[Signal Request] Adjust YFI-A Debt Ceiling

Overview

YFI-A is currently at ~75% utilization, DAI from YFI has doubled in less then 12h since the last increase of the Debt Ceiling. Seems there is a huge appetite for YFI as collateral.

This poll is about changing the DC to allow more DAI from YFI.

Pros:

  • more fees collected
  • more DAI in circulation

Cons:

  • we already have 5% of all YFI in existence locked in vaults - people are getting worried about the YFI-risk already (great post here). maybe @Primoz can chime in here? can we do this without changing the SF?

Ceiling adjustment

Here are a few polling options with +50%, +100%, +200%

  • 60 MM
  • 40 MM
  • 30 MM
  • 20 MM (no change)
  • Abstain

0 voters

Next Steps

Poll will run until December 3rd and depending on the result will move on-chain assuming the outcome of the poll deems it necessary.

Update

In the meantime another signal running about multiple options moving forward with YFI-collateral has been opened by @Primoz also considering SF/LR adjustments.

5 Likes

So we are currently running some risk numbers for each vault type since we need to propose maximum debt ceiling parameters for MIP27 implementation (DC IAM) next week anyway.

The maximum debt ceiling figure should in our opinion be a function of liquidity, where DEX liquidity carries a higher weight, because auctions can be more efficient. After performing volume analysis for YFI we determined maximum DC for YFI to be 46m DAI, based on a selected ratio of 30% to 60% of CEX+DEX daily volumes. At 175% LR this would mean Maker would hold more than 11% of YFI market supply, which is not insignificant (ETH has only 2.3%). But this high ratio is mostly due to high YFI volumes compared to its market cap.

This means we don’t have anything against increasing DC towards 50m if volume metrics for YFI don’t worsen. However risk premium will rise if we were to attempt such an increase.

Below are stress test results for YFI-A, which shows risk premiums based on different price drop and auction slippage inputs at a debt exposure of 46m DAI. The conservative base case inputs we are using for YFI is -40% daily price drop and -40% auction slippage which yields 15% risk premium for 46m debt exposure. For current 20m debt exposure risk premium would measure around 7% (lower debt exposure normally leads to lower auction slippage).

This high risk premium is due to the largest vault (9.8m DAI debt), which has only 223% CR at the time of writing. This means that if we have a higher drop in price, this vault would quickly get liquidated and could lead to significant losses if auction slippage is severe.

Now, there are a lot of “ifs” here and that is why I wanted to show this matrix so it gets easier to decide for appropriate risk tolerance. But we need to be aware that such an increase could lead to double digit SF, which could dissuade current vault users from borrowing. As said in any case we will need to increase SF towards 7% which would equal Maker with Aave, from where a lot of users came initially. We can either hope CR distribution improves or hope that some YFI vault users are sophisticated enough to unwind before a large drop occurs. In such case lower than here calculated risk premium and SF could be justified, even at higher DC.

16 Likes

What’s the worst that can happen? We print another $20-$30M worth of MKR?

On a serious note–As a YFI token holder I do know this thing can whiplash to under $10K and back up quick fast like my name is Flash. So, not sure where we draw the line. It’s against my better judgement–but I polled for $30M…

1 Like

Not to beat a dead horse but can’t we sort of force their hand on this front by raising the LR? Does that help us maintain our preferable position at least on the interest front at least when compared to Aave?

2 Likes

Raising LR might perhaps not work well because you would probably want to move it towards 220% or something and you could liquidate some vaults, including 10m one. But another vault type YFI-B is possible, I actually wanted to address this but I forgot :slight_smile: However Aave already has much lower LR of 154% and I am not sure who would want to borrow at 220% if rate is not really really low. But I may be also wrong.

1 Like

Increasing rates is preferable to LR increases from a UX perspective.

2 Likes

YFI cary the risk on contract, like pickle or hervest.
They already were closed to lose some found 2 or 3 times.

If they get hacked, YFI can lose easily 50% straight away. Can we deal with that?

Raising CR seems needed to me.

Yeah i guess I’m mostly wondering what our numbers look like at the 200% LR. Don’t think i would be in favor of purposefully liquidating people. FWIW I don’t think that ~200% seems that crazy to me. Nexo.io requires about 2x collateral for bitcoin backed loans. Which is to say there is probably some demand at that number otherwise they would be setting their collateral requirements differently.

This is starting to sound like a better idea to me all the time. Might be something to keep an eye on depending on how this vote goes and how quickly / if people take up the leverage opportunity.

FWIW i think this is a matter of opinion. Personally when i borrow the most important question i want to answer is what is the present value of the loan payments over its lifetime. Variable interest rates make figuring that out next to impossible bc I have no idea what the people on this forum are going to decide to do.

However i would admit that to be the case if we start to liquidate people.

1 Like

I posted about this on Twitter (lol) but one thing to consider is that one of the factors for the debt ceiling and stability fee of the YFI-A vault comes down to how the vault is being utilized. Do we have users using a very high collateralization or taking on close to the maximum leverage they can? This combined with the volatility of the YFI token should be a factor for Maker Governance to consider when deciding whether to raise the debt ceiling (increase our exposure to YFI). There may be a case where we want to gain exposure to a certain user, but not another. Traditionally we have done this user segmentation by having two Vaults (ilks) with different risk parameters as is the case for ETH-A/ETH-B, and USDC-A/USDC-B.

