[ARFC] wETH & wstETH Interest Rate Curve - Ethereum Network


title: [ARFC] wETH & wstETH Interest Rate Curve - Ethereum Network
Author: @Llamaxyz @MatthewGraham
Dated: 2023-01-16


Simple Summary

@Llamaxyz presents a proposal to amend wETH and wstETH interest rate parameters on the v2 and v3 deployments.

Abstract

After discussions with @bgdlabs, @MarcZeller, and @ChaosLabs, this proposal presents a revised wETH Interest Rate curve for v2 and v3, plus the initial conservative wstETH Interest Rate parameters for v3.

The recently updated wETH parameters on v2 have led to improved revenue generation but also a gradual decline in TVL and daily revenue over time. This proposal intends to implement the learnings from the v2 Ethereum deployment into the initial v3 Interest Rate curves whilst also updating the v2 deployment wETH interest rate to be the same across both deployments.

Motivation

wETH Reserve Aave v2

Since AIP 131 was implemented, the amount of stETH deposited as collateral to borrow wETH has been fairly consistent. However, the amount of wETH borrowed by users who deposited stETH as collateral has reduced from 515.21K to 413.74K between 18th December 2022 and 15th January 2023. The tapering of borrowing demand is reflected in the wETH nominated revenue.

The market is indicating that there is insufficient demand for borrowing wETH with the current Slope1 parameters for wETH. Three possible reasons are shown below:

  • Structured products charge a fee
  • Additional smart contract risk
  • Elevated exposure to DEX liquidity

Our analysis indicates the Slope1, 5.75%, is most likely the primary causes of the reduced wETH demand. Lowering the Slope1 parameter will enable users to achieve a higher return from entering the recursive loop strategy. The chart below shows how users that deposit stETH and borrow wETH have behaved to date. The date highlighted below is the date that AIP131 was executed.

We propose reducing both the Slope1 parameters on the v2 market to 3.8% and introduce a Base of 1%. The graphic below shows the current and proposed wETH interest rate curves. Introducing a Base reduces the gradient between 0% < Utilization < Uoptimal. However, borrowing wETH becomes slightly more expensive at lower utilization.

The chart below shows the relative change in borrowing costs for users, comparing the current versus the proposed interest rate curve. When utilization is less than 41% borrowing cost are higher, ranging 0 to 0.98%, and when utilization is greater than 41% borrowing costs are lower, ranging 0 to 0.95%. Current utilization is 46% on the v2 deployment which means the borrowing costs would be reduced.

wstETH Reserve Aave v3

Due to the wstETH and wETH recursive strategy, we propose deploying the wstETH Reserve with a fairly flat Interest Rate curve.

  • Base: 0.25%
  • Slope1: 4.5%
  • Slope2: 80%
  • Optimal: 45%
  • Reserve Factor (RF): 15%

Having a Base fee of 0.25% leads to a low pool utilization, generating a higher deposit rate than it otherwise would. This then provides additional yield for the leverage recursive strategy which will hopefully lead to greater wETH utilization.

Due to the recursive strategy, Slope1 and Slope2 parameters for wstETH shall remain lower than the majority of other assets on Aave. A material yield on wstETH deposits will provide a meaningful source of yield for the recursive strategy. Lower borrowing costs for wstETH leads to greater utilization of the wstETH and wETH liquidity, whilst the wETH interest rate largely determines the profitability of the recursive strategy. A notable non-incentivised deposit yield on wstETH is desirable for maximizing revenue from the wETH and wstETH Reserve.

A lower Slope1 parameter of 4.5% encourages users to borrow wstETH without generating a high deposit wstETH yield. Any lower and we may see a lot of wstETH borrowing due to the various incentive programs for wstETH across DeFi. ie: 6% on the wETH/wstETH Balancer v2 pool, noting user receives at best 50% of the wstETH yield. With Slope1 set at 4.5%, we are expecting users will only borrow wstETH if there is a reasonable return for the added smart contract risk.

A RF of 15% further reduces the yield to wstETH deposits. This acts to reduce wstETH deposit rates, however still passes on a meaningful amount of yield to depositors. Setting Slope2 at 80% is intended to not cause any rapid spikes in wstETH deposit rates. Fine tuning the RF, Slope1, and Slope2 parameters for revenue and risk will be an ongoing exercise.

Post v3 deployment, these variables can be changed. We will continue to monitor the borrowing demand for wstETH and wstETH deposit rate as the BorrowCap is increased over time and how it affects the recursive wstETH/wETH strategy. The Uoptimal, set at 45%, will likely need to be increased in time when the BorrowingCap is higher.

