[ARC] BUSD Offboarding Plan


title: [ARC] BUSD Offboarding Plan
Author: @marczeller - Aave Chan Initiative
Dated: 2023-02-27


Abstract

This proposal aims to implement an offboarding plan for BUSD on the Aave V2 Ethereum market. The plan will be carried out in two phases over two months and will involve making it more expensive to keep positions open and less profitable to have liquidity deposited in BUSD. The proposal will be divided into two AIPs to implement sequentially.

Motivation

Recent developments between Paxos and the SEC have resulted in Paxos stopping the minting of BUSD. This will cause the circulating supply of BUSD to trend towards zero over time. Therefore, it is necessary to implement an offboarding plan for BUSD on the Aave V2 Ethereum market to encourage users to switch to another stablecoin and reduce the amount of liquidity in BUSD.

Implementation

This proposal will be divided into two different AIPs, each corresponding to one of the two phases of the offboarding plan. The AIPs will be implemented in the order of the phases.
both AIPs will deploy and propose adopting new interestRate strategies for BUSD on Aave V2.

AIP 1

  • Decrease uOptimal from 80% to 45%
  • Increase reserveFactor from 10% to 30%
  • Increase base rate from 0% to 3%
  • Increase slope 1 from 4% to 7%
  • Increase slope 2 from 100% to 300%
  • Duration: 1 month

AIP 2

  • Increase reserveFactor from 30% to 99.9%
  • Increase slope 2 from 300% to 1000%

The increased reserve factor will allow the Aave Treasury to be one of the last aBUSD holder. Allowing current aBUSD holders to have liquidity to exit the reserve during the offboarding execution. Once most, if not all, positions are closed or liquidated, the ACI will propose to use aBUSD reserve to clear out potential excess debt, close remaining positions, and lastly, convert the remaining aBUSD to another stablecoin.

Process

If the current ARC gathers community consensus, a Snapshot vote will be published to greenlight the offboarding plan followed by the first AIP if a YAE outcome is the snapshot leading option.

Conclusion

Implementing this offboarding plan for BUSD will help to reduce the amount of liquidity in BUSD on the Aave V2 Ethereum market and encourage users to switch to other stablecoins.

The plan will be carried out gradually over two months to minimize the impact on users and ensure a smooth transition to other stablecoins.

The ACI believes that this is the most reasonable path forward given the recent developments with Paxos and the SEC.

10 Likes

I generally support the direction of this proposal but:

  • I’m wondering why Rf isn’t increased to 99 directly, but only after a month? On Fei the instant 99 bump seems to have worked quite well and if the goal is to offboard i guess suppliers shouldn’t be incentivized by a high apy.
  • what exactly does a 1k slope 2 mean in terms of apy on max utilization? To me seems quite agressive(300% already is) and perhaps unnecessary as 99rf will encourage withdrawals and increase apy for borrowers organically.
6 Likes

also in agreement with @sakulstra on a more aggressive 99% RF bump to encourage a faster off-boarding

3 Likes

hello and thanks @sakulstra & @oneski22 for your replies

The solution to do it in two months is a consensus between a rapid execution and letting user enough time to react without too much penalty, in a ideal world, most users will repay and withdraw during the first month and little to no positions will be liquidated or occurs losses.

with 99.9% RF & 1000% slope2 the maxVariableRate would be 1007% APR for borrowers and the APY for LPs will be close to zero.
The idea behind a very aggressive second phase is that it is likely that we will get “latent” passive positions of users that lost their private keys or are inactive.
Because we don’t know how long BUSD will remain liquid (and Coinbase de-listing is not encouraging) this second phase is designed to accrue debt rapidly on remaining borrowers and cause liquidation in a reasonable amount of time.

If governance agree with it, the snapshot can propose two options

  1. standard plan (ARC presented)
  2. aggressive plan (phase 1 & 2 merged)
  3. NAY
  4. ABSTAIN
3 Likes

Hi @MarcZeller,

Great initiative.

I think RF to 99.99% immediately is the right approach here. I am aligned with @oneski22 and @sakulstra on this aspect.

