[Pre GIP] Add LUSD as borrowable asset

Abstract

This proposal suggests adding an LUSD pool on GearBox, enabling the borrowing of the asset. LUSD/3crv is already one of the top leveraged collateral on GearBox. This proposal paves the way for a more synergetic relationship between GearBox and Liquity.

Proposal Subtype: New Pool (LUSD)

ERC20 token address

LUSD Token Contract - Etherscan

Token source code

The code for the whole Liquity Protocol, LUSD included, is open source: Github repository.

Audit of token

Please note that Liquity/LUSD are powered fully by immutable contracts: therefore, attack vectors such as infinimint are technically impossible. No multisig is involved in Liquity’s protocol operations.

The Liquity Protocol received four security audits, additional risk assertions, and economic modelings before launch. All can be found in the Liquity documentation.

Introduction to Liquity & LUSD

Liquity is a protocol offering interest-free leverage on ETH. The LUSD stablecoin, the system’s output, offers some genuinely unique features, the leading one being the trustlessness of the whole Liquity protocol. The Liquity protocol is geared for sturdiness: dual oracle system, various fallback mechanisms, recovery mode, and more, making LUSD the most resilient and credibly decentralized stablecoin available on the Ethereum network.

Project: https://liquity.org/

Twitter: https://twitter.com/liquityprotocol

Documentation: https://docs.liquity.org/

App (list of front-ends): Liquity | Frontends List - Use Liquity

Discord: Liquity

Motivation

While LUSD/3CRV was a great fit for GearBox product launch, with still $5.3M used as collateral, the impact for Liquity was neutral – not because of GearBox product or intent, but because of the parameters:

Since LUSD is not available as a borrowable asset on GearBox, users leveraging LUSD/3CRV have to borrow USDC/DAI to supply it. While the pool TVL improved, its token balance remained, at best unchanged.

The addition of LUSD as a borrowable asset on GearBox would heavily mitigate this side effect, as LUSD/3CRV leveragers will be able to borrow LUSD and even lever up LUSD/3CRV using only LUSD – helping rebalance the pool.

Besides, having LUSD as a borrowable asset on GearBox paves the way for further collaborations if they are to GearBox’s community interest. Along with Aave, the GearBox LUSD supply market would be one of the only places to earn a yield on LUSD with limited exposure + a native yield (LUSD/LUSD).
image

Proposal subtype: Interest rate model for pools

The interest model for LUSD should use parameters similar to the USDC/DAI markets on Aave, with the interest rate formula targeting a ~85-90% optimal utilization rate. Since the usage of LUSD on GearBox is limited to opportunities to keep LUSD liquid with no lock (LUSD LP supply), the parametrization can optimize capital efficiency.

The main differences would be in the pool caps, where a 15M pool max is suggested, and ideally, min and max pers borrow halved compared to the DAI/USDC market – to ease the market’s growth.

Additional context and information

You’ll find in this section additional information added to the proposal that were not fitting the proposal frameworks used.

Implementation Details

LUSD to be added as an asset that can be supplied and borrowed on GearBox. 15M max cap, 50K min per borrow, 500K max per borrow, with a utilization rate formula targeting 85% utilization rate.

If possible, it would also be worthwhile to consider crafting a streamlined migration flow for current LUSD/3CRV depositors borrowing USDC or DAI: a one-click migration recipe enabling them to switch their borrow to LUSD.

LUSD Supply Market Incentivization

To grow the LUSD supply market, the GEAR Liquidity mining rewards could be targeted at achieving 5% APY on a 15M LUSD pool.

It is competitive compared to the LUSD Stability Pool; currently, around 3% APY on a monthly average, the primary yield source for LUSD holders (100M+ LUSD deposited) is also usually seen as the safest (since it’s just another contract within the Liquity protocol).

Forward Thinking Considerations

  1. Synergies with ReDefine
  2. The Chicken Bonds bLUSD / LUSD-3CRV Curve pool is a 20M TVL Curve pool with attractive CRV yields depositors should be happy to leverage with GearBox, if LUSD is available to borrow.

Information on Proposal Writer

I am TokenBrice, Liquity DeFi Strategist handling the protocol relationships with other DAOs and LUSD/LQTY liquidity strategies. I’m also a Gearbox fan since its inception, I took part in the Credit Account minting and covered it on my DeFi show back then (French).I fight to maximize the range of DeFi services that are available in an unstoppable fashion.

8 Likes

I am 100% for this.

