[TEMP CHECK] GHO Stability Module

Title: [TEMP CHECK] GHO Stability Module
Author:@AaveCompanies
Date: 7th July 2023

Summary

This TEMP CHECK requests approval for the introduction of a Peg Stability Module for GHO (The GHO Stability Module) as referenced in this Development Update.

Motivation

Overview

A Peg Stability Module (PSM) is a contract that enables the seamless conversion of two tokens at a predetermined ratio. This mechanism has proven effective for numerous stablecoin projects, such as MakerDAO in preserving the stability exchange rates. Using this model as inspiration, the proposed Stability Module for GHO leverages the benefits of existing models whilst innovating upon them in several ways to help further maintain GHO’s peg.

Features

As mentioned, the GHO Stability Module (GSM) is influenced by existing designs, with users being able to convert between GHO and governance-accepted stablecoins at a predetermined ratio. However, the design of the GSM also possesses unique features.

Price Strategies
The GSM introduces the concept of Price Strategies which provide the ability to adjust the pricing ratio between GHO and the exogenous asset, based on different strategies. Pricing strategies oversee the calculation of the price ratio, and:
• Can be fixed.
• Can be dynamic, based on price oracles and markets, linear curves, stableswap curves, etc.

For the initial launch, we recommend implementing a Fixed Pricing Strategy.

However, should the DAO choose to transition to a dynamic pricing strategy in the future, this shift can be facilitated through an AIP. The change would involve deploying a new Price Strategy contract and update the GSM to make use of the new Price Strategy.

The Debt Ceiling

The DAO may want to limit exposure to specific assets backing GHO through the GSM. The Debt Ceiling gives the DAO the ability to control the maximum exposure to the exogenous asset.

If a user tries to mint GHO and the amount of exogenous assets backing the issued GHO is higher than the Debt Ceiling, it reverts.

The Capital Allocator

The Capital Allocator provides a method through which a pre-defined portion of exogenous assets in the GSM can be used by a “fund manager” to allocate and earn yield.

The amount of funds that can be used is defined through a configurable threshold which is proposed to be decided through a separate governance process.

Last Resort Liquidations

In case of a rapid increase in risk in an exogenous asset, the GSM features Last Resort Liquidations to liquidate the exogenous asset. This special contract role allows – in the worst-case scenarios – for the DAO to take control of the asset (as a whole), thus pausing the GSM.

Price Bounds and Swap Freezes

In case the price of the exogenous asset deviates from the 1:1 ratio, trading can be stopped using Price Bounds and Swap Freezes.

Last Resort Liquidations, Price Bounds and Swap Freezes are enabled by an external contract granted with a special role that can pause trading. This contract has a function that will look up the price of the exogenous asset, and if it is outside of a certain bound, react accordingly.

Next Steps

If this TEMP CHECK receives community approval at the Snapshot phase, we will:

  1. Engage community auditors for a review of the GSM codebase.
  2. Following the audit, we will share the codebase with the community alongside an ARFC based on the Facilitator Onboarding Process Framework

Moreover, governance will then be able to decide on configurations around many of the GSMs features during the next stage of the governance process. These configurations include but are not limited to:

  • Initial Debt Ceilings
  • Initial Exogenous Assets
  • Initial Pricing Strategy
  • Distribution for Capital Allocation
  • Price Bounds for Swap Freezes

Overall, we are excited to introduce our vision for the GSM to the DAO and believe it will play a key role in peg maintenance.

Disclaimer

Copyright

Copyright and related rights waived via CC0.

11 Likes

This proposal has full ACI support.

3 Likes

Let’s $GHO! 100% in favor of this proposal

Hi @AaveCompanies,

We are supportive of the direction outlined in the [TEMP CHECK]. A GHO PSM is definitely needed. However, the initial parameters require analysis and the DAO needs to decide what level of exposure to each asset is desired. The Debt Ceiling for each asset may have a subjective component as well as a quantitative component. Recent events with USDC are to be noted and the relative performance of various stable coins during periods of stress. Our preference is for selective counterparty risk and to lean into a more decentralised asset focus when selecting GSM collateral type debt ceilings.

Would it be possible to learn a rough ETA on when this upgrade is expected to go live ?

Prior to submitting this code base to an external audit, we would like to suggest @bgdlabs performs a preliminary review. If this has already happened then great. However there is the opportunity to perform a peer review in-house prior to the external review and we think this is a prudent approach.

Hey Aave community,

At Angle we recently also explore a different approach to peg mechanisms. With the USDC depeg in March we have seen the limitation to Maker PSM or our own stability module - less critic as it took into account current USDC price.

We also felt the need to have a more robust approach to those black swan events. Post depeg we came with multiple requirements on the module, and implement them in the Transmuter, which fit most or all your requirements:

  • Tokens are traded at a prices than can only increase the instant collateral ratio. Currently by following this strategy - but can also be another strategy as specified in this post.

