ARC: Re-enabling ETH Borrowing post-Merge

As AIP-97 has passed, pausing ETH borrowing into the Merge, it is now time to consider the opposite action: re-enabling ETH borrowing post-Merge. One of the main goals of pausing ETH borrowing was to mitigate the insolvency risk caused by liquidation failures due to high ETH utilization. To safely re-enable ETH borrowing, we recommend taking the following minimal set of precautions before executing the re-enablement of the ETH market. These features should be considered necessary but not sufficient conditions for the re-enablement of ETH borrowing.

  • ETH utilization returns to a healthy state (e.g. < 90%)
  • Prices tendered by ETH oracles from Chainlink have returned to their prior distribution
    • This involves performing statistical tests on the distribution of prices before and after the Merge. While we expect there to be some disruption to oracle prices within some time window around the precise Merge block at the target difficulty, this should subside relatively quickly. Given ETH2’s short epoch finalization time window of 12.8 minutes (see this blog post from client developer Prysmatic Labs for more information)
  • No other unexpected events occur during the Merge
    • e.g., Liquidity on on-chain and off-chain exchanges return to a stable, locally stationary state within < 1 day of the Merge

However, we also want to propose a set of further monitoring conditions that are important to qualify the security of restarting ETH borrow:

  • Monitoring new insolvencies caused by failed liquidations
    • The main reason for this is to understand whether the observed failures (if there are any) are due to, e.g., liquidity conditions or oracle issues around the Merge
  • Monitoring ETH pool utilization and ETH withdrawals to understand supplier elasticity
    • Given the rare nature of the Merge event, the data used to model supplier elasticity in the protocol will be hugely important for modeling edge case lender behavior
  • Comparing Aave’s ETH borrowing APY with the other money markets
    • This is mainly to test if the relative differences between the utilization distributions between Aave and other markets relax to their pre-Merge levels or if we observe a new equilibrium. This can be quantified, for instance, by measuring if the Kullbeck-Liebler divergence between various money markets’ utilization distributions pre- and post-Merge has changed dramatically
  • Monitoring Aave stETH insolvencies, stETH / ETH recursive positions, and Curve stETH pool liquidity
    • Given that the stETH/ETH market has the most reflexivity (and hence, recursive lending risk) with regard to the Merge, monitoring the state of this system will be crucial for understanding the conditional VaR of ETH borrowing post-Merge

These are not blocking the re-enabling of ETH borrowing, but we will utilize this data generated from the Merge as a learning opportunity to further refine Gauntlet’s risk modeling and risk management platform. Post-Merge, we will provide analytics and comparisons of risk parameters and borrower behavior before, during, and after the Merge, which can be used to complete a post-mortem on AIP-97’s performance. We will provide monitoring for these statistics on our new auxiliary Aave lending dashboard.

Next Steps:

  • Monitor the features as outlined above.
  • Welcome thoughts and discussion from the community.
  • Initiate a Snapshot vote, timing dependent on the state of the above features.

By approving this proposal, you agree that any services provided by Gauntlet shall be governed by the terms of service available at


Thanks @Pauljlei and the Gauntlet team for the analysis.

From a bit more technical/operational perspective, we would like to point out that, if targeting the re-enabling of ETH borrowing on Aave v2 Ethereum just after merge (let’s say having proposal’s execution time ~2 days after Merge, so creation by tomorrow), it is pretty important that:

  • The proposer (Gauntlet) is in quite close contact with their proposition delegator, for them to remove proposition power and that way cancel the proposal if anything is noticed and prudency says to keep it disabled.

  • As a fallback of the previous, the Guardian of the Aave governance is notified (BGD can help on this) to be alert of any potential cancellation.

Regarding the re-enabling itself, there is not really much more to comment on from a technical point of view; we will review the proposal as usual and will double-check that the Aave decentralized UI will function as expected.


Thanks, @bgdlabs. Given the community feedback that this proposal is just intended to revert the ETH market back to normal conditions, we will not put up a Snapshot vote in this case. We are targeting an AIP tomorrow (Tuesday 9/13) around 3 PM PT.


AIP is live: Aave - Open Source Liquidity Protocol


Update Post-Merge:

  • WETH utilization has dropped down to ~50%.
  • There have been no significant spikes in volatility after the merge.
  • The supply of WETH has remained mostly flat while sustaining a large amount of withdrawals and deposits post-merge.

So far, market conditions are trending towards a healthy state post-merge.


Update Post-Merge:

  • WETH utilization has continued to decline following the merge and is currently ~42%.
  • Having analyzed the liquidation behavior data from yesterday, we noticed no unusual behavior. See below for liquidation data.

We recommend implementing AIP 101 to re-enable ETH borrowing on Aave.

1 Like

As a Post Mortem of AIP-97’s performance, Gauntlet has analyzed market risk and borrower behavior before, during, and after the Merge. Here are several insights we would like to share with the broader community:

  • ETH utilization dropped to 45% within the same day post the Merge. Since the Merge, utilization has normalized to pre-September levels.

  • Aave’s ETH borrow rate has normalized post the Merge and returned to pre-September levels.

  • No significant new insolvencies post Merge. The chart below shows all new insolvencies, an estimate based on supply and borrow amounts.

  • Pricing feed from Chainlink did not indicate any price feed disruptions during the Merge event.

  • We are monitoring ETH pool utilization and withdrawals to understand supplier elasticity. After the Merge, Aave experienced an increase in supplied ETH.

  • We are monitoring stETH/ETH recursive positions, which is abstracted away by icETH token. ETH borrowing costs impacted the yield on this recursive ETH index prior to the merge but returned positive afterward.

  • Curve’s ETH/stETH liquidity pool had a highly disproportionate amount of stETH to ETH prior to the Merge, as shown in this dashboard. Since the Merge, the liquidity pool has experienced a high inflow of ETH. This rebalance of ETH/stETH within the pool reduces the price deviation risk between ETH and stETH.

1 Like