[ARC] ETH Interest Rate Curve Update

Governance
1

Title: [ARC] ETH Interest Rate Curve Update
Author: @Llamaxyz @MatthewGraham @Dydymoon
Dated: 2022-11-05

Simple Summary

@Llamaxyz presents a proposal to amend ETH interest rate parameters on the Aave Ethereum v2 and Ethereum v3 (when deployed) Liquidity Pools.

Abstract

With Aave v3 about to be deployed on Ethereum and a wstETH and ETH eMode likely, the ETH Interest Rate should be adjusted such that the looped recursive strategy breaks even within Slope 1 where the Interest Rate changes are less volatile.

The new interest rate curve has an Interest Rate of 5.75% at the Uoptimal point, which is 25 bps more than stETH yield at the time of writing. This proposal will lead to improved capital efficiency of the ETH Reserve and generate more revenue for the DAO when compared to just replicating the v2 Interest Rate curve on v3 with the introduction of an ETH eMode Category.

For the existing v2 market, @llamaxyz suggests adopting the same Interest Rate parameters as being proposed on v3. This will enable the ETH yields across the markets to somewhat equalize leading to higher utilization of the ETH Reserve and more revenue for the DAO.

Motivation

By changing the Interest Rate at the Uopotimal point from 3% to 5.75%, when an eMode for ETH and wstETH is created on Aave v3 the recursive strategy will break even prior to reaching the Uoptimal utilization within the Reserve. This is a desirable feature as the volatility of the Interest Rate on Slope 1 is less than Slope 2. By setting the Slope 1 parameter at 5.75%, 25 bps greater than the current annual wstETH yield, utilization in the Reserve is not expected to reach the Uoptimal level for any prolonged period of time. By reducing the Slope 2 parameter, from 100 to 80, the gradient of the second part of the yield curve increases due to an increase in Uoptimal from 70% to 80%. This is subtle and can be seen in the chart below. The Base Rate for the existing Interest Rate and proposed Interest Rate is unchanged at 0%.

With the new Interest Rate curve recommended for the v3 deployment, it is reasonable to expect the yield across the v2 and v3 deployment to be arbitraged. If the yield on the v3 deployment converges to the wstETH rate, then this will shift the v2 Reserve utilization beyond the Uoptimal level set at 3%. Because of this, it is recommended to also change the v2 Interest Rate to match the v3 deployment Interest Rate. This way both v2 and v3 Reserves will experience utilization along the Slope 1 portion of the Interest Rate curve whilst also benefiting from improved capital efficiency.

The graphic below shows the changes in the interest rate.

This proposal increases the Reserve Factor from 10% to 15%, directing 15% of the interest paid to Aave DAO. When combined with the Interest Rate changes, the v2 deployment will generate an estimated $13.1k per day, up from around $5.5k per day. Prior to this proposal, due the implementation of AIP-68, the Aave DAO is earning less revenue from the ETH Reserve post merge despite utilization within the Reserve increasing significantly over that time, [2].

Specification

The below table shows the current and proposed changes to the ETH Reserve on Ethereum v2 and Ethereum v3 when deployed.

Parameter Current (%) Proposed (%)
Uoptimal 70 80
Base 0 0
Slope1 3.0 5.75
Slope2 100 80
Reserve Factor 10 15

References

[1] Aave - Open Source Liquidity Protocol
[2] Optimize rate curve for ETH & improve revenue generation
[3] Creative Commons — CC0 1.0 Universal

Copyright

Copyright and related rights waived via CC0, [3].

5 Likes
2

I am in favour of increasing the Slope 1 and increasing the reserve factor collections, it makes sense given the arbitrage that exists and rewards depositors more fairly for the use of their tokens as leverage.

However the Uoptimal measure is a risk metric and you have not given any risk analysis supporting the change or showing the impacts. I don’t think we can simply look at increased revenue for the DAO in insolation.

The DAO is already the largest holder of stETH with >20% of total supply, more supply in fact than of wETH. Do we want to encourage a further increase on this? We need to remember that despite the merge, stETH is not yet redeemable 1-2-1 for ETH, and the same dynamics that lead to the depeg earlier in the year still apply.

We can also not ignore how concentrated aWETH holders are, with the top 3 addresses controlling >25% of the supply. With higher slope1 curve and Uoptimal rate, there is very little room if any of these actors pull liquidity, which could lead to stETH unwinds and depegs.

Maybe I am wrong, but would want to see some risk analysis supporting the changes. I think we still need to be cautious here until stETH is truely liquid and depeg risk is minimised, especially given v2 doesn’t have the ability to implement supply caps.

3 Likes
3

Thanks, @bayesiangame for your thoughts and analysis. Gauntlet is analyzing the potential market risk implications of this proposal and will return to the forums.

