[Signal Request] Adjust ETH-A Debt Ceiling (2021-02)

Overview

Since the last urgency exec raising the DC of ETH-A from 590 MM to 740 MM around 50% of the additional headroom has been consumed in the last 3 days, leaving us with a utilization of 90%. With the current growth rate, we are going to hit the DC around wednesday.

I would like to poll for two things here:

  • changing the DC to allow more DAI from ETH-A.
  • find out if we want to switch over to IAM for ETH-A

Changing the Debt Ceiling

Pros:

  • more fees collected
  • more DAI in circulation

Cons:

  • I personally do not see any, as ETH is still the most liquid and robust asset-type we have in the portfolio, but I am happy to adjust here if people raise good points in the discussion
  • even more (Rism team should decide)
  • 1000 MM (+ 260 MM)
  • 900 MM (+160 MM)
  • 800 MM (+60 MM)
  • 740 MM (no change)
  • Abstain

0 voters

Next Steps

The Poll will run until January 22th; its outcome will either result in a on-chain-poll assuming the outcome of the poll deems it necessary or its intermediate results are going to be taken to another initiative - another signal poll lead by the risk team, onchain-poll, urgency executive.

IAM for ETH-A

update

@LongForWisdom started a more formal with concrete next steps over here: [Signal Request] Rolling out the Debt Ceiling Instant Access Module (DC-IAM)

Please express your opinion over there!

IAM is already active for ETH-B and has been working quite good. I guess we don’t want to have executive votes for changing the ETH-A Debt Ceiling on a weekly basis, so maybe it is time to think about activating. I know that the idea isn’t new, but maybe it helps to find out if we as a community want to have it so the corresponding mandated actors feel empowered to move forward.

Pros:

  • fewer signals, on-chain-polls, execs working with the Debt Ceiling of ETH-A

Cons:

  • ETH-A is our biggest collateral-type, maybe it would be better to start with something smaller
  • Yes
  • No
  • Abstain

0 voters

Next Steps

Since IAM for ETH-A is planned anyway, I don’t think we need a corresponding on-chain-poll about it - but happy to move forward with this as well if needed. Nevertheless, this poll will also close on January 22th.

4 Likes

As the price of ETH rises and DC risk exposure increases, we should slightly increase SF.

2 Likes

yepp, and i guess this is already covreed in the current Rates Changes Proposal

2 Likes

Looking at ETH-A liquidation curve, it tells me that about 100m DAI gets liquidated if we see around 30% ETH drop. Of course some will react and unwind before liquidated, but some won’t. Plus don’t forget there are other larger risky vaults (25m liquidated at ETH-B, 30m at WBTC-A at same price drop). And finally, in terms of protection we have 6h box of 15m and 4m of surplus (even if we increase surplus to 10m, it will take some time to fill it).

So if somebody asks me whether we should be increasing risk exposure and me knowing that tomorrow 100m+ debt could easily get liquidated, I don’t need to say anything more.

However CR ratios are healthier compared to average historically observed distribution and we aren’t seeing aggressive risk taking on ETH-A vaults as is the case at ETH-B. Even though the exposure is rising, the pace of ETH price increase is much higher. Still, if additional DC leads to existing vaults minting more DAI and lowering their existing CRs, this would be a very bad and risky situation - we could see over 200m liquidations on ETH-A at 30%+ drop alone! Even if we are optimistic and think 50% of those vaults will react to price drops and unwind in time, we could have quite a large liquidation event in front of us. Auctions of such magnitude could be clearing for about a day or more - depends how many people unwind in between, but otherwise the time we risk additional drop is total liquidation size/15m*6hr if I am not mistaken. And if price goes lower for another 10-33% during that time (33% to reach 100% CR, but there will be auction slippage), we start having loss events. So it is about what people think, are we going to see short term crashes of ETH below $700 again or not and if vaults are going to become more aggressive on leverage?

Also by approaching the 1bn debt ceiling and potentially CR distribution becoming riskier, we might need to become much more aggressive on SF and Surplus, although this won’t save us from anything short term.

If somebody is feeling a bit paranoid about the whole situation as I am, would it be too unpopular from me to soon propose a higher LR ETH-C vault instead of increasing DC for ETH-A all the way up to 1bn? At least until we get more protection in Surplus Buffer or until this crazy growth is behind us and things stabilize.

The current liquidity metrics for ETH are justifying a higher DC though, but I don’t believe this community is also prepared to increase SF, box parameter and surplus buffer at the same time substantially and therefore I am looking for other risk mitigation tools such as introducing higher LR ETH vaults. I expect people here arguing higher LR ETH vaults won’t be competitive and usable enough, but note that the average debt weighted Collateralization Ratio on ETH-A is 473% right now.

Such measures would be similar to what margin trading platforms do when prices are becoming very volatile, which is to suddenly increase margin requirements for users. But in Maker’s case that isn’t doable instantly so we may need to prepare similar measures in advance.

10 Likes

Looking at the SC-team feedback for WBTC-B, this is doable but not as part of a urgency action.

I like that you are paranoid/conservative, but not raising the DC if there is obviously need for DAI cannot be an option

If ETH-C with similar SF/LR as ETH-A does help from the risk perspective, I am all in for that. But this is maybe a step after the next ETH-A increase. We need to avoid the situation of not being able to mint DAI-from-ETH

1 Like

I think we should increase the total ETH DC by allowing for new vault types, with higher LR. New vault types can be as simple as higher LR, lower SF, or even use something like Global Collateral Ratio (new vaults can only be open at current GCR or higher in that vault type). So lets assume we have ETH-C with 200 CR, and lower SF than current ETH-A - some large and highly collateralized vaults will probably migrate to ETH-C, effectively lowering DC utilization in ETH-A, allowing for new space.

