Proposal for Immediate Changes and Executive Vote

Hello,

This is an unprecedented time for MakerDAO. Thursday was a tremendous learning opportunity for the entire Maker community, and MakerDAO can immediately apply some of the lessons learned towards making the Protocol stronger than it was before.

Situation

Dai is facing a severe liquidity crunch. The Dai peg has been trading as high as $1.10. Dai generation from Vaults is low. Many in the community are buying Dai in order to participate in this upcoming Thursday’s MKR auction. Some are accumulating in order to pay down Dai debt in preparation for another downturn in the Eth price. There is an extreme ecosystem-wide shortage in the Dai supply.

As a result, the Interim Risk team is putting forth a proposal for an immediate Executive Vote in order to help alleviate these issues. We hope that this proposal both resolves the credit crunch and is a meaningful step towards restoring Vault use. The executive vote will include the following changes:

  • Monetary Policy

The drop in DSR and Stability Fee to 4% on Thursday has not had immediate results for managing the Dai peg. Clearly, Vault use remains low, and there is a surging demand for Dai. Nonetheless, MakerDAO should consider all options with respect to monetary policy. The Interim Risk Team is proposing a DSR of 0% and a Stability Fee of 0.5%. This change will result in a tiny DSR spread which may mitigate any potential incentive issues that might arise from a stability fee of 0%.

Additionally, the proposal also includes an SCD debt ceiling reduction to 20 million.

  • Governance Security Module

The 24 hour GSM has constrained the MKR holders’ ability to react swiftly when it has been needed most. Moving forward, given the continuing significant volatility in the markets, the community should consider having the ability to react more quickly. The Interim Risk Team is proposing that the community reduce the GSM to 4 hours. This should give the community more time to deal with any governance attacks or handle network issues, but also significantly speeds up turnaround time for any new executive votes. MKR Holders can, of course, revert back to the 24 hour GSM, or another time period, in the future.

  • Collateral Liquidation Freeze

The proposal includes a significant addition to MakerDAO’s issue-response tools. The engineering team at the Maker Foundation has created what — if introduced by the MKR Holders — will effectively be a decentralized circuit breaker, to be utilized only by the authority of MKR Holders.

To very briefly recap Thursday’s events, the $0 bidding was apparently affected by two marketwide issues: network latency issues and lack of Dai liquidity. While Thursday’s 6-hour auction bid duration extension, voted in by MKR Holders, should be effective in mitigating network issues, there are still some situations where insufficient keeper liquidity might lead to affected bidding. For instance, a sudden Eth-price drop(anything resembling Thursday) could trigger another series of parallel auctions. It is currently excessively difficult to source Dai to bid on auctions, as well as pay for gas, in times of such network congestion.

The proposed solution involves a form of rate-limiting in order to reduce auction triggering. Keepers seemingly have a sufficient amount of assets, and the exchanges can support the necessary volumes. If auctions were to be more sequential, as opposed to parallel, then keepers would likely have the ability to recycle their Dai from auctions and rebid again. The risk of excessive price volatility is outweighed by the alternative risk that collateral may be sold for an unreasonably low price.

As a result of this proposal, the MakerDAO governance community, by full executive vote, will have the ability to temporarily disable Vaults from being sent to the auction liquidation module. To re-enable liquidations, MKR Holders would have to pass a subsequent executive vote. Critically, this proposed module lives outside of the GSM and is not subject to any delay.

To be clear, today’s executive vote on this proposal will not trigger the circuit breaker. It merely enables MKR Holders to utilize it at a future date. Also, as a reminder, this vote is still subject to the 24 hour GSM delay.

A sample scenario in which the community might want to make use of the circuit breaker: Assume the Eth price spikes down to $90, liquidations occur and keepers bid on auctions, and then immediately after the price dives to $70. Given the OSM delay, the MakerDAO governance community, in conjunction with keepers, would have one hour to react to keepers requiring more time to recycle Dai from the first round of auctions in preparation to handle the incoming collateral queued up for liquidation. That reaction time includes the time necessary to make the decision to freeze auctions temporarily.

