[TEMP CHECK] Community Plan for GHO Stability and Peg

Summary

In this post, Gauntlet hopes to help formalize the community discussion around strengthening the GHO peg and to organize into its core components with actionable items. We will outline -

  • Current state of GHO minter behavior and relationship with GHO market price
  • Potential methods to stimulate GHO demand, as well as other suggestions for the liquidity plans discussed in the community
  • Our view on how key GHO updates should be scheduled to bring the highest impact

Motivations

Since GHO was launched with v3 Ethereum as its facilitator in July 2023, it has grown to ~25m in market cap, with ~12m locked in DEX pools across Uniswap and Balancer. Strong supply side initiatives help incentivize GHO LP positions to support a maturing stablecoin. Proposed plans to diversify liquidity pools will further provide variation in trading pairs.

One of the main challenges that may prevent greater GHO adoption is its discount to the $1 peg. GHO has traded < $1 since inception, exhibiting general negative drift, with some price spikes, possibly due to liquidations involving GHO debt and/or GHO repayment. As a result, GHO ADV is ~$1m and price movement is heavier-tailed compared to other stablecoins. GHO rolling standard deviation (GHO_std in graph below) also exhibits some degree of volatility.

The DAO has also taken various steps to swap treasury holdings into GHO to fund future strategic initiatives. However, bringing GHO back to peg will require sufficiently stimulating organic demand for GHO, in addition to the supply side initiatives already live. So far, a number of suggestions have been raised to help bring GHO back to peg. However, each of these have their own questions around their efficacy in doing so, as have been brought up via discussion.

Current State of GHO: How GHO minter behavior impacts GHO market price

Analysis of the relationship between GHO price and GHO positions further emphasizes the need for increased organic GHO demand. Currently, decreasing GHO price encourages existing profitable GHO mints to repay and realize profit.

  • We define the profit of a GHO position to be the difference between the current market value of that GHO borrow, versus the market value when it was minted.

GHO positions timeseries, mint price considerations

  • (Blue) Weighted averaged GHO mint prices trend downwards, with recent magnitude of downwards movement limited by decreased recent mint volume compared to GHO launch.
  • (Red) Total GHO minted has passed 50m.
  • (Pink) GHO Price briefly recovered by borrow rate increase to 2.5% on 2023-09-20, but price decay continued to happen

GHO positions timeseries, with position pnl considerations

  • (Green) Large upswings in realized profitability (GHO position close-out) occurred at declined GHO price levels
  • (Yellow) Positive weighted unrealized profitability is mostly due to legacy positions opened at GHO launch when GHO price > 0.98
  • There is slight negative correlation (-0.13) between GHO daily returns and GHO supply changes.

GHO mint prices for current positions

  • A large number of early minters repaid their debt and took profits on GHO price decline (late 2023-08).
  • Roughly half of all GHO minted (~25m) was minted by 2023-08-11, at which GHO price was 0.981.
  • There were 15m GHO minted with avg price above 0.981 on 2023-08-11.
  • As of 2023-10-21, roughly 6m GHO remains minted with avg price above 0.981, suggesting a large number of positions have been repaid.

Conclusions on current state of GHO

Given current status of GHO demand vectors, GHO minters could be broken down into two types of users: GHO shorters who seek to make profit when future GHO price < mint price, and GHO LPers who mint GHO to LP on DEX. Both can contribute to GHO sell pressure.

  • GHO shorters market sell their minted GHO.
  • LP can acquire double-sided liquidity positions through minting GHO and selling half of the GHO for USDC, contributing to GHO sell pressure.

This suggests the following may be playing out.

  1. 50 bps borrow rate increase per the schedule encourages a tranche of users to repay their GHO, some passive LPers and some shorters
  2. GHO repayment briefly triggers GHO price recovery
  3. New GHO shorters enter at this higher price point and mint GHO to market sell, new GHO LP mints GHO due to increased volume and chance for higher fees
  4. Influx of users from (3) may restart price decay.
  5. GHO price remains outside of (0.995, 1.005) per borrow schedule, leading to proposal for rate increase.