This would also partially solve the free riding problem, as users with higher CR are much less risky for the system, while they are charged the same SF.

2 Likes

I think we should have a higher ETH-A debt ceiling with a higher stability fee and commensurate stability buffer. I also believe that @yaronvel should design an ETH-C vault with B. Protocol to safeguard it from liquidations. This would be a win for the customers, Maker, and B. Protocol.

2 Likes

Just want to give the voice of the ecosystem that integrates with Maker.
Having ETH-A depleted, will affect integrated projects.
I our case it will not put user funds at risk, but will make the system less usable.
I think @niemerg from yield protocol will have similar issue.
All the yETH vaults.
Etc.

I get that this is only one of the considerations to have, but wanted to make sure is aware of it.

7 Likes

That is correct. Yield Protocol makes the assumption that it is always able to borrow via ETH-A. If this condition is not true, our user experience is significantly degraded because users may be unable to withdraw Dai they are owed.

6 Likes

Yes it seems like a no-brainer that the ability to create dai from ether should be available to anyone on the globe every day of the year 24/7. The parameters to make this a reality will be adjusted as needed but having ether not be available, having an “out for lunch” sign is not acceptable. For the spice must flow.

1 Like

To mitigate the additional risk of the higher debt ceilings, I think we should focus our attention on growing the surplus buffer aggressively rather than raising stability fee rates or collateralization ratios. Let’s seize the opportunity and step on the growth pedal!

4 Likes

Are you suggesting increasing the SF again above the rates proposed on Jan 6th? I agree with increasing the SB and the box parameter but am assuming you would want to see how the market reacts once new SFs are implemented.

Regarding the SB, do you feel the $10M currently leading the poll is high enough or should we increase this to $12M+? We’ve grown DAI by over $300M since the poll went live so I think it would be important to consider larger increases (as a side note, I feel like the signal request polls should be closed more quickly given the rapid growth we’re experiencing, or if sufficient consensus is reached).

Regarding the box parameter, are we able to get a sense from the Keepers of the level of capital they hold?

1 Like

love this idea … rewarding low-risk users with a lower SF seems like something that should exist anyway

2 Likes

It is my understanding that the main issue is liquidation. With the current box it would takes weeks to liquidate vaults which is likely to hurt us a lot (if I understand correctly that we can liquidate 15M DAI every 6 hours).

This is a broken system and should be fixed. We should either help Keepers to be up to the task of liquidating 100M DAI per hour or find another solution. Maybe Liq 2.0 is the solution already.

SB, SF and LR are not solutions to this problem. Sure a higher SB might let us live. But we will still take the loss because our liquidation system is not good enough.

We need to have a liquidation which we are confident that it can liquidate 5% of available ETH in 2 days. Then we should allow having 5% of all ETH as locked collaterals. If ETH is the pristine collateral of DeFi and Maker the biggest lending protocol, 5% is the minimum we can expect.

The speed of liquidation is a technical problem that is fixable. The ability of the market to absorb all those sell orders is a more complicated matter (just like portfolio insurance in 87’s Black Monday).

3 Likes

This is only half the problem and is fixed by Liq 2.0, IMO. The other half of the problem is using the OSM on a 1-hour delay for Dai generation. A suggestion I would make to the Oracles team (for the medium term, those guys are absolutely swamped) is to have Dai generation be governed by the lower of the delayed OSM and a live OSM, and have liquidations only be governed by the delayed OSM. This will require core smart contracts changes, so it’s not an easy lift.

3 Likes

Please note there is a new more formal signal request regarding IAM here: [Signal Request] Rolling out the Debt Ceiling Instant Access Module (DC-IAM)

Please participate over there as well!

This will move on-chain on Monday as it was wrapped up in @Risk’s more general debt ceiling poll here: [Signal Request] Debt Ceiling Adjustments, 11th Jan 2021

Thanks to all who voted.

Our proposal to experiment a Vault backed by committed keepers is in RFC phase, and you are welcome to comment about it.

Liq 2.0 abilities will be capped by DEX liquidity, which is far from 100M DAI an hour imo.
The solution we proposed are committed liquidators who will be incentives to build complex hedging systems, and thus could absorb some liquidity in extreme market conditions.

1 Like

the corresponding on-chain-poll about ETH-A DC increase running right now - please vote.

also - since we are about to hit the DC of ETH-A again pretty soon - there is an urgency executive for raising the DC to 1000 MM on-chain already. please help.

1 Like

So I’m a bit confused by the various hesitations around raising the ETH-A debt ceiling so that we do not need urgency executives every two weeks. The last time I asked about this, it seemed to me that the ultimate goal of Maker is to maintain the DAI peg, I supposed in such a way that users trust this asset, in turn increasing the demand for it and the cashflow generated.

I see some people in this thread pointing to the risk of liquidation as a an argument against raising the DC. Why is that? In what way do liquidations impact the peg? In what way do liquidations impact the security of the overall system? It seems to me that this risk is segregated on a per vault basis and only concerns the vault owners, not the Maker system as a whole.

In that context, and as I was saying in the thread linked above, I am under the impression that the DC should always be increased when the cap is about to be reached, and that there should in fact always be a comfortable margin, except if the value of DAI below $1, at which point there is a need to make people unwind their vaults by increasing the SF, thus allowing for a reduction of the DC. Indeed, why would we artificially limit the amount of debt generated by not moving the ceiling appropriately and raising the fees, as long as the peg is maintained?

1 Like