This situation would be an ideal time to use the circuit breaker. After allowing enough time for keepers to recycle their capital, MKR Holders could then unfreeze the circuit breaker and trigger the next round of auctions. MKR Holders would have one hour (before the OSM price applies) to trigger the liquidation freeze. While this is difficult and indicative that a long-term solution is required, the MakerDAO governance community needs to be cognizant of the fact it would have to rally an executive vote in under an hour to prevent severe underbidding.

Summary

The liquidation freeze/circuit breaker would be another tool in the governance community’s toolbelt. It is relatively simple to implement at the technical level, and decentralized at the social and governance layer. Without it, MakerDAO remains more vulnerable to certain price movements. This is not an issue of how much collateral the exchanges can absorb, but how much Dai liquidity keepers can generate and utilize. The circuit breaker would help by processing liquidations in a more orderly manner.

There are certain cases where an extreme price drop might necessitate additional aggressive actions by the community (such as intervening with the oracles or triggering Emergency Shutdown, which is better addressed in separate threads). The community should continue to explore additional potential actions, such as the addition of new collateral types, but the Interim Risk Team proposes the above changes for immediate discussion and consideration.

Thanks all, looking forward to hearing your feedback.

13 Likes

Let me add some new info @cyrus at the time may be unaware of. We got new minting of about 3-4M DAI in the past 2 hours that has entered the market (someone was saying to buy ETH) the PEG is now 1.027

Given the extraordinary circumstances and possible additional market downturn risk I think it is reasonable to do SF=0.5, DSR=0, SCD DC 25-20M and GSM 24hrs to 4hrs and the addition of the auction circuit breaker (god great job to developers on getting that together).

Great Job cyrus.

As pointed out by psybull in chat perhaps we want to additionally lower the MCD DC as well from 100-90M just to head off any kind of OSM attack minting issue. We are still around 75M here so 15M overhead seems reasonable.

I do have a concern the PEG will overshoot to the downside here so this may be an additional reason to clamp down on additional minting but we are going to have to play this day by day.

Two questions:

  1. The GSM of 4hrs is a reasonable change and I see above this delay will apply to the auction hold.
  2. And how does this play with the ttl of 6hrs there basically giving us the time we need to stop auctions ‘in process’ or will those close?
3 Likes

The circuit breaker I like in concept, but has this code been audited?

Another question, should we be considering a debt ceiling reduction for DAI-ETH? considering we currently have a ~25m DAI availability on ETH loans currently, is there any risk if someone were to mint all 25M at once? I was thinking maybe a DAI reduction 10M (to 90M) would be prudent.

4 Likes

Monetary Policy

Monetary Policy changes make sense, although I’m slightly worried about an over-correction that might require some future management (not that this is a reason to not do something). I second @MakerMan’s concern with the MCD Eth debt ceiling.

Governance Security Module

GSM change to 4 hours makes me nervous. Realistically, I don’t think that is necessarily enough time to rally MKR in case of a malicious executive. I recognise we have other concerns right now, but our situation gets a whole lot worse if someone tries to take advantage of this (especially given that more MKR is being printed in the near future.) Why 4 hours over 6, or 12? Does the Foundation believe that MKR for ES can be gathered in 4 hours?

Collateral Liquidation Freeze

Sounds good in principle as a short-term solution. Like @psybull I am worried about the potential for bugs in the code given the short notice and general atmosphere of panic. I’m assuming this will not be audited, but at the very least can we have an estimation of the worst case in terms of bugs in this module? How much authority does it have, what can it affect, etc.

As another point, 1 hour to rally the MKR to activate the breaker also sounds pretty difficult to pull off, unless everyone is already paying attention. I guess we won’t know how realistic this is until we try to activate it.

3 Likes

To me bundling the circuit breaker with the rest of this vote seems a little risky.