As a result, without organic demand and buy pressure to counter (4), the above sequence may cycle and repeat, potentially leading to decreased GHO supply (excess costs from mint) without repeg. Ultimately, minter behavior may change once borrow rates exceed benchmark rates (such as Maker DSR) or when borrow rates exceed expected fee revenue from liquidity provision.

Potential methods to stimulate GHO demand

Finding synergy between the features of Aave as a lending platform and GHO may help spur usage. The high level to achieve this could be to use GHO borrow rates to finance GHO demand vectors. We explore a couple additional ideas below and explore their tradeoffs. The ideas we suggest may allow for users to recycle between various use cases in their lifetime of holding GHO, across evolving market volatility.

Stake GHO to lower WETH/USDC/USDT/etc borrow rates

At a high level, users can stake GHO to reduce their borrow fees. The premise behind this approach would be to draw demand for GHO by providing rebates on borrows, especially during bull markets. For instance -

  • a user looking to recursively collateralize WETH with WSTETH can stake GHO to reduce their borrow fees for WETH loops, increasing profitability for the strategy.
  • a user looking to implement an SDAI ↔ USDC loop can stake GHO to reduce borrow fees for USDC loops.

This type of rebate may allow for more user-centric benefit rather than a blanket yield denominated in GHO (such as reducing borrow costs for leveraged LST-WETH strategies). A couple considerations need to be kept in mind.

  • The magnitude of borrow rate discount for each asset should be a function of asset volatility to USD, with sufficient spread to GHO borrow rates (i.e. stablecoins can have higher discount, WETH has less discount) We provide an example below.
  • Each asset would need a “discount cap” that bounds the amount of borrow with discounted borrow rate in order to preserve supplier APR

On borrow rate discount

Borrow rate discount should facilitate Aave generating more revenue under this framework as compared to the current regime excluding GHO with high likelihood.

  • For each asset
    Screen Shot 2023-10-27 at 11.21.57 AM
    i.e. rate_buffer = 0.005

  • USDC/USDT/DAI borrow_rate_discounts can be
    Screen Shot 2023-10-27 at 11.20.51 AM
    Since current GHO borrow rates are 3%, this comes out to ~2.5%.

  • For other assets, we can consider finding discount rate per asset X such that for some probabilistic bound k (say ~90%)
    Screen Shot 2023-10-27 at 11.19.34 AM

  • Using GBM (Geometric Brownian Motion) for asset returns as a rough proxy and that WETH volatility ~ 40%, we find that WETH can support a borrow rate discount of ~1.5% at probability threshold 90%.
    Borrow rates for $100k WETH borrows, as function of staked GHO percentage of total borrows and current utilization, with 1.5% borrow fee discount

On discount caps

  • Borrow rate discounts should not impact supplier APR, to avoid supply contraction. Let D be the discount cap as a percentage of current supply. The following should hold, for each asset X. Then we have

Screen Shot 2023-10-27 at 11.23.15 AM

Some observations -

  • The right hand side will be at least the reserve factor of X.
  • Discount cap as a percentage of current supply may be tricky, especially when for asset X, amount of GHO staked exceeds discount cap.
    • One way to counteract this would be to enact a penalty, users staking GHO for asset X in this situation would pay interest penalty (i.e. withdraw less GHO than they staked), and have decreased discount benefits.
    • This would be similar in principle to the current IR framework (slope1/slope2) for Aave.
  • As an example, WETH would allow for discount caps of ~30% current WETH supply.

Stake GHO to earn GHO Savings Rate, fueled by liquidation protocol fee (LPF)

Aave v3 introduced the Liquidation Protocol Fee, which takes a portion of each liquidation bonus and sends to treasury. Creating a GHO Savings Rate where interest is paid from this liquidation protocol fee may drive GHO demand and encourage people to hold GHO during volatile periods. Those who stake GHO in the GSR vault would be eligible to collect the GSR. We give some rough historical statistics.

