Collateral Enabled LTV LT LB RF LPF Debt Ceiling Supply Cap Borrow Cap
WBTC NO YES YES 70% 75% 6.25% 20% 0.10 N/A 43K 28K
WETH NO YES YES 80% 82.5% 5% 15% 0.10 N/A 1.8M 1.4M
wstETH NO YES YES 68.5% 79.5% 7% N/A 0.10 N/A 200K 3K
USDC NO YES YES 74% 76% 4.5% 10% 0.20 N/A 1.76B 1.58B
DAI NO YES YES 64% 77% 4% 10% 0.20 N/A 338M 271M
LINK NO YES YES 50% 65% 7.5% 20% 0.10 N/A 24M 13M
AAVE NO NO YES 60% 70% 7.5% N/A 0.10 N/A 1.85M 0

IR curve parameters are to be set to the current Ethereum v2 configurations.

In addition, we recommend configuring an Ethereum-correlated (wstETH & ETH) e-mode category with the following parameters:

Category Assets included LTV LT Liquidation Bonus
Ethereum Correlated WETH, wstETH 90% 93% 1%

Gauntlet Proposal

  • full details on the methodology and proposal can be found here
Symbol Isolation Mode Borrowable Collateral Enabled LTV LT LB RF LPF Debt Ceiling Supply Cap Borrow Cap Supply Cap USD Borrow Cap USD
USDC NO YES YES 8,650 8,850 10,480 1,000 0.20 0 2,000,000,000 591,700,000 $1,918,303,914 $591,631,355
WETH NO YES YES 8,250 8,600 10,520 1,000 0.10 0 600,000 469,000 $690,915,520 $552,732,416
WSTETH NO YES YES 7,200 8,300 10,720 1,000 0.10 0 600,000 94,000 $682,967,764 $106,998,283
WBTC NO YES YES 7,200 8,200 10,520 2,000 0.10 0 40,000 4,600 $594,982,530 $75,440,126
USDT YES YES YES 7,500 8,100 10,530 1,000 0.20 10,000,000 680,000,000 303,000,000 $678,186,530 $302,976,250
DAI NO YES YES 7,650 8,900 10,430 1,000 0.20 0 320,000,000 114,400,000 $316,922,850 $114,357,765
LINK YES YES YES 6,950 8,250 10,720 2,000 0.10 25,059,000 32,000,000 2,650,000 $183,541,004 $15,482,976
CRV YES YES YES 5,500 6,100 10,830 2,000 0.10 20,929,000 140,000,000 25,000,000 $67,634,449 $12,986,109
AAVE NO NO YES 6,600 7,300 10,770 0 0.10 0 3,000,000 0 $155,435,804 $0
MKR YES YES YES 6,500 7,000 10,770 2,000 0.10 4,592,000 60,000 1,100 $30,906,375 $579,653
TUSD YES YES YES 8,000 8,250 10,520 1,000 0.10 4,451,000 24,000,000 7,180,000 $23,507,765 $7,178,272
FRAX NO YES NO 0 0 0 2,000 0.00 0 29,000,000 10,900,000 $28,397,588 $10,844,134
BUSD NO YES NO 0 0 0 1,000 0.00 0 47,000,000 12,800,000 $46,961,233 $12,774,788
YFI YES YES YES 5,000 6,500 10,770 2,000 0.10 16,187,000 10,000 200 $24,279,888 $1,082,947
UNI YES YES YES 6,500 7,700 10,930 2,000 0.10 17,112,000 6,000,000 437,000 $28,204,849 $2,245,127
ZRX YES YES YES 5,500 6,500 10,770 2,000 0.10 6,062,000 79,000,000 933,000 $12,609,873 $149,189
MANA YES YES YES 6,150 7,500 10,770 3,500 0.10 6,869,000 34,000,000 1,980,000 $10,302,960 $612,605
1INCH YES YES YES 4,000 5,000 10,880 2,000 0.10 14,331,000 56,000,000 1,100,000 $21,496,425 $426,397
DPI YES YES YES 6,500 7,000 10,770 2,000 0.10 118,000 50,000 4,500 $2,350,324 $257,971
SNX YES YES YES 4,900 6,500 10,770 3,500 0.10 8,641,000 11,000,000 3,600,000 $16,753,891 $5,641,686
CVX YES YES YES 3,500 4,500 10,880 2,000 0.10 3,367,000 2,000,000 257,000 $5,049,037 $868,088
BAT YES YES YES 6,500 7,000 10,770 2,000 0.10 4,898,000 41,000,000 2,100,000 $7,345,858 $377,538
SUSD NO YES NO 0 0 0 2,000 0.00 0 5,000,000 1,850,000 $4,693,138 $1,868,455
BAL YES YES YES 6,500 7,000 10,830 2,000 0.10 3,390,000 1,000,000 185,000 $5,084,701 $980,370
LUSD NO YES NO 0 0 0 1,000 0.00 0 3,000,000 1,210,000 $2,355,611 $1,229,852
ENS YES YES YES 5,000 6,000 10,830 2,000 0.10 12,663,000 2,000,000 40,000 $18,993,939 $442,272