One further comment on this, LUSD was constantly in the 11-14M$ bracket before it reduced due to price fluctuation. By creating a LUSD/LUSD borrow, we can neutralise the effect of LUSD peg. This will help grow the LUSD CVX strategy

3 Likes

This is awesome, thank you for dropping in @TokenBrice! Big fans of ur educational work & Liquity <3

  1. LUSD is great, it would be awesome to have it as a pool mostly to support decentralization. Meaning, it’s nice to help grow it and maintain peg, so it becomes a more dominant stablecoin in DeFi. For that, Gearbox DAO could consider some emissions of GEAR as long as it’s on par with “value” the protocol gets (as a source of leverage liquidity and as a great asset to be working with).
  2. If the target is 5% on 15M pool, that’s 750K USD annual reward (monthly 62.5K USD). Comparatively speaking, for USDC and DAI the things now are:
  • DAI: 20M and 3.5%
  • USDC: 40M and 3%

As a result, 5% on 15M pool seems like an average level at this point. Plus, whatever the organic % will be (likely similar utilization curve to the stables above). Seems good overall.

3 Likes

Hi and welcome!

I’m on board with adding LUSD as an asset for lending/borrowing but wouldn’t this only make sense to farmers if the Chicken Bond pool was added aswell? I would not be supportive of providing GEAR rewards to lenders of LUSD unless there was an appropriately profitable farm for borrowers to use considering borrowing APR on aave is ~3% and LUSD+FRAXBP/3CRV pools are 2.54% and 1.35% on Convex respectively.

I’m not fully across Chicken Bonds so can’t comment on risk associated with the blusd pool.

Full disclosure: RiskDAO and B.Protocol communities are very much overlapped, and B.Protocol and has a tight collaboration with Liquity. In particular, all the Chicken Bond LUSD are deposited in an instance of B.Protocol.

Hence, of course we have positive sentiment towards LUSD, however there are also some negative subtitles that the Gearbox community should be aware of.

While LUSD is a prominent collateral in Gearbox, and overall the RiskDAO ranks LUSD as a decent stable collateral, it is important to notice that the liquidity of LUSD is annoyingly asymmetric.
While it is relatively easy to sell LUSD, it is notoriously hard to buy LUSD. In when liquidating LUSD debt, the liquidator has to buy LUSD.

1inch shows that it is currently possible to liquidate only $5.5m of LUSD debt in one shot
image

while it is possible to liquidate over $50m of LUSD collateral in one shot.
image

the situation was much worse in the recent past, where it was barely possible to liquidate $2m of LUSD debt in a single trade.
Even worse, the amount of LUSD that can be liquidated becomes smaller as the price of LUSD (the debt asset increases), as the demand to buy LUSD to close user positions on Liquity increases, and the demand to mint more LUSD decreases.

This means that the liquidation thresholds of volatile assets such as ETH and WBTC will likely be much lower in the LUSD pool.

@mugglesect as for stable assets, it should be noted that in the current Gearbox oracle implementation (which follows the industry standard) the price of the LUSD/3crv LP token pool is basically the minimum of LUSD, USDT, USDC and DAI. Since LUSD is the least stable of both, and tends to be traded > $1, users will actually be safer to leverage LUSD/3crv vs DAI or USDC.
This could be somehow mitigated if the oracle is changed so it would take to min price of LUSD and 3crv (as 3crv lp token itself > $1).

Conclusion

LUSD is more risky as a debt than as a collateral, but this could be mitigated by setting up more conservative risk parameters.

2 Likes

If the suggestion is to add Chicken bond as a debt asset (rather than a collateral asset), then it should be noted that there is no available oracle for it. And while floor price might be considered a good solution for it as a collateral, it is much more problematic when having it as a debt asset.

Do you mean the bLUSD / LUSD3CRV Curve pool? A significant obstacle to adding it to Gearbox is that it is a Crypto pool, where building a general purpose oracle is complicated. Perhaps a special-case price feed can be designed, given bLUSD has a special relationship to LUSD, but that would require more research and dev bandwidth.

1 Like

Adding the cvx_bLUSD/LUSD-3CRV as collateral on GearBox to harness a strategy enabling to leverage it through LUSD borrowing could be interesting to secure borrowing demand indeed.

As already pointed out, the main limitation is the oracle since the bLUSD / LUSD-3CRV is a Curve V2 pool made of bLUSD + another Curve V1 pool (LUSD/USDC+USDT+DAI).

However, if there is interest, I think a custom solution could be built that would provide a price merging two data sources:

  1. bLUSD [50% of the LP] can be burned for LUSD through Redemptions at the bLUSD Floor Price (1.1070 LUSD currently).
    Redemptions offer a way to convert back bLUSD to LUSD without using an AMM, securing a minimal and rising floor price. The amount of LUSD obtained corresponds to the share of the Permanent Bucket matching the % of bLUSD supply burned. LUSD Chicken Bonds Redemption Doc
  2. LUSD/3CRV [other 50% of the LP] is a more usual Curve pool, V1 meta pool pairing LUSD with the 3pool LP token, so sourcing the price for it should be easier than for the base pool.

one subtle issue with the floor price is that it is not really the floor price.
As the ETH part in the stability pool is not redeemable by the chicken bond redeem function. Hence, in extreme scenarios (when the stability pool is temporarily imbalanced), the redemption will be below floor price.