The GSM and Monetary values are more well understood and I would guess an executive vote with the backing of the risk team would would immediately pass.

Introducing new protocol functionality into that vote might scare off more risk adverse mkr holders and because of the rushed nature of the new code and the unknown runoff effects it will introduce to the protocol.

6 Likes

They can trigger ES themselves with the multisig. I think the reasoning is that governance right now needs to be able to act swiftly since the situation is constantly evolving.

Agreed.

1 Like

I’m glad for the community here as it is so easy to say yes to things.

On reading the added comments.

I think the auction hold while a good idea is probably risky to add at to an executive at this point.

I agree with LongForWisdom some justification for why 4hrs was good vs 6-12 I think is important.

We really want an easy executive to pass and the auction hold may have some contention and should be removed and offered for discussion in its own thread. I think it is important but needs consideration.

I also really want to understand what we think is really a solid GSM delay here. I have a 4hr number being proposed but no understanding why 1,2,6,8,10,12 whatever were not.

I am pretty sure we need to reduce the GSM but what are the ramifications of the quick fix vs. governance ability to act balance here? Generally I am going to want to trust governance and risk but I like that others constantly question not just for the action - but the actual reasoning behind such a choice of action being presented.

Thank you all again for hard work and hard questions - it is all of us that keep this ship from hitting the iceburgs floating around us.

It will be interesting to see if we can even get consensus on this in 4hrs much less an executive up and a vote passed in 12. :slight_smile:

Ehhh, they could. It stretches the unwritten contract that they won’t use it for voting, but I guess in that situation it would make sense.

I highly doubt they would hesitate if they thought the situation called for it. They could also rally VC mkr in four hours probably.

Same goes for the 1 hour window to enact auction hold. I am guessing the foundation has people looking at the situation close to 24/7. They could force that exec through quickly enough.

Has the situation not improved regarding keeper liquidity/diversity.

Side question, when people say “keeper liquidity” is that referring to the dynamics of the capital keepers access to participate in auctions or does that refer to the diversity of keepers actually participating? Both?

DSR should be greater than 0%. Even if it’s just 0.25%

Apps built utilizing DSR (Compound, Dharma, PoolTogether) are predicated on some form of a return. Sending it to 0% in the eyes of a normal user could throw the whole system into doubt.

1 Like

The vote is up now: https://vote.makerdao.com/

I completely agree with this. This will kill any new user coming into De-Fi space now.

It’s not a great situation but the DSR can change weekly (or more quickly in emergencies) presumably this is a short term solution.

It just seems leaving a tiny DSR is a massive perception difference and a very small economic incentive difference and therefore very worth it.

At this point part of the intention is to deliver a ‘get out of the DSR’ perception to the Dai that is currently in there. I would argue that perception difference is what we want to get more Dai liquidity.

1 Like

Note that this proposal has now passed as written.

SAI had no dsr yet still an attractive lending rate. Based on that, I doubt the secondary market will go to 0 for lending dai even with 0 DSR and .5 SF. Curious how they will respond.

1 Like

Some further context for some of the feedback so far

  • The monetary policy overcorrecting is the least of the problems right now. We have no way of knowing how much demand there is between $1.001 and $1.03
  • The liquidity situation has improved, but not nearly enough to protect us from a repeat of Thursday. Some sort of rate limiting is necessary until a better long term solution is found w.r.t auction process.
  • The liquidation freeze is an optional tool. If it’s deemed to not be so useful, we won’t use it. It’s good to have in the arsenal.
  • The new GSM of 4 hours means governance can be more nimble and add the DSR back in very quickly. It’s hard to overstate the important of liquidity over the following week.
  • GSM of 4 hours is enough time. the entire maker ecosystem is basically on 1 hour alert right now. It should go back up after things cool down.
  • The code is good, but the incentives need to be well understood.
2 Likes

Reminder that Compound, etc. had positive interest rates before a DSR even existed. Interest on lenders is also related (perhaps more strongly) to the Stability Fee.

1 Like