Let’s introduce Yearns delegated YFI vault as a new user primitive. Yearn delegated vaults deposit tokens into the Maker Protocol in order to borrow Dai which has much more favorable yield generating opportunities than the underlying asset. Yearn manages the collateralization of these positions through keepers which monitor the Oracle Security Module price and rebalance if the price of the underlying token drops significantly. As a user, this makes Yearn delegated vaults much less risky than “degens” even at the exact same collateralization ratio.

In light of this I want to put out the idea of having vaults specialized for individual “users”. In effect a “whitelisted” vault. Of course the Risk Team would need to chime in whether this is something that actually helps manage risk. Additionally, currently the Maker Protocol does not support a whitelisted vault, so the Smart Contracts Domain Team would need to chime in on how feasible something like this even is. I suspect we could do it through adding a whitelist to a collateral adapter.

14 Likes

I think this is an absolutely fantastic idea.

I’m an active contributor in the Yearn community, and am currently working on some ideas for upcoming YFI yVault strategies that combine aspects of current governance staking with yield farming (as well as potentially supporting protocol initiatives) all in one go. I think Maker is a perfect place to store YFI safely while still being able to utilize its value. Realistically, I think a YFI yVault that deposits YFI into Maker could end up replacing governance staking, and likely will if we offer the option.

With all of that said, I don’t know enough about Maker governance (or how the risk calculations are performed) to know whether having a special YFI vault for Yearn is an outrageous request. Paired with Yearn’s significant YFI holdings in governance staking and the YFI yVault, whitelisted access to the OSM, a health factor maintained by keepers, and the relatively conservative risk parameters that our DAI strategies tend to have (farming CRV or COMP, nothing too crazy), I think it could be very beneficial to both protocols. Maker would be earning fees on a large amount of capital, and Yearn would be able to utilize that capital more effectively and efficiently.

5 Likes

Not sure if this has flaws or is overly complex, but here’s an idea for creating a vault for use only by a YFI Vault, with no changes to MakerDAO smart contracts.

The YFI vault could create an internal accounting token representing the YFI it holds, and that token could be on-boarded as Maker collateral.

This YFI vault would need to provide a function to redeem that token for YFI, but it would only be useful in the case of Maker vault liquidations.

I guess there is a loophole, though, which is that in the case of liquidations those tokens could end up owned by someone other than the YFI vault who could decide to deposit them back in Maker instead of redeeming for YFI.

1 Like

Thank you for this really interesting analysis :heavy_plus_sign: :computer: :muscle:. May I ask what is the reason behind choosing -40% for both daily price drop and auction slippage?

2 Likes

That is not a bad idea at all. Some might say we shouldn’t make differences between users, but in distant future I do imagine loans would have parameters set per type of user risk. Currently Maker charges same rate for any user of YFI-A, regardless of your CR or ability to deleverage. Which means higher collateralized vaults are actually indirectly subsidizing fees for lower CR vaults being more risky.

Whitelisted vaults could also come handy for institutional user types who don’t mind paying a bit higher fee if they knew there is always debt ceiling available and rates are more stable. In return Maker gets long term and direct relationship, which helps us manage risks better because we can always reach out to him. If such user can prove they have systems in place to unwind before next OSM price and their balance sheet is strong to add collateral/make repayment, we don’t mind setting more favourable risk parameters. For retail users we could do the same thing, for instance we could make more favourable LR or SF for vault users protected by DeFi Saver.

We plan to publish analysis for each vault type next week, but in genera price decrease is based on largest daily drawdowns observed in past trading data. For YFI this might not be totally appropriate since the asset has a short trading history. But since we don’t adjust for the fact that some vaults might avoid liquidation before next OSM price, we thought 40% drop is conservative enough. High auction slippage of 40% is based on us modelling for max DC which can be up to 60% of daily trading volume. For lower debt exposure auction slippage gets toward 0, which then helps us create risk premium to debt exposure curve. Estimating exact auction slippage is really hard, but the easiest shortcut to it is to compare liquidated debt amount with Uniswap LP pool liquidity and estimate slippage.

5 Likes

If Yearn (or any other superuser) is using different tools to monitor and automatically unwind positions, I feel much more comfortable (since there is less risk).

We might want to let anyone become a superuser (they need to prove that they have mitigated certain part of the risk), but I do not think that we should treat everyone the same. That is penalizing the more conservative (risk-wise) users.

3 Likes

The 20M YFI-A Debt Ceiling has maxed out.

1 Like

This largest vault is a fund. Not a normal user. Is it worthwhile to create a separate vault category for funds. We feel that a fund can easily consume more than half of the 20M allocation and its not fair on normal consumers.

1 Like

Funds and enterprise solutions can definitely propose new vault types and whitelists for themselves if they wish. We have been discussing these more bespoke options recently and I think it would be nice to see new relationships start to develop.

2 Likes

We should raise the YFI debt ceiling and stability fee at the same time, please go ahead.

1 Like

Quick update: @Primoz has started another signal also reflecting options w.r.t. SF and LR - Thanks for you votes here, it helped on the options for the new signal. Please express your opinions regard SF/LR-adjustments over there.

Nevertheless, I will let this signal running until 2020-12-03 as initially planned.

2 Likes

I would also like to see YFI whitelisted vaults given that v2 is going to be more robust and less likely to be liquidated than the retail consumer.