[ARFC] Extend Safety Module Coverage to Polygon v3

Governance
4

Hey guys, before extending the Safety Module in its current form, can we introduce a discussion around the underlying risks in the current safety module design and some potential solutions?

During our coverage of dYdX, dYdX Grants commissioned some research by Xenophon Labs around their safety module, which was a fork of Aave’s. We find sincere congruities between the two, and think there is a more efficient way to protect borrowers.

We believe the primary issues with the current design are:

  1. Alignment of assets vs safety valve (AAVE).
  2. Efficacy of BPT approach in times of crisis.
  3. Incomplete implementation around the backstop module and, if completed, inefficient.

More detail on those and potential solutions:

  1. There is no risk of AAVE being the asset that makes up the bad debt (since there is no borrowing it!), so using that as the reserve asset seems like a sub-optimal and misaligned choice. The reserve factors better address this, but will take time to build up to be sufficient in size (thus the need for a safety module). The best reserve asset should likely target the most liquid asset, at bare minimum, which would be ETH or stables.

a. Are there other considerations on why the safety module should be made up of AAVE tokens? Or issues with using ETH or an LSD?

  1. BPTs are a great idea for liquidity but using the 80/20 weight can have the opposite effects on deep selloffs. In the research mentioned above, the impact on TVL in weighted pools from harsh sell-offs at different token weights shows the risk of an 80/20 pool as a reserve asset or even liquidity tool for reserve assets. We think, given the likelihood that AAVE price falls in the event of a safety module auction, a more balanced or even underweight pool would better serve the desired purpose. Gauntlet has also shared some thoughts on LP tokens during liquidations, which makes us think these are suitable as liquidity drivers but not as a reserve asset, if that is their usage here.

  1. In the docs there is a contract called the “backstop module” where users could deposit USDC and ETH and be the first bid for the Auction Module when Aave needs to be sold. Depositors would be rewarded with a revenue share. Although we couldn’t find any docs on implementation or usage of the backstop module besides the design, it makes sense as a way to turn AAVE into more liquid assets quickly. The question becomes though, why pay two parties (AAVE stakers and backstop module depositors)? We propose a better design would be to just incentivize ETH and/or USDC depositors and skip the auction step.
2 Likes