Gauntlet recommendation for WETH Uopt on Ethereum v3

[Gauntlet]

Summary

AIP-306 to onboard sDAI also increased WETH Uopt from 80% to 90%. After analysis, Gauntlet recommends returning WETH Uopt to 80%. Currently, there may be insufficient WETH to facilitate liquidations during increased volatility - this proposal aims to double the available WETH for liquidations should utilization for WETH be at Uopt = kink.

Recommendations

We present two options for the community, option 1 to lower Uopt to 80%, option 2 to lower Uopt to 85%. We include how the borrow APR will change under the current slope 1 parameterization, as well as our proposal to lower slope 1 to 3.3% here -

Option	New Uopt	Current Utilization	New Borrow APR (slope 1 = 3.8%)	New Borrow APR (proposed slope 1 = 3.3%)
1	0.8	0.83	0.1580	0.1530
2	0.85	0.83	0.0371	0.0322

Analysis

Comparison of the riskiest WETH collateralized positions with the riskiest USDC collateralized positions - another asset with 90% Uopt - shows there can be excess risk involved with setting WETH Uopt to 90%.

The below chart shows the top 10% riskiest WETH positions have LTV >= 0.68 (~$55m), compared to LTV >= 0.6 for the top 10% riskiest USDC positions (~$27m).

For the above riskiest positions, WETH also collateralizes more uncorrelated borrows (stablecoins, ~86%) relative to USDC (WETH/WSTETH/etc, ~50%). Uncorrelated borrows increases the chances of liquidation.

As a result, compared to USDC, 90% Uopt for WETH increases the chances that the available capacity for liquidations will be consumed during a significant market move, at which point WETH utilization may be 100%, preventing additional liquidations.

The above shows that WETH moves can be milder (riskiest positions have higher LTV) and eat up available liquidation capacity faster (borrows are less correlated) compared to USDC.

Next steps

Welcome community feedback and aim to put up snapshot with the above two options on 2023-09-11.

[MarcZeller]

The ACI stands firmly against these recommendations.

uOptimal has been optimized (pun intended) for growth and competitiveness reasons.

AIP 306 enforcement has been more than successful.

wETH liquidity for liquidations can be sourced by MEV bots from other venues such as Balancer,
additionally wETH borrow cost at 83.8% at 100% utilization will increase deposits and incentivize repays.

Concerns seem unfounded and risk-reward is in favor of keeping the status quo.

[Gauntlet]

Hey @MarcZeller thanks for the feedback here. To reiterate, our reasoning to return WETH Uopt back to 80% is because of potential 100% WETH utilization risks associated with 90% Uopt (which may inhibit healthy liquidations) during sudden increased volatility, similar to what happened on 2023.08.17.

Based on the analysis above, a ~23% drop in WETH triggers $60m in liquidations. Should WETH utilization be at 90% Uopt, that means at current WETH supply, only $52m can be seized for liquidation. This suggests that utilization will reach 100% and prevent healthy liquidations from going through. It is unclear how fast WETH repayment / new deposits can occur during these sudden shocks to mitigate those risks involved with 100% utilization.

On the other hand for USDC, 90% Uopt poses less of these risks. A 23% increase in WETH/WBTC price only triggers ~$11m in USDC liquidations. USDC supply is $280m, so USDC utilization at 90% Uopt implies that $28m can be seized for liquidations, which sufficiently covers these liquidations.

Let us know if there are any areas of this logic and analysis that you disagree with, always happy to discuss further.

[Gauntlet]

We put up the snapshot for the WETH Uopt updates here. Voting begins in 1 day and lasts for 3 days.

[Gauntlet]

We provide more color to the options presented in the snapshot here. Based on the current loan-book, our simulations show that a 30% drop in WETH price as seen on June 10-13 2022 would yield ~$900k bad debt for 90% Uopt, ~$500k for 85% Uopt, $0k bad debt for 80% Uopt due to lack of liquidation capacity.

At the point in which 80% Uopt starts accumulating bad debt (-37% market drop), 85% Uopt has accrued ~$2.4m bad debt and 90% Uopt has accrued ~$3m bad debt.

In order to adopt a more conservative outlook, the above assumes no WETH repayment. The borrow rate at 100% utilization is ~84%. In times of extreme volatility, it is unclear how quickly WETH repayment would occur.

From a revenue perspective, relative to 80% Uopt, 90% Uopt may facilitate an additional $55m, which equates to ~$290k additional revenue per year. 85% Uopt may facilitate an additional ~$145k per year.

[Michigan_Blockchain]

It seems during the 2023.8.17 volatility, WETH utilization did not increase. Is that because the price drop is not big enough, or that borrowers started paying back their debt?

Is there any evidence of the speed of debt repayment based on past episodes of volatility, eg: June 10-13 2022, etc?

It also seems this proposal somewhat offset the simultaneous proposal to decrease WETH slope 1 to 3.3%, which is meant to increase utilization. For the simulation above post, does it matter whether slope 1 is 3.3% or 3.8%?

[Gauntlet]

Thanks for the community’s participation in the snapshot. Gauntlet continues to recommend returning WETH Uopt to 80% because there may be insufficient WETH to facilitate liquidations during increased volatility should utilization for WETH be at Uopt = kink at 90%, but given the community’s preference in option 3, no change to WETH’s 90% Uopt, we will not proceed forward with this proposal.