E-Mode containers

[robert-zaremba]

Summary

Implement Aave E-Mode functionality.

Problem Definition

Today, when borrowing against a highly correlated asset, user has the same borrow limit as when borrowing against totally non correlated asset. For highly correlated assets we expect more efficiency. For example, when borrowing USDC against USDT, we don't expect much deviation, unless when some market situation will emerge.

Proposal

Aave implements an E-Mode containers.

E-Mode allows borrowers to restrict themselves in borrowing only assets belonging to a certain category (e.g., stablecoins). When this happens, if users supply assets of the same category as collateral, the borrowing power (LTV), and maintenance margin (liquidation threshold) are overridden by the E-Mode category configuration to allow higher capital efficiency.

E-Mode categories, enable a wave of new use cases such as:
• Highly efficient yield farming (for example, deposit ETH staking derivatives to borrow ETH)
• Diversified risk management

Solution

Create E-Mode containers. User will have to select when collaterizing tokens for an E-Mode container or when borrowing from the E-Mode container.

Each E-Mode container will have a set of lending parameters similar to token parameters from the factory. If a user makes an operation within an E-Mode container, the container parameters will be used instead of the token parameters,

Container composition as well as parameters will be subject to the governance.

Example

Let's say user want's to borrow ETH against sETH (some liquid staking of ETH). Normally he would would need to keep something like 130% collaterization ratio and the liquidation premium would be ~5%. However, when borrowing in an E-Mode ETH container, the collaterization ratio could be 105% and liquidation premium could be ~2%, making the supply more efficient.

Consequences

• we still need to think how this will impact the general supply
• for the final user it makes more sense to operate within E-Mode when possible

---

For Admin Use

• Not duplicate issue
• Appropriate labels applied
• Appropriate contributors tagged
• Contributor assigned/self-assigned