The initial 3k wstETH BorrowCap is a key variable for Aave to monitor. If both the BorrowCap and SupplyCap were reached, the utilization of the Reserve would be 1.5% and the yield to depositors would be 0.005%. If the recursive strategy utilizes 6 loops, then depositing wstETH would contribute an additional 0.03% to the strategy. The user would still have wETH borrow costs to consider. The wstETH deposit rate acts to counter the wETH borrowing rate within the recursive strategy.

The table below shows the wstETH deposit yield for the following utilization assumptions:

  • A portion of the SupplyCap (200K)
  • 100% of the BorrowCap (3k)
SupplyCap Utilization Deposit Yield (%) 6x Deposit Yield (%)
25% 0.043 0.26
50% 0.014 0.08
75% 0.008 0.05
100% 0.005 0.03

The below chart shows the proposed wstETH interest rate.

wETH Reserve Aave v3

As the initial 3k wstETH BorrowingCap being heavily restrictive, there is less immediate need to amend the wETH interest rate on v3 relative to v2. As the wstETH BorrowingCap and Uoptimal increases, the wETH interest rate’s Slope1 parameter may need to be increased.

With a desire to migrate liquidity from v2 to v3, we propose implementing the same Interest Rate curve on v3 as is proposed above for v2. Whilst the wstETH deposit rate is not yet a meaningful yield source for the recursive strategy, due to the restrictive BorrowCap, there is no immediate need to deploy v3 with a more conservative interest rate than v2.

Specification

The below table shows the current and proposed wETH interest rate curve for the Ethereum v2 and v3 deployment.

Parameter Current (%) Proposed (%)
Uoptimal 80.00 80.00
Base 0.00 1.00
Slope1 5.75 3.80
Slope2 80.00 80.00
Reserve Factor 15.00 15.00

The below table shows the proposed wstETH interest rate curve for the Ethereum v3 deployment.

Parameter Current (%)
Uoptimal 45.00
Base 0.25
Slope1 4.50
Slope2 80.00
Reserve Factor 15.00

Copyright

Copyright and related rights waived via CC0.

2 Likes

This proposal has ACI support.

2 Likes

From the technical side regarding the upcoming activation of Aave v3 Ethereum, we think it is reasonable to include as initial rate strategies for WETH and wstETH the ones suggested by @Llamaxyz, the rationale is the following:

  1. The modifications compared with the current WETH rate strategy on v2 are not too aggressive, mainly only reducing the Base and Slope1 parameters.
  2. The rate strategy configuration, even if important, is a quite dynamic component of the protocol: it has been designed to be adjusted continuously, following liquidity-driven considerations, like the ones explained in this thread. This means the community can just iterate over it just after release.
  3. The presence of borrowable wstETH is new for the protocol, so there is no reference for its rate strategy. This proposal introduces it, which is better than just using a default rate strategy for other volatile assets, especially given that there is a quite tight correlation between WETH and wstETH usage mechanics.
  4. Including this into the activation governance proposal doesn’t really add voting overhead, meaning we don’t identify any NO vote on the activation exclusively because of the rate strategies.
  5. If further steps via Snapshot are required, that would mean further activation delay.

It is especially important for the community to give feedback on this. For now, we will proceed with the final testing of Aave v3 Ethereum assuming the proposed rate strategies for WETH and wstETH.

4 Likes

Gauntlet Recommendation

tl;dr: The smart contracts for V3 wETH Interest Rate Strategy have incorporated the parameters that Llama has outlined in this post. However, these parameters from Llama are incorrect and do not represent the intent of their post, as shown by their charts and commentary.




The recommended strategy is confusing - the terminology used in the OP post is not correct according to how Aave docs/contracts define the variables. We are using the terminology as indicated in the Aave docs and interest rate curve strategy contract. Given the terminology in the Aave docs, Llama’s recommendation does not achieve the effect of lowering Borrow Interest Rates, though we are not clear on the full intent given the confusion in terminology.

If Llama intended the Slope1 parameter to represent the IR value at the Kink (which would seem to be the case based on their wstETH graph), that does not match the Slope1 parameter and should not be put into the live contracts.

All our analysis below uses the terminology as defined in the docs.

Analysis of WETH IR Curves

Llama justified their proposed WETH interest rate curve by suggesting that it would increase borrower demand by lowering the borrower interest rate, but the proposed curve actually increases the interest rate that borrowers pay at every level of utilization. The graph below shows the interest rates for suppliers and borrowers under the current and proposed interest rate curves, as the graph provided by Llama is incorrect:

The difference between the current and proposed curves is shown below:

As you can see above at every point in the curve, there is an increase in Interest Rates. This was not the intention of Llama’s post, given their charts.

