[ARFC] Aave Instances Strategy Shift
Author: @ACI
Date: 2024-11-25
Summary
This ARFC seeks governance feedback and approval to deprecate the Etherfi mainnet instance and reformat the Instances model in the Ethereum Mainnet Aave ecosystem. Previous TEMP CHECK passed the Snapshot
Motivation
Aave 3.2 introduced liquid E-modes, creating a new universe of options and efficiency improvements for the Aave ecosystem.
The Aave Instances vision, created & pushed by the ACI, is a massive success with the Lido instance that recently surpassed a billion-dollar market size.
That being said, the rationale of this instance was to isolate use cases to avoid friction and undesirable asset interactions within the same instance. This came at the cost of liquidity fragmentation, but when we initiated this vision, it was the most straightforward implementation option.
Aave 3.2 was a paradigm shift because Liquid E-modes allow us to isolate use cases without having to fragment liquidity. A suitable example of this is in the Lido instance. Without liquid E-mode, ezETH & wstETH would be competing for wETH liquidity, creating friction & sub-optimal interactions.
With the current setup, ezETH contributes to wstETH supply yield and therefore indirectly contributes to wETH supply yield. This created synergies instead of tensions by isolating inside liquid E-modes which assets are collateral for specific borrowing assets.
With Aave 3.2, it doesn’t make sense anymore to have many different instances as all these assets can coexist in a synergistic way inside the same instance and contribute to the same liquidity, with value added to each other.
For these reasons, we suggest a merge of Etherfi & Lido instances and further refinement of the 3.2 E-Mode strategies. EtherFi’s instance was meant to focus on LRT deposits with stablecoin borrowing interactions. But with 3.2, just like we implemented with ezETH, we can allow weETH to borrow wstETH and stablecoins via two segregated liquid E-modes inside the “Lido” instance, having the best of both worlds in the same pool.
Another important thing to consider is that Aave protocol revenue is heavily concentrated in a few assets. The money makers in Aave are: wBTC (and growing BTC-correlated assets), the ETH-correlated assets (wstETH, weETH & wETH), USDC, USDT & GHO. Everything else has minor contributions to the Aave DAO revenue.
Every asset that is used as collateral in Aave increases the potential for a shortfall event. It’s crucial we think as a DAO about the risk/reward of each collateral onboarding with consideration for potential revenue generation.
Until the launch of Aave V4, we do not have an option to impose a risk premium to borrow some assets using some collateral, so we are currently in a situation where we can create no premium and no added revenue to use lower demand assets when compared with blue-chip collateral.
As Aave DAO revenue is a volume-at-scale profit model, we end up with little to no revenue with some assets as collateral.
The Aave DAO earns the reserve factor on interest paid by borrowers, from the current market borrow rates. Using USDC as an example, the DAO earned an average of 35-50 bps per year on volume generated.
For some small collaterals in the main instance, actual revenue doesn’t cross 5 figures, which is a bad deal when considering the risk associated with any collateral onboarding, regardless of isolation mode, caps, and exposure. At the Aave scale, no amount of risk is worth less than 5 figures of revenue.
This is intrinsic with the Aave model of earning only at scale as interest paid by borrowers is largely sent back to LPs. The exception to this is the GHO stablecoin where the protocol itself is the LP and earns all interest paid by borrowers.
Still, having the same cost to borrow GHO using wETH as collateral compared to an emergent asset doesn’t make economic sense from the perspective of the DAO.
Despite these economic realities, it is important to be able to onboard new assets inside the Aave protocol to avoid them gaining traction in Aave alternatives. We also witness some frustration from ecosystem actors that consider Aave “too elitist”.
A solution to this is to create a new instance that we propose to name the “Emergence” instance. This instance will only onboard assets in isolation mode with the sole ability to borrow GHO. GHO will be onboarded as a 0% LTV asset with an interest curve higher than GHO on the main Aave instance.
This will have several benefits:
- Create a de-facto risk premium on GHO with some collaterals, adjusting reward in terms of DAO revenue with risk taken by protocol
- Create a path to onboard emerging assets in the Aave ecosystem, allow to impose goals and KPIs on these collaterals (i.e., it must generate $X of volume over Y days) before consideration for onboarding on the main instance
- Allow for quick and relatively painless offboarding if a collateral performs below DAO expectations. If KPIs are not met, we can set LTV to zero, reduce caps, decrease LT, and offboard in a short time period
- If an asset does hit KPIs, it would have a strong case to be supported & onboarded in the Main instance
In terms of risk management, it’s likely current GHO stakers might not be keen on being the counterparties of an emergence market. This TEMP CHECK seeks community feedback on the creation of a “Junior Tranche” stkGHO for stakers willing to earn more risk in exchange for more rewards.
As the Borrow rate of GHO is set higher than on the main instance, the junior tranche StkGHO (or jtStkGHO) can be financed by the borrow rate premium from the Emergence instance.
Using the upcoming Umbrella program, Risk and Aave service providers can define a target liquidity to protect GHO in case one of the collaterals on the Emergence instance experiences a failure and shortfall event.
Lastly, to reflect on risk, we propose to not provide any stkAAVE discount for this instance.
To clarify this new set of instances, we propose the following naming conventions:
- The Lido instance will become the “Prime” instance, focused on blue-chip collaterals and high leverage correlated assets use cases
- The main instance will become the “Core” instance, it’s meant to stay the most liquid and best risk-adjusted yield instance for LPs and borrows with more diversity of options
- The new instance will be the “Emergence” instance, meant to expand in a controlled and suitable risk/reward way the family of assets supported by the Aave protocol. Successful assets in Emergence are meant to eventually be onboarded in “Core”
- The Junior tranche stkGHO is named “jtStkGHO”
Specification
This AIP seeks to get feedback on:
- Freezing and offboarding of the “Etherfi” Instance
- Rebranding and merge of current “Etherfi” instance to the current “Lido” instance, to be rebranded as “Prime” instance
- Rebranding of “Main” instance into “Core” instance
- Freeze and offboard smaller collaterals from the “Core” instance
- Deployment of “Emergence” instance, with GHO onboarding or new facilitator creation and some of the offboarded assets of Core instance (to be defined during ARFC phase)
- Definition of a KPI framework for assets in Emergence instance
Disclaimer
ACI did not receive compensation for creation of this proposal.
Next Steps
- Publication of a standard ARFC, collect community & service providers feedback before escalating proposal to ARFC snapshot stage.
- If the ARFC snapshot outcome is YAE, publish the necessary AIPs for final confirmation and enforcement of the proposal.
Copyright
Copyright and related rights waived under CC0