Normalizing ETH Market Post-Merge

Proposal 122 has been passed to mitigate risk in the ETH market ahead of the Ethereum Merge. The proposal intends to protect against the risk of high ETH utilization that would interfere with liquidations of ETH collateralized positions.

We wanted to kick off community discussion around normalizing the ETH market post-merge. ETH borrowing demand is currently low on Compound - as such, there is limited downside for keeping Proposal 122’s borrow cap under current market conditions. As for the interest rate curves, it would be prudent to revert the interest rate changes when ETH utilization is back to a healthy state.

As market conditions evolve up to and after the merge, we will provide more details on the recommended precautions the community should take ahead of reverting any parameter changes from Proposal 122.

Next Steps

  • Monitor the ETH pool liquidity to evaluate the ETH interest rate curve rollback strategy
  • Monitor ETH borrowing to inform actions regarding the borrow cap
  • Welcome thoughts and feedback from the community
4 Likes

It would be strategically better to follow up with what was initially voiced if governance want to mantain healthy level of trust and just revert everything to a state it was before after merge execution.

It indeed looks like cap isn’t really a big deal for now. But it might be a bit premature judgement, and there is still a possibility of spike on a merge day. I am not big fan of caps, as protocol should be able to function without them, otherwise there is something wrong with design of protocol, if it needs them to exist. Manual intervention from multisig should be minimized.

3 Likes

Happy to see that the merge went through without a hitch! While Compound didn’t see as high of interest accrual as certain other defi lending protocols, I think being the only major protocol that maintained ETH liquidity through the entire process is a huge win.

Some thoughts on ideal parameter setting for the v2 ETH market going forward:

I think it makes sense to maintain the ETH borrow cap as is, or even to consider a somewhat lower level based on demand. ETH borrowings briefly hit the 100k ETH borrow cap, but have now declined back towards normal levels. The borrow cap provides a second layer of defense to users and the protocol against potential technical or asset specific risks that could drain borrowable ETH liquidity and collateral value from the protocol. In the long run, I think the best course of action would be to implement borrow cap automation similar to MakerDAO’s debt ceiling instant access module feature implementing a maximum borrow cap, as well as a target amount of available borrowable liquidity for each asset. This would help mitigate black swan risks without excessive governance overhead or impact on user experience.

I’d prefer to switch to a new jump rate model instead of reverting to previous straight line rate model. The old rate model topped out at around 15% borrow rate, which may not be enough to prevent full utilization of liquidity in stressed conditions. A jump rate model would allow for safety against illiquidity risk, while allowing for significant borrow usage (particularly if Compound v2 adds support for ETH staking derivatives such as stETH or rETH). The following parameters seem appropriate for a long term ETH rate model:

  • Borrow rate at 0% utilization: 0%
  • Kink: 80% utilization
  • Borrow rate at kink: 3% (or roughly 0-1% below expected ETH staking rewards rate)
  • Borrow rate at 100% utilization: 100%

I don’t think there’s any strong reason we should anchor to prior ETH market parameters, which had been unchanged since early days of Compound v2.

1 Like

Agree with @monet-supply that we should keep a jump rate model, since this was a basic advancement. Governance can also adjust rates without deploying a new contract, which reduces technical risks.

The suggested parameters look generally good, though I’d recommend setting a minimum borrow rate closer to 1-2% which is where it was before the merge. Raising the starting rate to 1%, and the kink to 4% (slightly above predicted rate), would ensure that the protocol is compensated for potential risk at low utilization rates, and that users will have ample liquidity to borrow.

3 Likes

We agree with these recommendations to change to a jump rate model since it will better address illiquidity risk moving forward.

In response to Sirokko’s concern - setting a robust long-term solution is in the best interest of the protocol and will naturally instill trust in governance.

1 Like

I’m not against kink-rate model at all, i was referring to original rates, rather than model itself, though we are speaking about quite abstract thing here. Utilization never reached even 50% in normal market situation, so what rate is going to do after 80% utilization does not really matter much.

Especially if sentiment is in leaving CAP intact, which will surely prevent utilization from ever reaching right end of curve at all. And please, don’t provide me with argument that multisig can adjust. It is very well known that multisig wouldn’t. It’s always slow to react, and it would never want to take responsibility in raising cap at higher than 80% utilization, so anybody can take it to the bank, that it’s never going to happen as long as borrow cap would be in place Thus rates for that section only matter when there is no cap on borrowing and we just doing theory crafting here.

But if we are talking about long-term solutions, than kink-rate model isn’t that great either. I would rather want to see 2 kink points to split rate curve in 3 sections, rather than 2. To have stimulative rate section, where utilisation rates are low, normalised portion, where utilization is somewhere in 20-70 percent utilization range, and restrictive, where utilization is high, with over 70% utilization.

We at Morpho Labs are generally against hard caps. Especially, as the borrow cap is reducing the borrowable liquidity, it is reducing by a lot the possibilities for integrators that borrows on Compound. Since this is something that seems to be already pretty rare, we think that everything should be done to encourage these use cases of the protocol, without compromising on risks of course.

For example, a “soft borrow cap” could be imagined, playing with interests rates to incentivize a target total borrow rather than prevent any borrow. If this is something that the community is interested in, we would be happy to work on a separated detailed proposal for this.

2 Likes

Hard caps have some risk advantages - if one of the other collateral assets on Compound failed catastrophically, the cETH borrow cap would limit maximum losses.

Overall, both parameter proposals from @rleshner and @monet-supply would increase ETH utilization. Gauntlet supports setting the minimum borrow rate at greater than 0% and the kink borrow rate at greater than or equal to 4% due to the following analysis:

  • Compound’s ETH borrow rate is roughly around ~4% post ETH merge. Understanding the market equilibrium takes time, but a 4% borrow rate at Kink would be a reasonable target based on recent market borrow rate performance post-merge. Setting the kink at or above current market borrow rate would be optimal from a risk perspective.

  • Setting the kink borrow rate higher than staking yield would ensure a lower likelihood of utilization exceeding the kink. This parameter setting will ensure sufficient liquidity and larger reserves for the ETH pool. The lower rate model suggested by @monet-supply might improve capital efficiency, but there is a risk tradeoff in which the ETH pool is more likely of high utilization. From a risk perspective, setting the kink at 4% or above better manages high utilization risk.

  • Given the success of the ETH merge, we anticipate the risk attributed to ETH staking to decrease while the demand for ETH borrowing increases. Given the increased demand to borrow ETH, it may not be necessary for Compound’s ETH market to require 0% borrow at 0% utilization to incentivize borrowing. Gauntlet supports >0% borrow rate at 0% utilization.

1 Like