When we noticed this we were concerned that this would launch with V3 based on the forum post here and markdown + contracts here. We recommend changing this ahead of V3 ETH launch.

Our recommendation

To summarize: We strongly recommend leaving the WETH curves as is on V2 and similarly setting the WETH curves on V3 to what is currently live on V2. We would be amenable to further adjusting these curves post-launch once we can agree on the correct settings for the interest rate curves and the community has more time to review.

Specifically, the below (using the terminology in the Aave contracts):

Parameter Current (%)
Uoptimal 80.00
Base 0.00
Slope1 5.75
Slope2 80.00
Reserve Factor 15.00

Analysis of wstETH IR Curves

wstETH is not currently a listed asset, but there are several active proposals to list it to various markets. It is similar to stETH, which has been listed on Ethereum v2 since last Spring, but since borrowing has always been disabled for stETH, we have no data about how users respond to changes in interest rates. We agree with Llama’s strategy of starting with conservative parameters that can be tweaked later. We would note, however, that the graph in Llama’s post is incorrect according to Aave contracts terminology (though that doesn’t change too much on the IR curve).

The proposed optimal utilization of 45% is in line with the more volatile assets available on Aave (1INCH, AAVE, CRV, ENS, LINK, MKR, UNI). The maximum borrower interest rate (base + slope 1 + slope 2) that comes from these parameters is 87%, which is rather low: the closest points of comparison would be the stablecoins BUSD (104%), FRAX (79%), LUSD (79%), and TUSD (79%), as well as WETH (85.75% currently, 85.8% proposed). While there is a greater chance of utilization getting dangerously high when the maximum interest rates are this low, the use of borrow caps in Aave v3 markets can reduce the risks associated with this. We will monitor utilization following the asset listing of wstETH and may propose an increase in slope 2 depending on how the market evolves.

As a minor aside, the GH markdown for V3 has a small typo on the IR strategy for wstETH. BaseVariableBorrowRate should read 0.25% instead of 2.5%, however, the contract itself is correct.

1 Like

Hi @Pauljlei,

Thank you for the feedback.

We have amended the forum post, correcting the Slope1 parameter to 3.8% down from 4.8%. The borrow rate at the Upotimal point for wETH is 1% + 3.8% = 4.8% or Base + Slope1 = Borrow Rate. We also took this opportunity to amend some of the text to more clearly communicate the proposed changes.

Lowering the Slope1 for wETH addresses feedback raised by communities who support products that utilise the recursive wstETH/wETH strategy. The current Slope1 parameter, set at 5.75%, was leading to less wETH borrowing over time. Reducing the Slope1 parameter from 5.75% to a comparable 4.8% yield is more supportive of these products. Introducing a Base on wETH, whilst having a lower Slope1 parameter, acts to flatten the Interest Rate curve. This is advantageous to users of the recursive strategy as it softens the rate of change in borrowing costs. However, it does lead to higher borrowing costs compared to the current Interest Rate curve when utilization is less than 41%. Current the utilization is 46%, therefore the recursive strategy generate a higher return.

We prefer keeping v3 and v2 wETH rates the same across both markets and using wstETH parameters to encourage migration.

A small Base fee was introduced on wstETH, which leads to a very small deposit rate on low utilization. A small deposit yield is advantageous to the recursive strategy and is unique to v3, but not really effective at low utilization. When the BorrowCap on wstETH is increased, we expect to increase the Uoptimal parameter and potentially reduce Slope1 subject to the new Uoptimal parameter. Directionally, if any further amendments are made, we are of the opinion it should be to reduce wstETH borrowing costs and increase available liquidity in the Reserve. This leads to higher utilization of both Reserves but is dependent upon borrowing demand for wstETH.

After @Llamaxyz correction on Slope 1, and given that the parameters are still sound and can be adjusted in the following days, we will proceed with these for the initial Aave v3 Ethereum configuration.

4 Likes

Hi Everyone :wave:

A Snapshot vote has been created that will start tomorrow 30th January for upgrading the Ethereum v2 wETH Interest Rate curve. The Snapshot vote is specific to Ethereum v2 and is inline with the original proposal.

https://snapshot.org/#/aave.eth/proposal/0x9ae28e9c82c5fc0d24cf1df788094e959d99f906d11b89e455a60ee16b071d6f

Thank you to everyone in advance who participates in the vote.