TLDR

Gauntlet Proposal:

  • Recommendations based on information and insights from simulations Gauntlet conducted on current Aave V2 ETH and Aave V3 AVAX
  • Full list of assets as previously voted on.

Chaos Labs Proposal:

  • More conservative, risk-off, and guarded launch - recommendations based on current market conditions rather than parameters configured on v2 and other v3 deployments.

note:
a third option was proposed by ACI in the gauntlet proposal thread, that being said, the two current options are good enough to move forward with voting and integrating a third option will mainly just add noise & delays to this process, the ACI retract their proposal and invite the community to pick one of the current option via their vote.

19 Likes

Thanks very much, @MarcZeller and the ACI, for consolidating the different threads and putting this into a format where the community can compare the two proposals.

Below, we wanted to summarize and provide transparency on the differences between the two proposals and methodologies, as well as outline some tradeoffs.


  1. LTV/LT setting
  • WETH, WBTC: Our LTV/LT methodology draws from simulation results for V2 ETH and V3 AVAX, and we emphasize the use of caps to initialize a more guarded launch. As we understand it, Chaos’s methodology for LTV/LT relies on other V3 deployments agnostic to different chains.
  • Stablecoins - Chaos’s recommendations for LT/LTV are derived from a weighted average of the debt against them (when those volatile assets are used as collateral). The idea, as we understand it, is to reflect the volatility of the debt.
    • From our research, this should not be an apples-to-apples comparison because the primary market risks of stablecoin collateral are the upwards price movement of non-stablecoin debt. Under historical usage, the majority of protocol usage is WETH collateral and stablecoin debt. Therefore the LT and LTV of WETH are more heavily weighted towards the risks associated with the volatility of downward price movements, not its upwards price potential. The distribution of positive price returns and negative price returns is not symmetric.
    • Ultimately, this methodology can have sampling bias since we do not know the true distribution of debt that stablecoins support. This methodology takes a snapshot of the debt against the stables at this point in time, and ignores the natural variance in the debt that these stables support (the composition of the volatile assets that are borrowed against the stablecoins change over time).
  1. Number of assets and isolation mode
  • Gauntlet’s recommendations provide more assets upon initial launch. This reduces the governance burden of adding additional assets. Isolation mode assets are also included, which again reduces the governance burden of re-identifying isolation mode assets in the future, which can slow down the V2→V3 migration. With caps in place (and isolation mode as needed), this mitigates risk while allowing Aave V3 ETH to immediately have a more diverse set of assets.
  1. e-mode recommendations
  • We do not recommend the 1% liquidation bonus for wETH/wstETH. A high liquidation bonus can be risky because it pushes accounts closer to insolvency, but a low liquidation bonus is also risky because it needs to provide enough incentive to liquidators to close risky positions. Given 1) the MEV space is highly competitive, 2) liquidators need to take into account gas costs to guarantee prices, front running when they acquire the debt side to repay positions, 3) flashloan and other fees, the 1% incentive can be dangerously low.