  • Aave liquidations bonus across all deployments totals $23m over past 1.5 years
  • assuming LPF of 10%, that amounts to roughly $1.5m in LPF, or 1.5% APY for 100m GHO staked in the GSR vault.

We can simulate GHO savings rate accumulation. Let Liq: X x P → M be the liquidation function that takes a loanbook matrix X and a daily price trajectory P that outputs a total daily liquidation M per asset. Let LB denote the vector of liquidation bonuses.

Then yearly expected liquidation bonuses can be modeled as

Screen Shot 2023-10-27 at 11.34.30 AM

where P is a k x s matrix for k assets and s oracle updates. Taking expectation of Et and noting that liquidations are convex relative to loanbook changes and price change,

Screen Shot 2023-10-27 at 11.35.47 AM

We make a number of simplifying assumptions to model the right hand side:

  • loanbook changes are 0 from day to day
  • correlations hold and nothing “abnormal” occurs (i.e. no stablecoin depegging)
  • Proxy P via bootstrapping price returns from the past two years for WETH and assuming asset returns are equal to the asset beta-multiplier to WETH (stablecoin 0 beta to WETH)
  • LPF is 10% for all assets

Simulated Growth of Accumulated GSR Rewards, assuming 100% of LPF goes into GSR, using Aave v3 Ethereum as loanbook

Simulated GSR vault growth over 100 runs reveals a median yearly GSR vault accumulation of ~280k, assuming 100% of LPF.

GSR could then be distributed proportionately across stakers in the GSR vault. There are a couple considerations with how the GSR is paid out, since a LPF based GSR could cause excess depeg > $1 when significant market downturn occurs (in addition to GHO buys for liquidations).

  • Vault distributes GSR daily based on average liquidation amount over past month to smoothen out liquidation spikes.
    • LPF from liquidation on day i would be distributed evenly over day i+1 to day i+30
  • GSR is only paid to GHO staked in vault. The more GHO staked in vault, the lower the APY from GSR.

This also drives a couple of options on how users receive their GSR.

  • GSR vault simply accumulates raw collateral from LPF and distributes rolling average amount to users.
  • Users then have the option to withdraw in GHO or native token. If they withdraw in GHO, at the time of withdrawal, their proportional share is converted into GHO.

On how GHO mechanism updates should be scheduled

  1. Bring GHO back to peg via community discussed methods
  2. Activate GSM to mitigate impact of temporary dislocations to peg
    • GSM activation should have several conditions, discussed in the Appendix
  3. Continued monitoring to evaluate GHO balance across liquidity and demand as GHO scales
    • Peg is trading within (0.995, 1.005)
    • GSM supply caps are reasonable and do not present GHO price manipulation risk
    • Aave v3 bucket capacity is properly managed

Next Steps

We welcome community feedback on the above analysis. Some questions to consider:

  • Does the community agree that stimulating GHO demand should be a priority?
  • Which potential mechanisms to stimulate demand resonate the most with you?
  • What are the technical hurdles and level of difficulty to implement these potential mechanisms, as they relate to other measures being taken (i.e. GHO borrow rate increase schedule)?
  • What are the community’s strategic thoughts around the interplay between the lending protocol and GHO?
  • What are the community’s thoughts on the GSM activation criteria?
  • What are the community’s thoughts on how GHO mechanism updates should be scheduled and the conditions for monitoring?

Appendix

(A) GSM Activation Criteria

For Gauntlet to recommend GSM initialization, the following should hold.

  • GHO should consistently be trading > $0.995, i.e. hourly simple moving average > $0.995 (price at peg)
    • GHO rolling 100 hourly SMA is $0.975
  • GHO annualized volatility < 10% (price stability)
    • current GHO annualized volatility is ~6% from past 90 day
  • GHO ADV > 20% total supply (price maturity)
    • current ADV ~5% of total supply

Gauntlet discussed previously that GHO utility to drive demand pressure is necessary for GSM sustainability.

