Episode 78: March 17th, 2020
8:26: Vamsi’s presentation on Debt Auctions(flop auctions)
23:10: Dockerized auction keeper presentation overview by Marc Andre
32:00: Potential executive vote to lower minimum bid on Debt Auctions
SCD shutdown discussion, tabled for Wednesday
1:05:54 Circuit breaker discussion and USDC overview/recap
1:26:33: SCD migration
SCD: The Single-Collateral Dai system
MCD: The Multi-Collateral Dai system
CR: Collateralization Ratio
DC: Debt Ceiling
ES: Emergency Shutdown
EV: Executive Vote
GP: Governance Poll
SF: Stability Fee
DSR: Dai Savings Rate
Summary & Introduction
Welcome to the Daily Governance and Risk call. We’ll be having these daily to discuss the rapid series of measures to combat market instability and risks.
We all know that one year in crypto feels like ten years in the traditional world. Real-world is moving at a fast pace, so we have to move even faster. With open governance, we need consensus, decision making, and signaling. But we also have to be agile to get things done.
This is an experiment in how governance happens when markets are frothy. Though this is a challenging series of lessons, these calls will serve as a learning experience for posterity. We’ve already made decisions in the past week to strengthen the protocol, but, as we’ve seen, nothing is permanent. We have the coming weeks to refine and review everything that happened over the past week to get things right.
Market frothiness has raised a few other issues, also agendas for post-mortems. The quick moves in the market recently accelerated several plans we’ve had from the beginning. USDC was one of them, alongside discussing conditions for SCD shutdown. As the ecosystem becomes murkier, it’s essential to understand these dynamics.
DISCUSSION HAPPENS IN THE FORUM
Debt Auctions(flop auctions) Presentation
- We have both debt and surplus. (Which is why that calculation shows as a negative total on Dai stats.)
- Slight typo, lot size starts at 200. Later on, Cyrus talks about potentially adjusting this to a number that makes more sense.
Questions and Comments
chat Kurt: Last time I calculated, the flop start time was approximately 14:30 UTC on March 19th (10:30 AM EDT)
Dockerized auction keeper presentation overview
Link to the Dockerized Auction Keeper
Dockerized auction keeper is up for easier out-of-the-box use.
Keepers require optimization to improve auctions; the dockerized model is a solid start but requires individual tweaking for further optimization.
- New sharding functionalities allow for multiple accounts to work around the limit of 50 transactions per specific address.
Questions and Comments
- Kentonpresscott: As a comment on the last slide, in terms of keeper bid-only settings to limit node and server constraint, the latest GitHub commit now has bid-only as the default option.
Governance and Risk
Potential executive vote to lower minimum bid on Debt Auctions
Flop auctions are beginning at approximately 10:30 AM Eastern this Thursday. If the system accrues further liquidation penalties, this time could push back. Due to yesterday’s surge in liquidation penalties, it was pushed back from 6:30 AM to 10:30 AM. On the accounting side of the system, some of the bad debt clears out and the auction pushes back.
One of the critical questions that have been discussed is the starting price. The current structure is 250 lot of MKR for 50,000 Dai. While illiquid, the implied USD price is $200 per MKR.
That 200 number was chosen back in November during the MCD launch. Due to recent development, a massive crash in collateral prices could lead to a crisis of confidence in the amount of Dai. As a result, flop auctions were created to get bad debt out of the queue as soon as possible, 48 hrs after bad debt accrues. A 2/3rd discount was built into the system at $600 for the post-crisis MKR price was set to about $200.
We did not see that crisis of confidence in Dai. The Dai price has remained stable, reflecting the community’s belief that Maker’s long-term prospects are fine. There is less urgency to clear out the bad debt, so long as it gets cleared out eventually; which is partially why the extension from 48 hours to 6.5 days was suggested and passed.
- Now, there is a case where if the MKR spot price is below the 200 Dai per MKR starting bid, then we may see no bids. In this case, the auction would be put back 72 hours(three days.) In this case, the price would drop from 200 to 166 Dai per MKR. The question is: should we propose a governance vote to change the starting price to avoid potentially pushing the auction back? Or do we want to keep it as is? There is an argument to be made that even if the spot price is below 200, due to illiquidity, the auction might stil clear just fine. An executive vote can be used to either increase the lot size or decrease the bidding amount such that the starting price is reduced a bit. We would need to do this before the auctions start this coming thursday(with a 4 hour governance delay.)
Questions and Comments
- Charlie Noyes (From Paradigm): We agree with Cyrus’s point regarding not having to wait three days for the auction reconfiguration. I can see anxiety in the community regarding the debt, not clearing. I think we’d be supportive of adding a margin of safety to minimum bid price or reducing the time for the spot price in going down to three days. Those aren’t super strong feelings, but it makes sense optically. It is best for this debt to get out as clearly and cleanly as possible.
Emilainobonassi: A price reduction is an opportunity for speculators. If there is a reduction to, say $100, it could lead to people shorting MKR. If lowering the price could affect the system by introducing speculation, it may be better to reduce time delays instead of providing for more agile responses. This, to me, is a better signal than lowering a price.
- Cyrus: There is no good rationale for lowering the price arbitrarily. We could drop to a dollar, but that could open us up to an attack. Some people may look at that approach as a free-market, but we should understand the risks involved in this decision.