To clarify, Gauntlet’s recommendations take into account both current market conditions as well as stressed market periods (via simulation).

Overall Tradeoffs

  • Chaos’s proposed LTs are generally lower than Gauntlet’s. As part of Chaos’s proposed plans to encourage V2-V3 migration, a potential path is to lower LTs across the board on V2. While certainly reasonable, this strategy requires many things to work in order to be successful
    • Having lower LTV and LT on V3 will reduce the capital efficiency of V3 relative to V2, which is a disincentive for users to migrate to V3. Potential solutions are lowering V2 LTs, and additional incentives on V3. Reducing Aave V2 LTs may result in just losing TVL in general instead of promoting migration.
    • Of course, the community may elect to embark on these strategies, but it is valuable to acknowledge their complexity and ambiguity.
  • Gauntlet’s approach is different in that we rely more heavily on supply/borrow caps for a conservative launch instead of relying on lower LTs. This approach avoids the problems (disincentives and complexity) outlined above. From a methodology perspective, LTV and LT values for V2 and V3 should be similar, as market risks in both deployments are driven primarily by the liquidity of supported assets.

We hope that this helps provide clarity on the tradeoffs for the community. We couldn’t be more excited about the launch of Aave V3 on Ethereum.

1 Like

Hi @MarcZeller,

Given GHO is to be launch on Ethereum v3, I think we need to understand the mechanisms supporting GHO and how it will interact with each Asset Reserve. There is no documentation on GHO and no mechanism design modelling shared with the community, so I can only comment based upon what is heard on the grape vine, which could be very wrong.

From what I understand, GHO can be minted against any collateral asset on Aave v3 and there is only two parameters that are governance controlled which is the Interest Rate and the global amount of GHO which can be minted per Aave deployment. From what I understand there is no way to control how much GHO can be minted against each Asset Reserve. Therefore, GHO can be backed 100% by high volatile assets, it is possible, just not likely.

In a risk off event, the correlation across asset prices converges to 1. This means all higher risk assets tend to trade the same way. During these risk off events, there is a rush into stable coins and even redeemable stable coins trade above $1. This was definitely the case when the Luna crash happened, USDC, DAI and USDT all traded above $1. It is likely GHO will trade the same way.

Let’s consider a tail risk scenario… What happens when the GHO mint limit is reached, or is close to it, and there is a market crash where GHO trades above $1 and there are a lot of long tail assets backing GHO which collapse in price. This seems like a recipe for a greater chance of bad debts emerging as the stable coin trades above the redemption price, gas is typically high, governance token liquidity often contracts or DEX pool become overweight the high risk asset leading to higher price impacts on swaps. This all happens around about the same time which compounds the issues. This might be offset by a higher liquidation reward but the CRV event shows us there is still risk of bad debt despite a functioning liquidation process. This scenario is a by product of not being able to control which assets can back GHO which could be considered a design flaw/trade-off in how GHO integrates to v3.

It is hard to tell if the above is true when there is no public docs on GHO and no mechanism modelling shared with the community to fully understand the limitations / risks presented by GHO given the current design limitations of Aave v3. How easy would it be to upgrade Aave v3 to better accommodate GHO before deploying it on Ethereum. Seems like now is the right time to discuss how GHO interacts with v3 and then discuss if we upgrade v3 in production or do it now when there are no user funds in the contracts.

From what I can tell, the only way Aave can deploy GHO safely on v3, is to heavily restrict which assets are listed as collateral and then use SupplyCap to risk how much collateral can be deposited. There is no way to prevent any asset from being used to mint GHO and to my knowledge there is no way to introduce a GHO Reserve and then only enable GHO borrowing by certain types of collateral. It seems the current design of v3 is rather restrictive and sub optimal given the trade-offs with supporting GHO. Without upgrading v3 to have better risk controls for GHO, it seems the only logical way forward is to heavily restrict what assets are onboarded to v3 on Ethereum.