  • The whole system is govern by dynamic fees to target an exposure to each collateral. If governance thinks the best backing should be 35% USD1 and 65% USD2, the fees will be low when close to these targets and high otherwise. Contrary to debt ceilings you can control the proportion of each assets backing which is more robust for the peg.

  • Emergency state should always be running and should not require any trigger from the DAO or any other address, as it breaks the anti bankruptcy mode. This is why the module includes a redeem function, the account will exit the protocol with a proportion of all the collateral backing the stablecoin and therefore does not jeopardise the collateral ratio for slower participants.

  • Strategies can be implemented in the Transmuter to invest part of the funds. This equivalent to the Capital allocator.

If the Aave community would like to know more about the Transmuter and how it could be adapted to GHO, we would love to showcase even more the advantages of it!

The module is currently being audited on Code4rena and it should be done by the end of next week.

Guillaume from Angle Protocol

4 Likes

Thanks for this interesting proposal. I really like the idea of having a PSM from scratch for GHO.

Coming from Angle like Guillaume and wanted to share my thoughts on the overall design proposed here.

Here I think that the overall design makes sense and it’s more robust than any other design that is live out there, but I think there are some points that could make it slightly more optimized. I’m no so deep in the Aave governance, so please perdon me if what I suggest does not apply.

On the debt ceiling: I think it’s a must to have a debt ceiling, I’m just curious to know if the debt ceiling is going to be absolute or relative. I believe that relative debt ceilings make more sense, especially as it naturally automates a lot of the process of controlling the exposures of the system to the assets it has in reserves. Let’s say that GHO has LUSD and USDC in reserves with debt ceiling of 1m, and there is an alert on USDC (like during the USDC depeg), then people will withdraw their GHO for LUSD and potentially try to mint USDC, hence the debt ceiling of USDC will be reached after some point, but the problem is that 100% of the exposure of the system will be to USDC. Dynamic control on relative exposures may make sense → so you prevent people from exiting with like LUSD if it underexposes you to it or overexposes you to USDC.

On price bounds and swap freezes: this feature is also a must and I’m glad you thought about it. That being said, I think that it is possible to pause the system without having to rely on a special contract role. The contract can be effectively paused by setting a mint or a burn price equal to the minimum between the exogenous asset current price and a form of target price. Like, it can make burns or mints purely improfitable so no one actually uses them and no trusted address is needed to payse.

I do agree that its important to have last resort liquidations, but it sounds also like a manual process involving governance and leaving GHO stakeholders quite uncertain about the fair value of the reserves in the GSM while the proposal hasn’t been executed. In the meantime, do you think that adding a redemption feature which enables GHO to be redeemed against a proportion of each asset in the reserves might make sense?
That way, in case of distress or a rapid increase in risk in an exogenous asset, anyone can leave the protocol reduce its exposure without having to wait for any form of special contract or governance intervention!

Happy to ideate around these ideas: designs of Peg stability module have become something we’ve spent a lot of time on recently, and we’d love to leverage our thought to try making GHO safer and reduce the operational overhead for maintaining the system in case of distress.

1 Like

@TokenLogic: Thank you for your reply.
Regarding the initial parameters, if the following TEMP CHECK moves to the AFRC stage we will provide our recommended initial parameters for community discussion. Regarding the ETA, because of external audits and other factors, it is not possible to provide a rough ETA. Once the code is finalized, we will request a review from @bgdlabs.

@gnrv & @sogipec thank you both for your insights!
We will take time to review the Transmuter along with the other feedback provided.

2 Likes

Chaos Labs fully supports the introduction of the GHO Stability Module. We believe that incorporating a stability module or similar mechanics is crucial to ensure the stability of decentralized stablecoins, especially considering the current state of the crypto markets. Drawing from the analysis of Maker PSM’s historical behavior, we must avoid the significant pitfalls that arose from excessive reliance on USDC.

Regarding the specific innovations proposed:

Price Strategies: We consider enabling dynamic strategies to be vital for addressing unforeseen events and providing effective mitigation. However, it is essential to define these strategies carefully. Initially, we recommend implementing a fixed price strategy, as suggested, since it can be deployed quickly and should suffice while the PSM is still relatively small.

Debt Ceiling: We view the debt ceiling as a potent tool for mitigating the adverse effects of over-reliance on centralized stablecoins. Building on the approach proposed by @sogipec, it is important to set the ceilings in a relative manner. However, we differ in our opinion on the approach itself. Instead of setting debt ceilings relative to each other, we believe they should be relative to the circulating supply of GHO. Furthermore, we advocate for the inclusion of a hard cap as originally proposed.

In the event of a depeg event, users can supply the depegged asset and mint GHO, similar to what occurred with DAI during the USDC depegging. With responsible debt ceiling settings and secondary market liquidity, withdrawing other stablecoins from the PSM becomes a secondary consideration for the protocol. The primary mitigation for such scenarios should be employing dynamic price strategies.

