[ARFC] wGHO Aave V3 Onboarding

Governance
1

Proposal Updated without emode and with lower caps

Title : [ARFC] wGHO Aave V3 Onboarding

Author : Marc Zeller @marczeller - Aave Chan Initiative, Chaos Labs

Date: 2023-09-15

Summary

This proposal aims to onboard wGHO, an ERC-20 token wrapper of GHO, as a collateral-only asset in a newly created stablecoin emode within the Aave V3 Ethereum pool. The integration will enable using wGHO as collateral to borrow other stablecoins, facilitating arbitrage and leveraged positions for under-peg GHO situations.

Motivation

The onboarding of wGHO presents an opportunity to enhance the Aave ecosystem by allowing wGHO to be used as collateral for borrowing stablecoins such as USDC, USDT, DAI, and LUSD.

This integration will enable arbitrage and leveraged positions in cases of under-peg GHO, providing the markets with tools to act on peg-deviation of GHO in both directions.

Alongside the upcoming GHO PSM, this proposal will further strengthen the market’s ability to maintain the GHO peg.

It’s important to note that although wGHO is proposed as collateral in Aave V3, it will not be a collateral of GHO due to its collateral-only, isolation mode, and emode asset status.

wGHO will only be available to borrow DAI, USDC, LUSD & USDT.

Specification

The ACI proposes the following parameters for the integration of wGHO into the Aave V3 Ethereum pool:

  • Onboard wGHO as a collateral-only asset & isolation mode asset.
  • wGHO parameters:
LTV 77%
LT 80%
Liquidation penalty 4.5%
Supply Cap 1M
Borrow Cap 0
Debt Ceiling 1M$

Note:

wGHO is a simple ERC-20 wrapper around the GHO contract. A simple smart contract call allows anyone to wrap and unwrap GHO and wGHO making it as liquid and with the same risk profile of GHO.

Onboarding wGHO instead of GHO, is due to technical implementation and risk isolation reasons. From the end-user perspective, wGHO & GHO are functionally equivalent. If this proposal is validated by governance, the UI frontend is invited to abstract this away.

Disclaimer

The Aave Chan Initiative is not affiliated with or paid by any third parties to publish this ARFC. The he author have both large GHO debt & GHO holdings. The ACI uses GHO to fund the Orbit program.

Next Steps

  1. Gather community feedback and consensus on this ARFC.
  2. Escalate this proposal to the ARFC Snapshot Stage.
  3. If Snapshot outcome is YAE approved, escalate to AIP stage.

Copyright

Copyright and related rights waived via CC0.

11 Likes
2

This ARFC should be considered potentially synergistic with [ARFC] sDAI Aave V3 Onboarding

here’s a FrenchChart of a possible “leverage long wGHO” strategy enabled by this proposal allowing up to x10 leverage long on GHO to bet on an asset repeg. the Green section mobilizes the sDAI yield to cover the borrow cost for the strategy, but this part is optional.

image
image2545Ă—971 212 KB

for situation when GHO might be valued over 1$ on secondary markets, the upcoming GHO PSM will allow users to act in the opposite direction

5 Likes
3

Overall seems reasonable and I’m sure would be a popular position to go long wGHO / short other stablecoins given it currently is available at a discount. However, if the wGHO is not borrowable then the position will have negative net interest earnings and collateralization will decline over time, and users are effectively betting GHO discount will decrease fast enough for their profit to not be eaten up by interest payments. If people are forced to bail out of the trade at a loss because it isn’t returning to parity fast enough, this might cause some downside volatility in GHO price.

Also curious if wGHO is planning to use 1:1 / fixed $1 pricing, or using a price oracle based on GHO market rate? I think 1:1 would generally make sense, but then it may not be profitable to liquidate positions if GHO is trading at price below $0.99 (given only 1% liquidation penalty).

3 Likes
4

this proposal intends to use the Chainlink GHO/USD price feed,

2 Likes
5

It might be purely theoretical scenario, but say GHO price is above $1.2, for simplicity say $2.

Then I can deposit 1 wGHO, borrow 1.5 DAI, deposit the DAI, borrow 1.1 GHO (as afaik GHO price is hardcoded to $1).

I get free money this way, isn’t it an issue? What am I missing?

1 Like
6

You would probably sell that 1.1 GHO for dai tosecure your profit/ mint more GHO to get more money right ?

Congratulation, it’s called an arbitrage, I just hope you will be fast enough as you will be competing with bot trying to do the same

1 Like
7

No, it is not an arbitrage, I started with 1 gho, and ended up with 1.1 gho. Now I can just repeat it and get more gho. I don’t have to sell it.

In theory I can start with 1 gho and end up with they entire wGHO supply cap, without selling a single GHO.
But maybe I am missing something.