3 Likes

We would like to clarify a few of the points made in the comment by @pauliej

Our recommended parameters for the launch are indeed conservative. The V2 LT and LTV configurations (and the basis for the Gauntlet V3 proposal) are generally set high and do not reflect the current market conditions, as proven in the recent CRV attack. Migration to V3 is our highest priority, as stated with our recent V2 coverage extension AIP. However, we don’t believe that the path towards that is setting aggressive risk parameters in V3 to optimize for capital efficiency. Currently, Aave has risk debt, as parameters in V2 were not updated as market conditions deteriorated. It will take a few months to reduce the LTs in V2 responsibly and incentivize a migration to V3. We urge the community not to launch V3 on Ethereum with risk debt from day zero and take on unnecessary risk exposure.

We can always be more aggressive with these parameters if market conditions improve. Scaling them back is a much larger challenge and can be quantified in months of work and prolonged risk exposure, as seen in V2.

We’ve provided an analysis (which can be found at the bottom of our original post) to support the recommended configurations of LT, LTV, and LP for the e-mode category, taking into account observed stETH deviations. Given these parameters, alongside the recommended supply and borrow caps, we feel comfortable with the prospect of profitable liquidations.

As we’ve said in previous posts - we are open to hearing opposing recommendations and considerations. However, lacking data to support claims like the ones above leads to “hand-wavy” conversations that are challenging to manage.

We know that the distribution of positive and negative price returns is not symmetric. However, as the volatility of upward price movement is less severe than that of downward price movement, using this heuristic leads to more conservative LT and LTV recommendations as intended.

As we wrote in our original post, we have devised this interim methodology for the initial launch risk parameters. Under current conditions, our risk-reward tradeoff leans heavily toward risk aversion; therefore, our primary motivation with the interim methodology was to provide a risk-off strategy for the community to opine on. Optimizing for risk-averse parameters while producing recommendations promptly has intentionally yielded less capital-efficient results. Once the simulation framework is complete and published for review, we will iterate on the recommendations to produce more capital-efficient risk parameter recommendations while keeping protocol risk at acceptable levels.

Our analysis has taken into consideration recent historical data, not just a single snapshot. Indeed, it is not guaranteed to remain so, but this is true for any protocol measure. Risk management and parameter recommendations are ongoing processes, and parameters such as these are monitored over time and modified when needed.

Our recommendation is to have a 2-phased approach regarding the assets initially listed. It’s important to emphasize - per our understanding with core contributors - GHO is just a normal V3 asset. It doesn’t benefit from any extra feature, such as only being mintable by some assets or debt ceilings. The only way to restrict collaterals would be with isolation mode. Therefore, the limited assets in our proposal are by design. We want to support a safeguarded GHO launch, ensure high-quality collateral for GHO minting, and allow more time for a community discussion regarding the long-tail assets to be listed, given considerations (GHO launch, economic viability) that were not taken into account in the original Snapshot.

5 Likes

the snapshot vote for this ARC has passed with a majority selecting @ChaosLabs proposal.

V3 will be deployed with the phase I Assets as following:

Note: Quorum of 320k YAE vote was not reached due to a bug in snapshot platform preventing some delegates from voting but as shown in the snapshot below provided by @lbsblockchain their vote of 80k AAVE + the 283k AAVE that was able to vote is superior to the 320k AAVE quorum required.

We can thus consider this snapshot vote canon and quorum achieved without the need of a re-run.

7 Likes

With this blocker resolved, we have published an update of the release items for Aave v3 Ethereum on BGD. Aave v3 Ethereum. New deployment vs Aave v2 Upgrade - #16 by bgdlabs

With the activation really close, we appreciate any last-minute feedback from the community

5 Likes