Daily Limits: In addition to the debt ceiling, we believe that imposing limits on the velocity of stablecoin influx is an effective risk mitigation measure. This was observed during the USDC depeg incident in the DAI PSM, where Maker DAO limited the daily amount of USDC that could be deposited into the PSM. Therefore, we recommend considering the implementation of such a mechanism for the GHO Stability Module that will limit the total amount of GHO that can be minted daily by the PSM.

Control over Last Resort Liquidations, Price Bounds, and Swap Freezes: We strongly advocate for the development of a contract that can control Price Bounds and Swap Freezes based on external asset prices. Such a contract would significantly improve response time during crisis situations. We emphasize the importance of the logic within the external contract and suggest further exploration and auditing before finalizing it. While we agree that swap freezes and price bounds should be governed by a smart contract, we believe that Last Resort Liquidations are better suited for explicit DAO votes, allowing decisions and mitigation to be made based on the specific circumstances at hand.

Thanks for your comment and reply.

I agree on the soundness of everything that is proposed here but with some caveats.

For the ceiling relative to each other rather than relative to the total supply, I think this approach makes less sense when there are going to be multiple facilitators to issue GHO.

Like, let’s say it’s only possible to borrow GHO (through an Aave facilitator) and mint it from a PSM. In the end in case of a depeg, all GHO borrowed should be repaid because people are happy to repay their debt at a profit. But you end up with only the GHO from the PSM.

So, my concern is that each module within GHO should be fully functional on its own, and be built without any consideration for what’s happening outside, especially PSMs as this might lead to flawed assumptions and parameters setting.
The winning setup for me is the hard cap (even though I’m not sure it’s needed and I can see @MarcZeller or other delegates spend a lot of time making proposals to increase this hard cap) and exposures relative to the different assets in the PSM which is a form of cap as well.

The cool thing with relative debt ceilings is that it’s an automatic measure with no maintenance or management as the system grows (or as PSM grows but not other GHO mechanisms), also removing the need to rely on a daily limit (which is also a manual parameter requiring maintenance).

On the control over swap freezes, I think the contract you’re describing can be made fully permissionless provided that there is a sense for each asset supported of a target and a current price → burning should be impossible if one asset depegs, and minting from a depegged asset should be automatically be made impossible with this.

Anyway, glad to see some discussions on this, and end up on a solution that:

  • 1: guarantees GHO stability, resilience to black swan events and robustness
  • 2: comes with minimal maintenance and parametrization to reduce the burden on delegates and external monitoring tools, which will in the end make the system even more predictable for all its stakeholders
1 Like

We are in support of the proposed Stability Module for GHO. Adequate research has been leveraged in developing this proposed design, and we look forward to seeing it in action.

We are in support of the GSM!

Gauntlet is looking forward to the implementation of GSM for GHO. The innovative features of the GSM can help mitigate risks we discussed in our previous analysis, including low liquidity and usage and high supply relative to liquidity. For instance, the GSM will be highly important when GHO goes < $1 to help restore it back to $1. In general, the GSM will help mitigate potential price manipulation for GHO, especially since Emode minting can lead to premature GHO supply explosion relative to insufficient DEX liquidity.

  • The price strategies feature of GSM can help stabilize the peg of GHO by adjusting the ratio of price between GHO and the exogenous asset. Gauntlet supports the initial use of a fixed pricing strategy at launch. There are many potential dynamic pricing strategies that can be considered in the future, however, their impact on GHO can be complex and would require additional in-depth analysis.
  • By limiting exposure to specific assets, the debt ceiling can help mitigate the impact depreciating underlying assets can have on GHO price. Without a debt ceiling, the GSM can be overexposed to these depreciating assets, making it difficult to maintain the peg of GHO.
  • The capital allocator mechanism can be a potential growth generator for GHO. Setting the configurable threshold would require analysis of the risk-reward tradeoff of the exogenous assets, depending on factors including their liquidity, volatility, and yield.
  • The last resort liquidations, price bounds, and swap freezes are important measures to be put in place to protect the stability and peg of GHO in the worst-case scenarios. These methods will help serve as last resort levers in case underlying assets in the GSM are suddenly depreciating.

Ultimately, carefully balancing underlying assets in the GSM - especially during abnormal market events - will be something to focus on, especially since Emode minting capabilities for GHO can rapidly contribute depreciating underlying assets to the GSM and concurrently depress GHO prices.

3 Likes

The Snapshot for this proposal has been queued and will be live for voting on Jul 20, 2023, 10:47AM UTC+1

You can find it here: Snapshot

1 Like

Hellos, just 1 question. Makerdao had to profit share for using AAVE’s intellectual property for Spark Protocol, is AAVE going to reciprocate for using Maker’s PSM intellectual property?

Thanks

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