Moving the Uoptimal from 80% to 45% when utilisation is 56.45% is a rather aggressive move. If users are front of mind, the changing the Uoptimal to just above current utilisation and increasing Base + Slope1 to be greater than the current Slope1 parameter will provide encouragement for users to migrate funds.

I think the Uoptimal parameter should be changed from what is presented here. A borrowing rate of 10% at the Uoptimal should be sufficient for what we want to achieve. I support combining the proposals, but only if Uoptimal remains greater than current utilisation. Pushing users onto the more aggressive slope is not ideal and I think we should avoid this.

4 Likes

hello @MatthewGraham and thanks for your feedback

here’s an updated plan

Conservative option

AIP 1

  • Decrease uOptimal from 80% to 57%
  • Increase reserveFactor from 10% to 30%
  • Increase base rate from 0% to 3%
  • Increase slope 1 from 4% to 7%
  • Increase slope 2 from 100% to 300%
  • Duration: 1 month

AIP 2

  • Increase reserveFactor from 30% to 99.9%
  • Increase slope 2 from 300% to 1000%

Aggressive option

AIP 1

  • Decrease uOptimal from 80% to 57%
  • Increase reserveFactor from 10% to 99.9%
  • Increase base rate from 0% to 3%
  • Increase slope 1 from 4% to 7%
  • Increase slope 2 from 100% to 1000%
2 Likes

The solution to do it in two months is a consensus between a rapid execution and letting user enough time to react without too much penalty

I think it makes sense to have a closer look at the current situation & the levers existing.

There’s currently:

  • ~14M aBUSD on aave ethereum v2
  • there’s around 7.9M vBUSD on aave ethereum v2

I guess the goal here is to get users to repay the 7.9M vBUSD eventually.


Reserve factor

aBUSD is not collateral, so it’s factually not used to keep positions healthy. aBUSD positions only exist for yield farming.
Therefore one could consider bumping rf to 99.9%, effectively dropping supply apy to 0, and encouraging users to withdraw (as the single usecase is eliminated).
With this measure, i don’t see any negative side effects. The positive side effect would be, that by encouraging users to withdraw the borrow apy would organically increase.

Imo this would be a reasonable thing to do in AIP 1

uOptimal

Decreasing uOptimal will alter when slope 1 is reached, but not affect max interest rate.
Adjusting uOptimal, will move the apy within predictable bounds, but not suddenly inflate it.

As the goal is to encourage repayments, i would argue that decreasing this parameter could be reasonable in AIP 1 (possible even lower than 57%)

baseVariable borrow rate

Adjustments to base variable borrow rate will be reflected on maxBorrow rate. I think 3% bump could be justified.

slope1

Slope 1 essentially defines the apr at uOptimal, adjusting it will essentially flatten the curve, but not affect maxBorrow rate, so the rate still is within the predictable bounds of the current IR.

Collector

The increased reserve factor will allow the Aave Treasury to be one of the last aBUSD holder.

The collector doesn’t hold much funds(~12k). It could be reasonable to withdraw to BUSD to increase utilization slightly. It could still be redeposited if the reserve reaches max utilization, but in FEI this was also a concern that turned out to be unjustified.


Slope2

I’m not sure if this is a good idea to do, if yes it should not be done so aggressively imo.
Looking at the top vBUSD holders Aave variable debt bearing BUSD (variableDebtBUSD) Token Tracker | Etherscan seems like most of them have healthy positions with a variety of stables borrowed and BUSD not being overly exposed, so to me doesn’t seem to be BUSD shorters per se.
If one really wants to increase the slope2, perhaps would be more reasonable to slightly increase it to 150 or similar, and on aip2 consider bumping it to 300 (which is close to 1% per day!, without compounding).

I think the 1000% apr is far to aggressive.
This would mean ~3% payed per day(before compounding!), forcing borrowers into liquidation at warp speed.

1 Like

We will support an aggressive option, increasing the RF to 99.9%

Thanks to @sakulstra for the input and @oneski22 & @MatthewGraham for community support.

2 Likes

Thanks for this proposal, @MarcZeller - the Gauntlet team will review.