1 Like
8

yep, but as said before you won’t be alone, other peoples will do the trade and it turn into a rush to sell gho as you have lots of people who will wants to cash out, making gho price go down (let’s say 1$ for simplicity). You then can repay your loan and get back you 1.5 dai, wich means you are .5$ at loss since you bough gho at 2$ (or you had gho before but it would have been more profitable to swap directly)

Note: I am assuming you only got the time to do one loop, and couldn’t save your loan.

But in reality, as soon as gho is over peg peoples will sell it to another stable, making a profit. Should the price increase enough for the trade you mention to become possible it will only increase the selling pressure. Acting as a sort of hardcap.

1 Like
9

The attack doesn’t come with any selling pressure, with enough loops I make infinite gho out of my 1 gho (in practice of course it is bounded by supply cap and gas costs).

But if the assumption is that gho price will never be significantly > $1, then it is is fine.

2 Likes
10

@yaron

wGHO collateral value will reflect secondary market prices. Even if you can do what you describe, you’ll need to sell the minted GHO to keep the loop going, and this will affect secondary prices and thus the value of wGHO collateral.

there’s no infinite money glitch.

the forum is not designed to have this kind of conversation, you’re kindly invited to continue discussing this on Discord or elsewhere. please stick to the proposal itself in this topic.

3 Likes
11

the Snapshot vote outcome is YAE.

But after discussing with @Gauntlet about theoretical risk scenarios, the ACI will follow the conservative path maximizing user safety, and won’t escalate this proposal to the AIP stage.

4 Likes
12

Can this risk be mitigated in the future or whats the deal here? Can you elaborate?

1 Like
13

Thanks @MarcZeller for putting this proposal forward. Below is Gauntlet’s analysis regarding wGHO on Aave v3.

Gauntlet recommends against listing wGHO as collateral in the proposed format. Listing wGHO may increase the surface area for a liquidity driven attack and introduce excessive recursive borrowing risk, without contributing positively to GHO peg stability in a material manner, especially with the GSM. Any parameters associated with wGHO collateral may be stale, making it difficult to guard against such attacks.

Enabling wGHO as collateral introduces the possibility of the following non-atomic manipulation. The magnitude of attacker profitability depends on a number of factors - such as available GHO liquidity and current GHO price and is inherently uncertain due to this non-atomic nature.

  1. Borrow large amount of GHO
  2. Deposit half into wGHO, emode borrow maximum USDC. Keep half borrow as GHO.
  3. Sell borrowed GHO in market for USDC. At current liquidity levels for GHO, the impact may cause GHO << 0.9 (since pool is only $3m on the bb-a-USD side on Balancer)
  4. (**) Use new USDC to self-liquidate your wGHO - USDC position, recover all of GHO + keep some initial borrowed USDC.
  5. repay your remaining 5m GHO debt from the recovered wGHO position.
  6. You end up with free USDC, initially borrowed from the wGHO collateral.

(**) Whether or not self liquidation can happen is for debate, since the GHO CL feed does not update atomically.

  • The attacker is able to self liquidate their wGHO collateralized position.
    • Profitability depends on liquidity levels for GHO.
  • The attacker is unable to self-liquidate (other agents liquidate the wGHO position).
    • use USDC borrowed from wGHO position + USDC from GHO market sell proceeds to buy GHO and repay GHO debt
    • Amount of leftover USDC depends depends on GHO price and liquidity.

To conclude, while the profitability and success of the attack is dependent on GHO price and GHO liquidity, this mechanism introduces excess risk with respect to its GHO repegging benefits, which will be a significant use case for the GSM.

8 Likes
14

As mentioned in our response provided here and as detailed in the original proposal, enabling wGHO as collateral to facilitate the borrowing of other stablecoins can contribute to maintaining the GHO peg.

We acknowledge the potential risk stemming from liquidity and price manipulation attacks, especially considering GHO’s short history and limited liquidity, as illustrated in the aforementioned theoretical scenario. While these attacks hold theoretical feasibility, assessing their actual viability and profitability is notably complex.

As risk managers, our responsibility involves effectively balancing risk and reward. V3 was built with mechanisms designed to mitigate risk, particularly for assets considered “riskier.” Through the implementation of suitable supply limits and debt ceilings, we can reduce the likelihood of such attacks and curtail risk dynamically in response to evolving community risk preferences.

Should the community wish to list wGHO as collateral, and given the present liquidity and limited historical data for GHO, we recommend listing in isolation mode with an initial 1M supply cap and a debt ceiling of $1M.

5 Likes
15

Can we re-evaluate this proposal?

16

the proposal has been updated to remove emode & implement lower Caps.

2 Likes
17

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.