[TEMP CHECK] Safety Module Upgrade Part II - Asset Diversity, SM Categories & Slashing Updates

Governance
1

title: [TEMP CHECK] Safety Module Upgrade Part II - Asset Diversity, SM Categories & Slashing Updates
author: @Llamaxyz - @dydymoon & @TokenLogic
created: 2023-04-13

Summary

This publication presents the community the opportunity to discuss Safety Module (SM) upgrades that enhance asset diversification, introduces risk tranching and potential growth synergies.

Abstract

This publication presents the community with the opportunity to expand the assets accepted within the SM. Currently, the SM is overly concentrated in AAVE via direct AAVE staking and ABPT staking which is a Balancer v1 80/20 AAVE/wETH liquidity token.

This publication seeks to explore the community’s appetite to diversify the SM asset base and vary the portion of users’ funds which can be auctioned in the event of a shortfall. Currently, 30% of user deposits are at risk when depositing into the SM.

Motivation

@bgdlabs is currently proposing to upgrade the SM architecture to support the launch of GHO, by enabling stkAAVE holders to access cheaper GHO borrowing rates. Further to this, Llama has kick started the conversation around migrating AAVE liquidity from Balancer v1 to v2 in this forum post.

Several comments relating to the AAVE concentration and auction mechanism have been made on the forum which can be found here, here and here.

This publication presents three distinct categories for inclusion in the SM.

  • Single Asset
  • Volatile Asset Liquidity Positions
  • Stable Asset Liquidity Positions

The net change to the existing implementation is the introduction of a third category, Stable Asset Liquidity Positions. The existing AAVE and ABPT represent Single Asset and Volatile Asset Liquidity Positions.

Although any Liquidity Provider (LP) position can be included, G-Uni, crvLP, BPT etc… This publication will initially focus on the overarching concept and how BPTs could be integrated.

BPTs were selected due to Aave’s relationship with Balancer DAO and the veBAL holding which the community may elect to utilize here. It should be stressed, Aave DAO will elect which assets are to be included and the varying levels of slashing for each category.

Assets considered for inclusion should have minimal counterparty risk and display great on-chain liquidity amongst other considerations such as price correlation with Aave Protocol’s TVL during varying market conditions.

Llama proposes three initial asset categories to be created within the SM.

The slashing parameters update aims to reduce the price impact on the AAVE token in the case of a shortfall event. We propose adding different levels of slashing to each category, as follows:

Single Assets

The current single asset, with 30% slashing, is AAVE. This proposal introduces other assets to this category with the goal of reducing the concentration of AAVE inside the insurance fund.

Possible assets for consideration inside the SM Single Asset Category include, not limited to:

  • AAVE
  • ETH and/or wstETH

If the community elects with the introduction of additional assets other than AAVE, then each asset could be considered individually. This publication retains the 30% slashing feature for Single Asset.

The inclusion of wstETH may attract additional incentives from Lido DAO, therefore Llama prefers having the ability to distribute multiple rewards types with external teams able to distribute their own token. Something similar to the Emission Admin role on Aave v3 will suffice.

Llama preference is for wstETH over wETH due to improved capital efficiency by enabling users to earn an overall higher APY from being a SM depositor. wstETH has become a major trading pairing token on Balancer and it has ample liquidity across various pools to route swaps through. Additionally, the stETH-ETH discount has reduced significantly (0.9938:1 for 100K ETH sold, which also demonstrates very deep liquidity.

Llama would also encourage Lido DAO to indicate any potential support for this type of integration between the two protocols as inclusion in the SM would act to help retain or grow wstETH adoption via the creation of a unique use case for wstETH relative to other LSTs.

Max slashing proposed: 30% retained for single assets.

Initial assets proposed: AAVE / ETH and/or stETH

Risks Considerations

If LSTs (such as wstETH) are included in the SM, it creates a price divergence risk which, if sustained, can reduce the cover in the SM and create an unfair distribution of the rewards between ETH & wstETH. Post redemption the main risk consideration for LSTs becomes counterparty risk which if it materialized would lead to a significant discount to the LST price in ETH terms.

  • To limit the de-peg risk, the SM could stop incentivizing below a defined ratio, so the total value would then be deduced from the total cover value. This acts to mitigate counterparty risk without requiring governance intervention.

Based upon technical feedback relating to the ease of implementation, Llama may revise this section of the publication.

Volatile Liquidity Provider Tokens

In addition to the replacement pool for the 80/20 AAVE/wETH pool on Balancer v1, to be determined by an earlier publication. This publication presents two other pools of strategic importance to Aave DAO for inclusion in the SM.

This publication presents the prospect of creating an 80/20 GHO/wETH (or wstETH) which offers less impermanent loss than a 50/50 pool whilst also providing very deep liquidity for risk off events which see spikes in stable coin swap volumes which can often lead to stable coins trading above $1.

For the following categories, up to 75% of the current budget could be allocated.

Max slashing proposed 45% for volatile BPTs

Initial volatile BPTs proposed: B-80AAVE-20wstETH (or Part I proposal results) & B-80GHO-20WETH

Even if not yet created, this is probably the most interesting addition to this category as the trading volume of this pool will generate an important yield for LPs. The second pool could also be converted into a Balancer Core Pool by replacing ETH by IBTs as described in Part I.

Stable Liquidity Provider Tokens

In this third category, this publication presents an initial pool consisting of the Balancer 3pool and GHO. Like other categories, in the future, additional assets can be added. If the community elects to launch BB-A-USD/GHO on Balancer or a stable coin pool on Curve, then this section can pivot to include the Curve pool.

Max slashing proposed: 60% for stable

Initial Stable BPT proposed: B-3Pool/GHO

This pool introduces a stable asset to the SM design and generates a significant amount of the coverage as 60% of the TVL can be slashed with very low price impact across the broader ecosystem. This is especially the case if something like Cowswap’s auction mechanism was to feature in the new SM auction mechanism.

By integrating a GHO pair to the SM, Aave is essentially helping bootstrapping GHO adoptions through incentivised liquidity. Aave will also earn revenue from the GHO borrowing rate.

If the community elects to proceed with 3CRV/GHO, then this would increase the SM smart contract surface area through diversifying exposure concentration away from Balancer to include Cuve Finance and potentially Convex Finance risk as well.

There is also the option of adding additional pools which could include GHO, MAI, LUSD, FRAX and others once live like crvUSD etc…

This proposal can be implemented independently from the following parts. In this case, “classic” gauges would be advised. Otherwise, the strategy & smBPT gauges discussions will be outlined & discussed in Part III.

Technical Specifications:

  • Remove the BPT 80/20 from the current main module (30% slashing)
  • Deploy two other modules on the SM for the new categories (45% & 60% slashing)
  • Implement the new assets voted on Part II for the new categories
  • Create the new pools depending on vote results

Voting Options

YAE - Implement Changes
NAE- Rework Proposal
ABSTAIN

Disclaimer

The Llama is not compensated by any of the mentioned communities outside of Aave.

Llama is an unpaid delegate within the Balancer ecosystem.

Members who contributed to this proposal are not angel investors or advisors to any of the mentioned communities but some do hold small holdings in those communities tokens.

Copyright

Copyright and related rights waived via CC0.

References

  1. [TEMP CHECK] Safety Module Update Part I - Migrate AAVE/wETH Balancer v1 Pool to Balancer v2
  2. [TEMP CHECK] Safety Module Upgrade Part II - Asset Diversity, SM Categories & Slashing Updates
  3. [TEMP CHECK] Safety Module Upgrade Part III - Enable gauges on BPT in Safety Module (smBPT)
  4. [TEMP CHECK] Safety Module Upgrade Part IV - Incentives Management Upgrade
  5. [TEMP CHECK] Safety Module Upgrade Part V - veToken Holding Management Framework
  6. [TEMP CHECK] Safety Module Upgrade Part VI - Future considerations
7 Likes
2

Great post – thanks for starting the discussion on asset diversity and slashing parameters.

Two key points I’d like to raise:

1) I think we should consider the safety module not just as a mitigation tool for a shortfall event, but also as a way to optimally allocate $AAVE incentives.

Our two current choices of collateral for the safety module – single-sided AAVE and AAVE/ETH – have direct benefits to the DAO and to tokenholders. In both scenarios, they provide a direct opportunity for Aave tokenholders to generate yield. And with the latter, it also meaningfully increases liquidity for the AAVE token.

While ETH and wstETH are pristine collateral from a security perspective, incentivizing them in the safety module offers no strategic benefit to the DAO. Maybe one could argue that we really should have at least some of it for safety – if that’s the case, totally fine, but we should be thoughtful and only allocate limited incentives (i.e. the bare minimum to achieve whatever level of security we’re aiming for).

On the other hand, liquidity pool tokens that are GHO or AAVE pairs have strategic importance to the protocol. Incentivizing those should be the priority.

2. I think the max slashing should be lower on single-sided AAVE or AAVE liquidity pairs (30%) and higher on non-AAVE collateral (45%-60%).

Was there any reasoning behind the specific 30%/45%/60% tiers described above and the assets assigned to each one? If so, would be helpful to hear.

I understand that the whole reason for the 30% max slashing in the first place was to align incentives – $AAVE token holders would be the ones voting for a slashing event, and $AAVE holders within the safety module would obviously be hesitant to slash themselves. They’d never vote to slash the majority of their own holdings, so it makes some sense to offer a cap and make them more amenable to self-slashing for the benefit of the protocol.

However, ETH or wstETH depositors do not have the same conflict of interest. They are not AAVE depositors. I don’t think we need to (or should) offer the same 30% max slashing cap?

I think the max slashing for non-AAVE deposits would likely need some iteration, since the higher you set the cap, the less depositors you’ll have. This is offset by the fact that you’re potentially slashing a greater % of those fewer depositors (so overall safety dollars may still be higher). There is a point along that curve that would provide the maximum safety, and we’d need some experimentation to find it. That 45-60% range is somewhat arbitrary, but does feel right as a starting point.