A high-level comment to support the discussion, especially related to the timing aspects of it.

Even if a different order of magnitude, BUSD’s case is similar to renFIL’s on Aave v2 Ethereum. A basic asset dynamic is affected (minting), which technically disturbs the fundamentals on the asset in Aave.
Whenever that happens, as proposed & already executed, off-boarding needs to start immediately, but the timing of further steps is also important.
When variable borrow debt (e.g. BUSDs) grows due to its borrow rate dynamics, that means that at some point, that “growth” will need to be covered by somebody, being 1) the borrower itself via repayment or 2) an eventual liquidator to cover the debt of the liquidated user and take the collateral. In both cases, the borrowed asset (BUSD) needs to be acquired before interacting with the protocol, one way or another.
Not having new minting enabled means that all BUSD liquidity to cover the whole debt of Aave needs to come from existing liquidity, both on Ethereum and other venues.

What the previous explanation means is that no matter if collateral or not, an asset with no active minting mechanism should be off-boarded completely as soon as possible, to reduce any type of technical risk of the protocol, in this case, being unable to source from anywhere BUSD.
So it is more optimal to execute the main levers (e.g. RF update + main update of rates) without any delay to “give time to users”, especially when there is no radical measure, like for example can it be lowering Liquidation Threshold in other cases.

That being said, user experience should be taken into account, even if subordinated to the health of the protocol, so:

  • RF increase to 99% is clearly required, and as @sakulstra points out, there is no rational reason not to proceed immediately with it. This is a technical situation that requires exercising the levers available on the protocol, completely unrelated to the fee-collection nature of the RF.
  • Setting really high slopes is by general rule not recommended, as the dynamics become pretty chaotic from a high-level perspective. Decreasing the optimal point of the rate strategy is acceptable because it is just an abstraction of the rate dynamics, but (risk providers should give a more precise recommendation) increasing effective rates above ~150% total APY (considering base + slope 1 + slope 2) is something that most probably will not really help much and could create important problems for users borrowing BUSD.

The ACI wishes to express its gratitude to all participants for providing valuable feedback during this discussion.

It appears that there is a consensus to increase the reserveFactor to 99.9% as soon as possible rather than making a progressive modification. Additionally, feedback on the aggressiveness of slope2 and uOptimal tweaking has been taken into account.

Incorporating this community feedback, the “conservative plan” option has been scrapped. The following is the new consensus plan outlined in a single AIP:

  • Decrease uOptimal from 80% to 20%.
  • Increase reserveFactor from 10% to 99.9%.
  • Increase base rate from 0% to 3%.
  • Increase slope 1 from 4% to 7%.
  • Increase slope 2 from 100% to 200%.

If this AIP were enforced today, it would increase the APR cost of current borrowers to 101.25%, sending a sharp signal to repay without significant impact on their positions.

With these new parameters, if the reserve utilization reaches 100%, the resulting APR cost will be 210%, which is much lower than the previously proposed aggressive plan.

Upon analysis of the vBUSD debt token holders, it appears that most top positions are relatively active addresses and are expected to manage their positions actively.

The ACI would like to thank @bgdlabs for providing valuable tooling support for rateStrategy simulations.

5 Likes

Gauntlet Analysis

From a market risk perspective, Gauntlet supports the updated plan.

Current utilization is ~56%, so this would send the variable rate to ~100% immediately. BUSD is not collateral on Aave v2, nor can any additional supply or borrows be made. 90% of the total borrow amount is done by 100 users, and the top 5 borrowers account for almost 30% of the total borrows.

The top borrowers are usually fairly active (elastic), suggesting that the increase in borrow rate could be effective. If BUSD debt does not decrease meaningfully following this proposal, then the community may consider toggling rates higher.

2 Likes

It looks like there’s already significant withdrawal of BUSD by suppliers and repayment by borrowers, so users have been acting based the expectation of future offboarding. At the same time, it looks like the utilisation is up to 90%. @MarcZeller From a user experience perspective, are we confident that the remaining loans will be paid off before suppliers want to withdraw liquidity?

Overall, @Michigan_Blockchain supports this proposal.

1 Like