0age(person talking for the backstop syndicate): From my perspective, regarding the backstop-syndicate, the risk is one sophisticated keeper mounting an account to steal collateral and execute malicious upgrades. A community of users that rely on Dai is stronger when there is less opportunity for attack. Regardless of where the auction price starts, a 100 Dai for 1 MKR is a good shelling point with broad consensus. Some people are trying to buy into Maker who are willing to pay, then some opportunistic people are trying to scalp Maker, and finally, there is the third group of people who just want to help.
Auctions that begin higher and then trend down is preferable. This gives those in the community without the capital and sophistication a chance of participation and contribution to the backstop syndicate. The real goal, in our case, is not to buy Maker but to make sure that Maker is sold. If more people come in at a higher price, it’s a win-win. For us, if the price starts at $1, the idea is still to bid at $100.
There will be a hundred auctions going on at once. Our intention is signaling support for the Dai ecosystem and helping prevent a considerable chunk of Maker from selling at low prices and centralizing into a few hands. We’ll go in whatever direction is decided. But I don’t think it makes sense to have no one bidding if the spot price is not accurate. The thing that worries me about the time that the auctions are running is: if the
ttl is being pushed out, then the auction compresses into a smaller window.
Cyrus: Another idea is the step size in between failed auctions. It is currently at 20%, so if an auction fails at $200, it would fall to $160. We can set the step to 50%, aggressively lowering the bid price between auctions.
Cmooney: I think 20% is a decent parameter because it stops slippage. The three-day delay may not be great.
- Nik Kunkel: Changing the auction model as a whole right now, as some are suggesting in chat, is not reasonable. Smart contracts are not something you can whip up overnight; we are constrained by the current auction model. We should be focusing on when the auctions start, starting price, and duration. Let’s focus on what we can change now.
USDC and Circuit breaker discussion
Super brief recap. On Thursday, we had two main issues: network congestion and Dai liquidity. We haven’t had a chance to properly address these issues and how they interrelate with each other. Increasing
ttl(auction duration) to six hours was primarily to hedge against network infrastructure issues. Eth congestion led to bidding issues etc. Adding the delay allowed keepers to deal with that.
What this didn’t give us was a proper solution to the Dai liquidity issue.
Questions and Discussion
chat Matthew Light: It might be worth telling everyone on the call that people can immediately generate USDC from USD on Coinbase with no cost. The addition of USDC is essentially equal to adding US dollars as collateral
chat is there any evidence of active keepers with large USDC balances?
- (see the following section)
chat Matteo Leibowitz: Are keepers correctly incentivized to alert MKR holders to trigger the circuit breaker? It seems like in most instances, they’d be better off just letting auctions proceed and bidding low.
chat cmooney: we can know if keepers are tapped out too. If we start seeing liquidations below market that are accumulating to system debt.
- Last week we had one of the most active weeks this year in migration. Sai supply dropped by $4.5 Million, and the migration contract was $2.5 Million.
- Since SCD CR was much higher than on MCD, we didn’t see nearly as many liquidations.
- CDP repayment activity: 170 CDPs repaid or got liquidated in the last week, about $4.7 Million in debt repaid.
- The day after Ether crashed, many CDPs migrated.
1st hypothesis is: in the last few weeks, CDPs could deposit additional Ether to fight liquidation. Now they capitulated, extra exposure wasn’t worth it. The only way to close the position was to migrate.
2nd hypothesis: SCD is a bit different than MCD; it could have been fear of more significant losses in the system, which would dilute PETH. That would lead to compounding losses.
There is less flexibility in SCD. You can deposit more ETH or migrate, so it seemed that my guess from the past few months (that falling Ether would force their hand) turned out correct. Also, a reminder that penalty fees go to PETH holders. $200K in interest paid to the system as a result of these migrations.
- This is a chart of Sai flowing into the migration contract. Spike on Thursday, likely because there is a premium on Dai, which you can get by migrating. The other option is depositing Dai in secondary markets. There were bite’s today, but no one was bidding. This morning I saw many of them clearing out. There are still a few CDPs with liquidation ratios below 150%, when the peg returns that should alleviate itself.
- Cyrus showed how much debt liquidates in MCD. In SCD, the situation might be better, but if ETH price goes to $60, most of the debt gets wiped. $80 or even $70 is fine since there is liquidity in the migration contract. If the price drops that low, then we have a Sai liquidity crunch. We mentioned SCD shutdown, however these should be some polling about that.
- I checked how much Sai supply is inactive, holders that don’t move have almost $10 million Sai. Hard to say when the incentive to migration will arrive.
Cyrus: Also, just to point out, the keeper liquidity problem is even worse in SCD than in MCD. In MCD, we have the USDC collateral type and the liquidation freeze. If the price drops, the CDPs go underwater, but keepers can’t come up with the liquidity.
- Primoz: We have the buffer in the migration contract.
David: Also wanted to mention this, it’s a little out of order, but
cat doesn’t have authority on the USDC flippers, so USDC liquidations aren’t available without another executive vote.
- Cyrus: Great point! In the forum post, we mentioned that the collateral type initialized with frozen liquidations. Even SF accruals will not trigger USDC vault liquidations.
- Cyrus: As a brief early analysis of Dai backed by USDC, the primary value of backing by USDC is that it allows keepers to maintain liquidity. You can see here that much of the USDC minting was used for peg arbitrage.
Links from the Chat
Tim Black produced this summary.
David Utrobin produced this summary.
Igor Teslya produced this summary.
Everyone who spoke and presented on the call (listed in the headers.)strong text