1 Like
3

Do you guys think the main concern here should be total yield for depositors or “spread” ie additional yield depositors can earn in exchange for the risk they are taking? The spread required to incentivize stakers is a function of liquidity prem + slashing risk, and slashing risk is equal for any asset while the liq prem itself is likely a function of price volatility. So in the context of choosing which assets to include, we think the ones that minimize this spread, as smaller spreads means more efficient spending by the DAO, while maximizing liquidity availability. But curious your thoughts on that framework.

Based on the framework, we posit that stables and ETH/stETH would require the lowest additional spread to depositors. However, open-minded to other arguments for most efficient assets for the SM.

Do you think the higher slashing cost will simply increase the premium required for stakers? Not sure what the value add is here unless it gets better coverage for depositor assets at a cheaper cost for the DAO. I can see how the attempt here is to balance allowing riskier assets with a higher penalty (slashing percentage) to get more liquidity in the SM. Again, think the focus should be on attracting liquid assets sustainably rather than going out the risk curve.

2 Likes
4

Hey @BarryLime thanks for the feedback !

Definitely agree here, and this is a major part of the SM Upgrade proposal. You can find the details in Part IV, but I’ll provide more details below.

Incentivizing single assets other than Aave make sense especially wstETH of the auto compounding balance which will help increase the SM cover amount.

However, I agree about the limited incentives and it’s important to remember that we based the proposal to remain below the current budget. If the DAO was to consider that growing the budget can be beneficial since it would increase the cover & the liquidity on native assets, updated estimations can be provided. Note that GHO could also be considered in the single asset category.

On the estimation proposed, 25% of the current SM reward budget is allocated to single assets (275/1100) over which 200 are for StkAAVE & 75 for ETH or wstETH.
The remaining 75% could be used to incentivize pools including native assets using the vote incentives leverage when profitable.

Sure, creating different slashing parameters for each category also creates risk tranches representing different potential earnings and increasing the total SM cover. Agree that we could consider reducing the slashing on single assets, but this would also impact the max cover, and current users seem comfortable with 30% so far.

The reasoning was to keep the single assets slashing unchanged since the design is very similar to the current one, except that several assets can be incentivized & that the stkAAVE APR would be lower (Currently around 3,7% with 200 AAVE/day vs 6,2% APR now with 550 AAVE/day)

For volatile BPT, the strategy is to aim for a range APR between 8% & 12% below a defined TVL threshold. Above that, the APR per asset in the SM might be lower depending on the Aave DAO strategic voting power capacity.

Considering that depositing in a volatile LP has more risks and would offer a better yield than single assets, the goal was to create an intermediary slashing category, with initial considerations between 50 to 60%, as we were aiming for a 75-80% slashing on the Stable BPTs. However, following several feedback, we reduced the volatile one to 45% and the Stable one to 60%.

For the Stable BPT, the strategy aims for a range APR between 18% and 22% APR below a defined TVL threshold. As for the volatile category, above that threshold, the APR per asset in the sm might be lower depending on the Aave DAO strategic voting power capacity.

Offering a sustainable 20% APR on stables while supporting GHO liquidity growth should attract an important amount of depositors which could be willing to take a higher risk of slashing since the position would be relatively safe. Considering that the APR would be above 20% most of the time, having a 60% slashing should be ok and will considerably contribute to increasing the SM cover value.

Note that we base our estimations on the current SM incentives budget value, however the DAO could definitely vote for a budget increase/decrease or add a complement using GHO profits once live.

It’s not the majority of their holding since it’s max 30%, and i’d say never say never here. It didn’t happen so far because the shortfall events were limited and small enough to be contained by the treasury without a significant impact, however, it doesn’t mean that it won’t happen and Aave holders might just not have a choice to save the protocol at some point.

Open to discuss changes for these parameters, and slashing can be a midterm solution but this solution is not sustainable long term in its current management because it requires to keep incentivizing depositors indefinitely.

As mentioned in Part VI, I strongly believe that the DAO should, at least partially, self insure over time, and could even use the POL positions to generate additional revenues and strategic voting power, but we’re not there yet.

Thanks for the feedback @John_TV_Locke !

As explained above, we based this strategy on several parameters, and estimated yield for depositors is one of them, as well as the spread between incentives spent by the DAO and emissions received for it. As for the assets selection we initially focused on the native assets liquidity growth and with a limited selection to easily implement it but one of the future considerations of part VI is to define a more detailed framework to support more assets.

This answer was co-written with @TokenLogic.

If no additional comments arise, the vote will be published on snapshot over the weekend, with voting to start on Monday 29th.

5

Hi Everyone :wave:

This proposal has been progressed to Snapshot.

https://snapshot.org/#/aave.eth/proposal/0x8e60bbebf59a6911cd0d137aa06780aefc5b5b0cd0f8ccfa12427cf153f58b05