1 Like
4

Gauntlet has analyzed Llama’s interest rate curve adjustments for wETH:

  • wETH’s Top 5 suppliers constitute 30% of liquidity within the pool. Concentration risk for the wETH asset is not as highly concentrated as other assets within the v2 market. If the top 5 suppliers withdrew their 30% supply of wETH and borrowing remains constant, wETH utilization would increase to 75% and still leave ample liquidity for further liquidations or withdrawals.

  • The top 10 borrowers of wETH are utilizing stETH as collateral in order to enhance returns with a leveraged liquid staking strategy. Demand for borrowing wETH will continue to be driven by wETH/stETH recursive yield.

Top 10 Supply Positions for top wETH Borrows

  • Post the Merge, the average wETH pool utilization is 51%. Utilization rate has not been relatively close to the kink since the Merge. The average borrower variable interest has been 2.19% since the Merge.

  • As noted by @bayesiangame, Aave’s stETH collateral position represents 19.5% of stETH circulating token supply in the market. Concerns of stETH depeg have subsided since the ETH merge, but the risk remains until the Shanghai update.

Our systems observe ample liquidity in the stETH/wETH market. The slippage/liquidation curve suggests that a $220mm market sell of stETH through 1inch only incurs 1.5% slippage, implying a liquid market.

Gauntlet currently has no recommended adjustments to Llama’s interest rate recommendations. We also compliment the @Llamaxyz team for following best practices here and waiting for risk analysis before submitting a vote.

6 Likes
5

Hi @Pauljlei,

Thank you for sharing Gauntlet’s insights on the forum and supporting the proposal.

We have proceeded to create a Snapshot for the community to vote on with voting to start in a little over 24hrs.

We hope everyone can participate in this vote and we thank all participants in advance for taking the time to vote.

4 Likes
6

Hi Everyone :wave:

During a review of this proposal it was identified that the following parameters for Stable Borrowing should also be amended in line with the changes to the Variable Borrowing rates.

The AIP is almost ready for submission. But if the community feels a new Snapshot is needed, we are happy to schedule one for next week. If there are no comments, we intend to submit the AIP on Monday 12th December.

Parameter Current Variable Borrowing (%) Proposed Variable Borrowing (%) Current Stable Borrowing (%) Proposed Stable Borrowing (%)
Uoptimal 70 80 70 80
Base 0 0 3 3
Slope1 3 5.75 7 7
Slope2 100 80 100 80
Reserve Factor 10 15 10 15
2 Likes
7

Hi @everyone :wave:

The [ARC] ETH Interest Rate Curve Update AIP has been submitted. Voting starts tomorrow.

Thank you in advance to all those who participate in the vote.

3 Likes
8

I really appreciate that Llama and Gauntlet essentially worked together on this proposal. Beyond coordinating in the forums, can either Llama or Gauntlet describe how they worked together on this outside of the forum and how they normally work together? I’d just like to be assured that the process is working well, and I’m also wondering if other risk service providers like Chaos are getting a chance to contribute. Thanks.

2 Likes
9

Thank you! We work closely with many Aave service providers and contributors including Gauntlet, Chaos Labs, BGD, and others. We worked with Gauntlet on this interest rate curve update proposal, while we’ve worked with Chaos Labs on proposals such as AIP-125 to update risk parameters for Aave v2 Ethereum.

3 Likes
10

Hi Everyone :wave:

A gentle reminder, when v3 is deployed on Ethereum, the following ETH interest rate curve is to be implemented reflecting the community snapshot linked above:

Parameter Variable Borrowing (%) Stable Borrowing (%)
Uoptimal 80 80
Base 0 3
Slope1 5.75 7
Slope2 80 80
Reserve Factor 15 15

With stETH currently yielding 4.8%, @Llamaxyz will likely present a proposal to reduce the Slope1 parameter to 5.0%, or a slightly different premium depending what other LSDs, like rETH, are added once v3 is deployed on Ethereum.

Ideally, the Slope1 parameter is around 20bps higher than the highest LSD yield to ensure the Reserve’s utilization is less than the Uoptimal parameter.

11

llama has ACI support to lower slope 1 & 2 % on ETH.

Current parameters makes wETH/stETH less competitive and Aave is leaking liquidity for competition.

We’re even supportive of a more aggressive change to place the slope1 threshold at 4.8% but won’t fight if that figure is 5%.

Let’s keep in mind V3 has emode and this is an essential strategic feature for stETH/ETH trade on V3, with shanghai confirmed to be on testnet in Feb & very high odds of being mainnet in March, LSDs pegs will be more and more ressilient.

2 Likes
12

Hi @MarcZeller,

Changing the Slope1 parameter to 4.8% would be equaling the current Lido yield. This is the lowest the Slope1 should be changed to with only product fees applied to structured products keeping utilisation under the Uoptimal value.

Do you have any thoughts on the Base parameter ?
We like the idea of introducing a small Base fee which will flatten the gradient up to Uoptimal utilization.