  • Without organic GHO demand, GHO price naturally decays and trades < $1, which is as we hypothesized before.
  • Equilibrium GHO price < $1 naturally eats away at GSM reserves.
  • GHO price may intermittently spike due to liquidations from market downturn (such as 2023-08-17), or user repayment if $$ \text{GHO market price}{t{mint}} > \text{GHO market price}_{t} + \text{threshold}$$ where threshold is user preference function on size minted, GHO liquidity, time GHO position has been open, and GHO borrow rates, and t is the current time.
    • Before Maker introduced DSR in late 2019, DAI price followed similar trajectory, with liquidation volume spiking the equilibrium DAI price of < $1.
  • GHO demand pressure will incentivize market buying which will help bring GHO closer to peg.
11 Likes

I totally agree with all the points in this post, the current situation is quite precarious for GHO’s growth and tackling this issue early and swiftly should help build back trust in the future.

3 Likes

This proposal is an excellent approach for driving GHO Stability and Peg. We are particularly impressed by the Stake GHO to lower WETH/USDC/USDT/etc borrow rates and Stake GHO to earn GHO Savings Rate, fueled by liquidation protocol fee (LPF).

By giving additional utility to GHO, we can drive demand for GHO to meet existing Aave user needs, in our opinion these ideas should be explored on a trial basis to see the results they yield in the short term at least.

3 Likes

Not bad suggestions.

Do I understand correctly that per your numbers $10m sGHO would yield something like 15% a year if liquidations/tvl stay same? That is actually attractive, but 1.5% (at 100m staked) wouldn’t fix the issue I don’t think.

The reality is that you don’t have any stable coin holders willing to buy GHO with fresh stables and hold and no mechanism for them to realize upside on a return to peg because as stated until borrow cap is reached, there will be incentive to borrow and sell GHO when it gets towards $1.

Without new inflow of stablecoins into the system through LP or buying outright, peg issue won’t resolve.

What happened to the proposal to make GHO collateral? I would suggest that GHO as collateral with a fixed price of $1 would solve the issue as people could then take leveraged long positions.

3 Likes

We are thankful to @Gauntlet for taking the lead on this matter. We love the idea of staking GHO to lower borrow rates, but we also suggest enabling GHO as collateral as @0xTogbe has suggested. We understand Gauntlet’s concerns about this subject but have enough risk parameters to face the challenge.

2 Likes

I think the community agrees that GHO adoption is lacking its expectations. Here, I wanna share my two main insights. I am curious on whether you guys agree:

Insight 1:
GHO borrowing rate below s-DAI interest(5%) and a stable GHO<>Dollar peg is not possible!

The reason for this insight is straight forward: Assuming both holds, then “rate-Arbitrators” will simply provide collateral on Aave to borrow GHO and exchange it to sDAI. Then they can deposit their sDAI into Aave and repeat the borrowing of GHO and exchange to sDAI. This way the rate-arbitrators build a leveraged position to exploit the rate difference between sDAI interest and the GHO borrowing rate. Rate-arbitrators would do that until the peg between GHO and DAI breaks again. (See appendix for more evidence **)

If both - having a stable peg and a below sDAI interest GHO borrowing rate - then the question is what is more important.

Insight 2:
A stable peg is more important than a below market competition interest rate

If the peg is not holding, then borrowers have an additional risk: They might realise losses due to the added volatility. The history of finance has shown that people advert non-calculateable risks and prefer products with a clear, predictable trajectory.
Also the accounting of GHO for all kind of purpose is harder, if the value of GHO is freely floating.

Do you agree with the insights?

  • Insight 1 - peg and below market rate borrowing not possible
  • Insight 2 - peg is more important than below market rate borrowing
0 voters

If people agree with these two insights, then I think there is only one reasonable option:
=> Increase the borrowing rate to at least the sDAI saving rate.

The only question is how quickly this should happen. Currently a rate increase of 50 bps is scheduled for each month, which would take at current circumstances another 3,5 months. Maybe we should do it quicker, but I am not sure. It depends on the end vision of GHO.

According to my current understanding from GHO, the strategy was to enable below market-rate borrowing to drive adoption. If this is not possible, what should be the vision?
I am really curious on your visions.

My humble proposal: Drive adoption with sGHO
This would entail:

  • raise borrowing interest to at least sDAI saving rate and ensure the peg
  • earned borrowing earnings are delegated into a sGHO contract (similar to sDAI).
  • people can stake their GHO in the sGHO contract to earn the borrowing fees. This would allow them to earn at least the 5% and make the GHO coin attractive to hold.
  • long-term there will be a small spread between sGHO interest and GHO borrowing rate that will be captured by the aave DAO.

This would make GHO (or sGHO) attractive to hold for its users. People would still want to borrow GHO, if they wanna build leveraging position via Aave.

Appendix
** A similar situation appeared just recently within the DAI ecosystem. The sDAI saving rate was 8% and the borrowing rate for DAI was 5%. The immediately - within days, the people were putting up a lot of collateral on spark to borrow at 5% and earn 8%. It’s easily visible in the trajectory and the maker governance had to change the interest rate of sDAI back to 5% (see here )

2 Likes

To be clear, the only way I think this helps is if GHO is treated as pegged to $1 within the Aave system, otherwise there will be minimal reason to use it as collateral.

With the current stablecoin rates situation in DeFi and larger macro environment, a fixed borrow rate on various collateral is going to be attractive even if rates are far above sDAI yield.

Borrowing other stables has seen volatile rates since the increase in the DSR as a lot of stables have flowed out to stake DAI through leveraged rates traders and also people pulling deposits to go to a fixed DAI rate.

Once the stablecoin rate curves are adjusted up, it will only be more expensive to borrow, and will keep GHO an attractive option.

The price situation could be seen as an additional fixed cost on top of the fixed rate, but the reality is that almost everyone who shorted GHO the last 3 months is in profit.

We are supportive of the sGHO plan as this does add some native demand for GHO at current market cap and revenue from liquidations.

We also recommend considering wGHO be added as collateral with a fixed $1 oracle price. This will allow capital efficient leverage longs to come in and support the market by borrowing stablecoins and buying GHO below peg.

While these are discussed it seems prudent to more quickly raise the interest rate and also heavily consider additional liquidity mining incentives. While not a sustainable for the long term, putting GHO liquidity yield at better than market rates would incentivize LPs to come with fresh stables to hold in the system until the other parts of this plan are worked through.

4 Likes

Agreed on the fixed borrowing and using wGHO as 1usd worth collateral. Driving demand for GHO should be top priority for Aave atm, since it is simply a bad look.

1 Like

The Stability Fee is is algorithmic interest rate model influenced by the oracle price of the Dai token.

As the Dai token exceeds $1, the stability fee increases, to incentive less minting, as Dai receeds $1, the interest rate decreases, incentivizing more minting.

Currently, the stability fee is presumably used to fund the DSR. This means users who mint DAI are paying the open market to hold sDAI. This allows the open market to purchase a stable coin and deposit it for passive yield.

The DSR was introduced two years after the inception of DAI. Prior to the creation of the DSR, the stability fee was as low as 0.5%. After the DSR, interest rates were observed to be nearly 20% for minting the token.

Prior to the DSR, the stability fee was not meant to be market competitive, it was meant to manipulate the minting behavior in efforts to support a peg.

There is currently two facilitators for gho. The over collateralized minting facilitor, and the flash minting facilitor. None of these facilitators are designed to interact with market competition, they are designed just to mint GHO.

Thus, interest rates and the interest model for GHO, especially within the two current facilitors, should not be designed with market competition in mind, but simple minting behavior manipulation.

Maintaining the GHO peg is essential, but I have concerns with staking GHO for lower borrow rates. I am skeptical there is a discount rate that is both profitable to Aave and receives significant interest from borrowers.

While a GHO Savings rate is generally useful and hard to game, a borrow rate discount targets a subset of sophisticated users who are trying to maximize capital efficiency and are already using Aave. These users have a simple calculus: “Do I lower my costs more by staking GHO or by selling GHO to pay back debt?”.

As such, these users are unlikely to create long-term growth. They will stake if the discount is generous, then dump the moment the discount gets cut.

2 Likes

Thanks for the proposal @Gauntlet. We are in agreement with the sentiment of the post and the current state of GHO. GHO stability is a critical issue and we believe the outlined suggestions are a good approach.

In regards to next steps:

  1. GHO demand should be a top priority. Strengthening the peg is crucial to build trust and encourage adoption.
  2. The proposal’s suggestions, such as staking GHO for lower borrow rates are well-thought-out and worth exploring. These ideas may attract users and provide additional utility to GHO.
  3. However, it’s important to consider the potential drawbacks. For example, staking GHO for a borrow rate discount might attract sophisticated users seeking capital efficiency. While this could provide short-term gains, we need to carefully assess its long-term impact.
  4. Additionally, liquidity mining incentives could be considered to attract LPs and fresh stables into the system, albeit temporarily.
6 Likes

What’s the status of this proposal?

As an update, we are still soliciting community feedback and plan on moving forward with Snapshot in the coming weeks if the community is interested.

Something must be done rapidly to fix GHO state. I like this proposal approach, specially the staking GHO to lower borrow rates part. That could bring lot of attention to GHO. But before peg issue has to be solved.

GHO launch has been kind of “disappointing” given Aave’s reputation and credibility. Finding how to maintain the peg on the long term sustainably is key. Market fit and usage will come along right after.

2 Likes

Why wait a few weeks?

There hasn’t been a single dissenting voice. At most there needs to be some further discussion about the specific parameters and implementation of wGHO as collateral and sGHO staking, but seems this could move to snapshot much sooner.

4 Likes

The whole GHO situation is a big mess. The community is fully blind to the things that are being developed behind the doors, by seeing all the proposals that are related to GHO one can tell that no one “owns” the strategy and everyone is trying to avoid the hot potato beginning with “Aavecompanies”.

2 Likes

100% agree. It has been a mess as a product roll-out goes despite it’s obvious success with product-market fit.

My current understanding is @TokenBrice and the liquidity committee have given themselves until the end of the month to correct the peg (not looking great tbh).

Post that, we should come together with a more strategic and long term plan and also provide more visibility into the options for governance to think through. The current situation is obviously not ideal for the protocol or for users.

The liquidity committee received 406k in funding back in October for a 3 month period to manage this process. They have already spent 150k with minimal results. I would be interested to hear further post-mortem from that team once this month has ended and further information about what their plans are going forward with the remaining budget.

4 Likes

I agree with the sentiment. The peg really needs a fix!
Please move and report to the community more transparent.

@MarcZeller or anyone else who feels responsible. Please also schedule the AIP for 0.5% rate increase. According to my understanding, it should have been already submitted, such that it can be executed on the 23.th of Nov. We can’t get it in time anymore, but at least lets keep the delay minimal.
See the agreement here: Aave - Futher Increase GHO Borrow Rate

1 Like

hello @josojo, here’s the proposal, [ARFC] Increase GHO Borrow Rate to 5.22% on Aave V3

with the “direct-to-AIP” process, I think we are still on track to publish on the 23rd. there’s actually room for interpretation here but I envisioned the 30-day delay between the execution of the precedent AIP and the publication (vote opening) of the new one.

But it also makes sense to consider the delay between the two publications. I’ll let the community debate this on the new ARFC thread.

1 Like