Generally very amenable to add LUSD as a borrowable asset (on a side note, we need to unify the nomenclature used, think an outsider would have a hard time parsing what side of Gearbox we are talking about with “debt asset” and “collateral asset” and so forth)

@yaron in terms of the issues you mentioned (LUSD is difficult to buy), is it possible that the addition of LUSD to Gearbox as a borrowable asset could affect this equilibrium?

Because if you can borrow LUSD in Gearbox, now there is a viable way to arbitrage it back to $1. And indeed, even participating in the curve pools while borrowing LUSD would help alleviate the imbalance between LUSD/3crv in those pools. Or are there other reasons to think that this will be a relative non-factor?

2 Likes

in theory yes, because it gives you an option to short LUSD when it is > $1.
But it is not really an arbitrage, just a short position, betting it will go back to $1 quickly.
In theory, if gearbox would let you do x10 on it vs USDC collateral, than you could make easy 30% whenever it is, e.g., > $1.05. Short at $1.05 and sell at $1.02 (well, not that easy because max short might get you liquidated).

But the problem is this:
typically the deppeg is worse when ETH is crashing, and we don’t know how quickly the shorters will come in such a situation. And if these short positions are not automated (which I assume today they are not), then it might be too late to help liquidations.

2 Likes

So from a risk perspective, what do you suggest the LTV should be? Or is it too risky to onboard as collateral at the moment?

1 Like

It is definitely ok to on board with the right liquidation thresholds.

I would recommend to start with 10% lower than the liquidation thresholds that are used for dai and usdc.
Overtime there might be room for 3-5% increase and optimization.

10% lower means that if the threshold of eth vs usdc is 80%, then eth vs lusd should be 72%.

For assets with 25% threshold, it is fine to keep as is.

3 Likes

Hello Gearheads, and thanks for the valuable feedback provided so far!

@yaron It seems like a parametrization we can get behind; it would already be a massive improvement compared to the current flow when GearBox cvxLUSD-3CRV leveragers borrow USDC/DAI and thus screw the balance of the pool even further.

So, allow me to attempt to recap the feedback so far so that we have a better idea of the following steps:

  1. In principle/alignment, the community seems happy to welcome LUSD as a borrowable asset.
  2. Adding LUSD as borrowable on GearBox will help alleviate the pressure on the peg
  3. However, there are concerns regarding the available liquidity on LUSD: there is enough liquidity to dump > $50M LUSD at once, but on the buying side, the limit is more around 5M
  4. The other main concern was the yield source: currently, only cvxLUSD-3crv is supported as collateral on GearBox, and the base APY is quite limited (~1.5%)
  5. To address this, the community also considered adding cvxbLUSD/LUSD-3crv, which outputs a ~9% base yield on Convex.

Regarding 3 - as @yaron pointed out, the situation has already improved tremendously over the past months, and further liquidity-related initiatives are being pursued on LUSD that should help further.

Regarding 5 - I highlighted above a pathway to solve the main technical limitation for this flow - having a reliable price source for cvxbLUSD/LUSD-3crv. Now to move things forward: should the addition of cvxbLUSD/LUSD-3crv as collateral on GearBox be added to this (revised) proposal too? Or would it be better as a separate proposal?

I hope that helps; looking forward to your answers to amend the proposal accordingly.
Cheers.

1 Like

Note that users would still prefer to borrow DAI vs LUSD-3CRV, as their risk for liquidations would be higher with LUSD debt.
For once because the liquidation threshold is lower, but also because the oracle price of LUSD-3CRV is pegged to min(LUSD, USDT, USDC, DAI), so its price does not increase when LUSD increase, and thus liquidation might happen.

But this is just a side note.

For onboarding cvxbLUSD/LUSD-3crv as allowed tokens it’s better to create separate proposal. btw, sorry for dumb question, how I can get blUSD? coz if it requires some staking/withdrawing it will require building adapter for it…

1 Like

No dumb question @apeir99n I imagine that not everyone here is a Chicken Bond pro!

The short story is that you bond LUSD to accrue bLUSD over time
During that time, the bond can be canceled and the full initial LUSD amount recovered (that’s the Chicken Out)
Else you wait for enough bLUSD to accrue to be in profits, and you claim them, forgoing your bonded to the protocol in favor of the accrued bLUSD: that’s the Chicken In.

bLUSD is like a supercharged LUSD, offering amplified and auto compounding yields. Once you hold it, there are two paths back to LUSD:

  1. Swapping on the bLUSD Curve pool
  2. Redeeming bLUSD for LUSD with the Chicken Bonds protocol itself at the floor price.

Note taken on the separate proposal for bLUSD cvx LP as collateral, I’ll start working on it.

To wrap it up here’s a neat infographics giving you an overview of the bLUSD-related strategies: