[Agenda/Discussion] Scientific Governance and Risk #168 - Thursday, November 18 17:00 UTC

Disclaimer

This communication is provided for information purposes only. The views and opinions expressed in this meeting are those of people involved, and do not reflect the official policy or position of MakerDAO or any of its contributors, Core Units, or affiliates. This communication makes no representations about the enduring accuracy of the information or its appropriateness for a given situation. This content is provided for informational purposes only, and should not be relied upon as legal, business, investment, financial or tax advice. You should consult your own advisers as to those matters. References to any digital assets and the use of finance-related terminology are for illustrative purposes only , and do not constitute any recommendation for any action or an offer to provide investment, financial or other advisory services.

Call Information

The zoom waiting room will be on, and a password is set to: 748478, please ping us in the Maker Discord’s #governance channel if you aren’t let in from the waiting room.

2021-11-18T17:00:00Z

Slides - updated before the call

Introduction

Governance Round-Up

Selected Updates / Discussions

Prepping for Alternative Market Conditions;
Various Parties: Risk, PPG, etc…
Prepping for alternative market conditions;
Collateral Bear Markets
PSM drainage
Soft peg underneath the PSM-caused hard peg
Monetary Policy levers for peg control
DSR
Emergency Playbook, for when things happen fast.

Other Discussions and General Q&A

  • (Time permitting, after general Q&A) - Ilk Hole size, Keeper health, liquidation ecosystem discussion.

Leave your questions in our anonymous question box and we’ll do our best to bring them up during the call.

2 Likes

G&R #168 Snippet

This snippet includes Governance, MIPs, forum updates, and Core Unit team discussions from the MakerDAO Governance and Risk Call #168.

General Updates

Votes

Executive:

  • Last Week’s executive - PASSED & EXECUTED
    • Parameter Changes to GUNI, WSTETH, D3M
    • 218,059.10 DAI Returned to the Surplus Buffer
  • Executive proposal is up for vote tomorrow. Will include:
    • WBTC-B Onboarding
    • WBTC Stability Fee Changes
    • Local Liquidation Limit (ilk.hole) Changes
    • Change of Covenants for P1-DROP (Peoples Company Series 1)
    • Offboarding Collaterals

Polls:

MIPs

Weekly MIPs Update #62

  • The Ratification Polls for November are now entering their second and final week.

Forum at a Glance

Forum at a Glance: November 11 - 18

Team-led Discussions

Bear Market Market Condition Prep

Are We Prepared? How Do We Prepare?

  • PSM gets filled during bear markets because people rush to repay loans.
    • But how high can we afford to go in USDC on centralized stablecoins?
    • People will liquidate volatile positions to less risky assets.
  • Bear market may drop yields during farming and reduce stablecoin saving/lending due to increased risk.
    • A stablecoin unwinding is a low-chance possibility.
  • Any immediate spike in demand will deflate over time as fear alleviates in the markets.
  • Have there been any simulations created for multiple collaterals over time in bear markets?
    • Primoz: Yes, you can simulate that yourself in the Risk Dashboard.
    • We began developing a framework called Asset Liability Management.
  • Some assets like renBTC and UniV2 are difficult to liquidate large vaults.
    • These assets are not ready for a giant liquidation wave.
    • Auction demo keeper did not perform well.
    • PE CU is trying to work with UniV3 to handle such a situation and add a good liquidator.
  • Things to do to be prepared: update liquidation keepers.
  • However, liquidating large Vaults would take way too long, which provides potential open market risk.
    • Liquidating smaller chunks allows borrowers to save themselves.

What Happens in the Event of Sudden Loss or Invalidation of Collateral?

  • Emergency Playbook by Davidutro:
    • During an emergency, the following occurs:
      • Communications coordination.
      • Technical coordination.
      • General coordination.
      • Etc coordination.
    • The Emergency Playbook documents and connects all coordination processes as clearly and successfully as possible.
  • It is important to communicate and get auction keepers to participate in auctions during such bad events.
  • Need to double-check Signal Group and check all the auction keepers’ info.
    • Make sure everything still works and who is active/inactive.
    • However, keepers are very private and competitive about their operations; it is difficult to get any contact information.
    • Profits highly incentivize Keepers.
  • Should new collateral types require a period of mandatory fire drills for their liquidation systems?
    • Test experiences help Maker improve and better prepare for actual events.
    • We need to immediately update keepers for the ability to perform easier liquidations
  • What happens if we lose network access?
    • Losing access to memory pool or nodes causes prices to continue to fall.
    • However, we have a cusp where after X% of a price fall forces auctions to be redone.
    • We should have a backup network like L2 where we can source liquidity on it.
      • Wormhole should help solve this.
  • In the future, most capital will move to L2s.
    • We will need everything on L1 to be integrated and adopted into the new L2 world.
    • Also, how long will it take for this to happen? Potentially, sooner than we think.
  • How can we educate our Vault users to use risk-mitigation tools?
    • Yes, this is super important, and Vault protection promotes more income because users apply higher risk.
  • Alternate unhappy perspective:
    • Market crash happens, and we proceed with buy/burn.
    • We get surplus from penalties to buy MKR, but then the market dips even more! We end up with more bad debt.
    • Will cause to reprint more MKR than was burned.
  • We should be agnostic towards what solutions to use instead of educating Vault users on all solutions.
    • Basic information is currently available in the community portal; more information is coming soon.
    • Reach out to Content Production for any ideas and feedback on this.
  • What if, during a market liquidation, users stick to central stablecoins and do not use Dai?
    • This can cause peg to break highs.
  • We do not need USDC or other stablecoins to do a PSM now! We can do it ourselves.
1 Like

Semi-transcription Summary

Episode 168: November 18th, 2021

Agenda

Video

https://youtu.be/2GcrJwElB0Y

Introduction

Agenda and Preamble

Payton Rose

00:00

  • Hello everybody. Welcome to the Scientific Governance & Risk Meeting #168 of MakerDAO. My name is Payton Rose, and I go by Prose11 online. I am one of the governance facilitators moderating our meeting to keep things on track. We appreciate all the smart, dedicated people joining us on the call today and those who will be watching the recording later. For those of you that are here, we encourage open participation. Ask your questions. If there is a lull in the conversation, feel free to hop on the mic. If you are unable or otherwise do not wish to speak your question, you can drop it in the sidebar. I will be happy to ask the question at an appropriate time. This meeting is recorded, be mindful and try not to speak over one another. Thank you all for joining us.

General Updates

Votes

Payton Rose

Executive:

  • Last Week’s executive - PASSED & EXECUTED
    • Parameter Changes to GUNI, WSTETH, D3M
    • 218,059.10 DAI Returned to the Surplus Buffer
  • Executive proposal is up for vote tomorrow. Will include:
    • WBTC-B Onboarding
    • WBTC Stability Fee Changes
    • Local Liquidation Limit (ilk.hole) Changes
    • Change of Covenants for P1-DROP (Peoples Company Series 1)
    • Offboarding Collaterals

Polls:

MIPs

Pablo

MIPs

04:23

Weekly MIPs Update #62

  • The Ratification Polls for November are now entering their second and final week.

Forum at a Glance

07:48

Forum at a Glance: November 11 - 18

Video: Forum at a Glance

Team-led Discussions

Bear Market Market Condition Prep

Are We Prepared? How Do We Prepare?

15:02

Prompt 1

Prompt 2

Prompt 3

15:20

  • David Utrobin: We are going to try a different format. This is a broad topic that has many subtopics within it. I prepared a handful of slides to prompt us in the discussion. There are three prompts. The first prompt is: Are we prepared for alternative market conditions? What does being prepared mean? The second prompt is: What happens in a collateral bear market? That is the main alternative market condition that people are thinking about. We have been in a decent bull market. Markets happen in cycles. When markets go up, rates and debt ceilings behave, and Dai demand and supply behave one way. When things go down, it behaves a different way.
  • It is better to start with prompt 2. What happens in an alternative market condition like a bear market? How does Dai supply behave? It is understood that when people’s collateral is falling, people rush to add more collateral or buy Dai off the markets to repay their positions. In a bear market, Dai demand becomes excessively strong. With the introduction of the PSM this year, it has been okay for the peg because we have a few billion of buffer to eat that extra Dai demand. Does anybody have other views that we might not be considered for what happens in a bear market? What happens after the PSMs get drained? What are our fallback levers?

17:37

  • Kirk: In a bear market, the PSMs do not get drained. They get filled because Dai is above Peg, and people will fire in USDC to mint cheap Dai and then sell it towards the peg. That is why we have got to 60% USDC in the first place. We already know what happens in a bear market. That is what we experienced after Black Thursday with Dai persistently above the peg, resulting in the introduction of the PSMs and taking on a bunch of centralized stablecoins. The biggest concern with that sort of situation is that if it happens again, it gets even worse? What if we go to 90% USDC? How high can we afford to go on centralized stablecoins before the market proposes that we use the decentralized stables instead?

18:34

  • Primoz Kordez: I agree with this. The PSM is a residual of Dai demand and Dai supply from borrowers. If there is a bear market, people will rush to repair it to repay their loans. On the supply side, Dai minted should shrink. On the demand side, you may be going to see more people liquidating their crypto positions’ wallets from volatile assets moving to stable assets. This will mean the PSM will grow. They have an alternative view of this. It is unclear what happens with stablecoin holdings in a bear market because most of the stablecoin demand was driven by yield farming. It is possible if you are in a bear market, these yields will drop. Now we are seeing them 10% higher, and it is possible they dropped to a few percent. In that case, it is questionable how many people will still be saving or lending their stablecoins at 2% or 3%. It might be too risky. Even though this yield is better than you would get in a bank, some people might step out of their stablecoin positions. This could affect Dai. A bunch of Dai is now on a curve. Imagine if, for some reason, there is the total unwinding of stablecoins. I do not believe this will be the case. It is one possibility. You could see Dai demand going down in a bear market.

20:34

  • Kirk: That is an extremely good point. It is a question of how big the different components are. Another component that increases Dai demand and bear markets is that if people think prices are going down, they do not want to hold something volatile like ETH. They might want to sell and wait for lower prices. After Black Thursday, when the Dai Peg was persistently high, do not forget that was before yield farming was even a thing. Compound did not start until the summer after Black Thursday. We had a high peg until then. There was enough demand from people wanting to be in stables because of the uncertainty of the market. They did not care about yields. They did not want to be in something that was going down.

21:21

  • Someone: We have a sample of this from 2018 during the bear market with PSY. We did not have as many tools for maintaining the peg upwards. People want to exit crypto in the early parts of the bear market. Things have gone up. Then they are going to pull their dollars out to a bank account or something like that. I see an initial spike in demand but, in the longer term, we could lose demand.

21:57

  • Chris Mooney: Our default view here should be May 2021 crash/correction. It may be a bit bigger, and some different things might happen. But by default, we should be thinking of that as the most likely scenario.

22:20

  • David Utrobin: Has Risk or anybody done any simulation? I know Risk does some sort of simulation stuff when you are doing risk assessments. Has there been a kind of simulation of a bear market across multiple collaterals at the same time?

22:40

  • Primoz Kordez: We are doing these simulations to calculate the risk premiums and the surplus buffer needed. You can go on our dashboards and by yourself pick parameters and see how bad it goes. You can simulate a 50% drop in a day versus a 40% drop. You can test whatever you believe is possible to happen. But this is related to risk. We are talking about bank management, and this is something else. We started developing a framework called ALM, Asset Liability Management. This is what banks do. It was not problematic for Maker before because all of our assets and liabilities are short-term. We can increase rates and decrease supply. We can match the demand, which is the Dai holders. When it comes to offering fixed rates products, such as institutional rewards or different retail fixed-rate implementations, this becomes important. We plan to develop a framework where a percentage of PSM would always serve as a stablecoin backing at Maker, considering the different scenarios of bear and bull markets. We plan to run some simulation analyses. Let us say there is a bear market. What happens on the demand side after considering inputs like the yields drop and the delta catch whales? You have Dai supply, which is borrowed, and then it is booting the AMMs. You have demand and supply matched, which needs to be subtracted. We will try to solve this. It will be difficult to simulate all different versions of next year’s bull/bear runs and then suggest a reasonable amount of stablecoin backing in Maker that does not lead us to bad back management or rely on increasing rates if we drain it.

25:29

  • Christopher Mooney: I ran the Auction-Demo-Keeper in a Hardhat forked environment to simulate a market crash like this. I fork it, and then across the board, I drop the price of our assets with Hardhat down by 10 or 20. I have even done 99% to see what would happen. The current Auction-Demo-Keeper is set up to use Uniswap V2 liquidity. There were several assets. When I go and liquidate Nexo in simulation on larger vaults, Uniswap V2 taps out right away. We get terrible prices with that one liquidity source. Across the board, it seems like Uniswap V2 is bad for many assets. You tap it out after a little while. A couple of these, like renBTC, Uniswap V2, was not enough to find debt to liquidate an even reasonably size vault. The Auction-Demo-Keeper, one of the myriad approaches to handling liquidations, is not ready for a giant liquidation wave. I am building into it the capability of handling Uniswap V3. The APIs and everything has changed in V3, so it is taking me longer. We do not have a backstop for wrapped stETH, which is an asset that I will pump the brakes on for debt ceiling increases until we know that we have at least one good liquidator for it. We have written the smart contract side of both things because our team is good at that. We do not have the portion to run a keeper or put this into the backstop keeper. In a simulation, the current Auction-Demo-Keeper did not perform well. It performed very well on Wild Wednesday because Uniswap V2 was the source of liquidity. We might not get that same level of performance next time around. If you are running a keeper, please speak up and know what liquidity sources you are using.

28:28

  • David Utrobin: Is there a way for future keeper versions instead of relying on a single venue to use a proxy like an aggregator and outsource the problem of trying to choose it at the bot level?

    • Christopher Mooney: If you can tool up one version of the Auction-Demo-Keeper, it may make sense to integrate with one 1inch or any DEX aggregators. You can use their APIs to find the best circuit and DeFi and construct that in the client. Then fire that off at one 1inch’s routers. It would end up making the trade. There is a good argument for doing that. You do not need to put together all of these things. We need someone to be working on this.
  • David Utrobin: The trade-off is that you get better pricing and better venue access, but it is less gas-efficient for the keeper.

    • 1inch is good about trying to find you a gas-efficient circuit. At the point that we are, liquidating a ton and we have tapped out one liquidity source, I would rather that we had keepers that were sourcing, even if it was more gas expensive, that could find liquidity that would reclaim the debt.
  • Kirk: One other issue with 1inch is that their web API is required to be online. You cannot route through the smart contracts. If they go down for whatever reason, those keepers will be screwed.

    • This is the defensive debt.
  • Kirk: You are taking a certain amount of trust risk on 1inch. Maybe you want to have the ability to sell through 1inch, Matcha, and Pairswap, all three, and build more.

30:26

  • David Utrobin: Is there a way to build fallbacks into the bot? For example, if 1inch fails, use Matcha. Then if Matcha fails, use something else. You build a schema.
    • Kirk: The more you add, the more complicated and expensive it gets in terms of effort. It is theoretically possible.

30:47

  • Christopher Mooney: There is a keeper group being signaled for now. I hope that they will take over a ton of development on this. The Foundation did development on keepers. This played an extremely critical role in helping us whenever we had large liquidations.
    • David Utrobin: Are you talking about Sidestream Auction Services CU?
  • Christopher Mooney: Yes, I am.

31:21

  • David Utrobin: It sounds like an essential part of being prepared is getting the liquidation keepers on point. What about the collaterals that currently have liquidations off altogether? What about the parameters of the liquidation system in terms of the ILK and the maximum amount of liquidations you can have at any given time? What about the price risk? What is Maker’s comfort level or approach with a billion dollars of price risk on ETH?
    • Christopher Mooney: From PE’s perspective, ILK was meant so ILK could be a collateral type maximum amount of on auction debt. That value should approximate how much liquidity we think will be available for that collateral type on-chain. The global whole should approximate how much liquidity we believe is available for all of Dai pairings on-chain. Those are more complicated circuits that you could make on-chain. The rest of the ecosystem is also liquidating. Gas is probably costly. You want to curtail these a little bit. It is a balancing game. You want to liquidate as much and as quickly as you can before you tear the whole market down and you get bad rates. If those are capped, we allow the auctions to take 40 minutes to an hour to recapitalize. You have this 40 minutes to an hour buffer where arbitrage can happen from centralized exchanges and begin to push itself into these pools or from more esoteric places and ends up pushing itself in. That is the answer to how we set those parameters.

33:47

  • Primoz Kordez: The ILK, which is this auction throughput parameter, should not be too big, or the on-chain liquidity will be completely drained. We look it through slippage numbers that we measure then limit it. The keeper system also constrains it. If we only have an implementation with V2, although there is a ton of liquidity elsewhere, you do not want to have too high a throughput. Even if you have high throughput and there is liquidity, you are risking these cascading liquidations. You might affect the prices. These are the limits to why you cannot liquidate as much as you would want. It is not ideal. If you liquidated the largest vault, it would take the 1.5 billion vault on ETH-DAI about 10 hours; That is a lot of market risk. Imagine that the ETH price drops further. There is a loss right there. It is not as bad as it sounds because, in some cases, if you are liquidating in smaller chunks, the borrower can still save itself. If you believe he can react within, if not in the first hour, that he can still save himself, if he can save himself in the second hour, then it is okay. Because you do not liquidate and you do not have the market risk. The user still stays with you, and it is not wiped out. You still collect to the stability phase. But you need to believe he is going to set himself. Exchanges liquidate as much as possible. They do not want to carry any marketplace. All these factors limit us.

36:02

  • David Utrobin: There is a third prompt. What happens in the event of sudden loss or invalidation of collateral? This question is less related to a bear market but more related to what happens when a major government deems a particular collateral illegal or Maker as a whole illegal. What happens if collateral recourse is stuck in limbo for ten years in some lawsuit and we are just sitting on a few million dollars of stuck Dai that cannot get liquidated? What should be the response if recourse is unenforceable?

36:58

  • Chris Mooney: Can we talk about the emergency playbook before we address prompt 3?
    • David Utrobin: I am the facilitator of Governance Communications CU. A part of our mandate was to support Maker DAO with emergency comms. An emergency is a whole domain in and of itself. There are proactive measures that the DAO does, even before an emergency happens, like threat monitoring at various levels. When an emergency happens, there are five or six different emergency types. Underneath those emergency types, there are derivatives within those types. When such an emergency happens, there are a few different coordination tracks that happen. There is a communications coordination track where internal and external comms have to be drafted, forum posts have to be drafted, information has to be aggregated, summarized, and shared, impact assessment has to happen, and written on and documented. There is also a technical coordination track where the engineering teams and other key people come into a war room and assess the exact situation regarding the emergency. There is also a general coordination track where it is not just engineering, but there are a few different things. Our team’s emergency playbook encapsulates everything into a single playbook with deeper links to more granular resources. We have a good first draft mostly done. It documents everything, as well as the processes for specific emergencies and emergencies in general.

39:35

  • Christopher Mooney: One of these specific versions of the playbook should be what happens in a massive collateral crash or a huge market tear down. In that case, we have several resources available to the Maker community and others that are partaking or are users in the space to help with liquidations. It might be hard to spin up a keeper, but the Auction-Demo-Keeper is not that difficult to spin up. We may have a whole track of people capable of spinning one up, and they may end up doing that. People may show up and have capital as well, and they should use the Web3 liquidators. They can also use the session we had last week of the Web3 liquidator that uses flash loans or this sidestream one. We have to communicate all of that on Twitter and other people to participate in those auctions and aggregate information about it. Something may happen if things go wrong, like on Black Thursday. We spin up a whole apparatus to try and deal with that and communicate.
    • On the market event, which is one major category of an emergency type that we have identified, we met with the risk team three weeks ago to define what a market event emergency is. We also have a preliminary bucket of updated and outdated resources. We have to check keeper resources and links to auction front ends. This preparation is something that our team at GovComms is focused on. We are focused on building it out in a high-quality way with others involved in this process, including many CUs like PE, Risk, Oracle’s, and more.

42:12

  • Nadia Alvarez: Christopher, do you have the list of the people running those keepers? Who can I contact to understand the current situation of the keepers? Is the signal group that you mentioned still active? We should double-check everything that used to work during the Foundation times and see if everything is still working. If not, we need to set up a new group or create a keepers room on our Discord. In case we need to help with liquidations, we can ask for help there. This could be private or public. During the Foundation, Greg and Joe were usually the ones coordinating with keepers. Do you or anyone else have these contacts?
    • Nikolai: The professional market makers who would run keepers out there are hard to get tabs on. They are not very open about what they are doing because running these keepers is like a war game. You want to have the best algorithm to make the most money. They are very intransparent about how they run their operations. With Maker, there is not a constant flow of liquidations. There are lots of opportunities in the market. They are always chasing after the highest yields. For example, in Maker, the person running a keeper would often shut it down suddenly because it is not worth optimizing their algorithm if there are only two liquidations every two weeks. It is very hard to keep tabs on because profits only incentivize these actors. There are many ways to earn profits and crypto. We must have open tools that anyone can spin up. We cannot rely on these private actors to run the keeper or have them up to date for a market crash. I know of some keepers that I can reach out to, but we cannot rely on them.

45:23

  • David Utrobin: Some people have voiced that our past experiences have been a blessing in disguise. They help us test in practice the health of the liquidation system. In my conversations with Unified Security and others, the non-technical side is interested in doing fire drills around emergency preparedness. When we launch collateral types, is it good to have standardized three, six, one-year processes to do some fire drills for the liquidation sites? How can we fire drill our liquidation systems better?

46:34

  • Nadia Alvarez: I like the idea of the grant, hackathon, or bounty. We should have the keepers chatroom on Discord. Explain how to run your keeper and have people answering questions and more. Our last experience with a hackathon was that the Maker protocol could be complicated for many new devs. They prefer to develop things for other protocols. We will need to have people helping these new devs or new keepers implement one, especially for Maker.

47:27

  • Someone: That is also the reason for SES. I am not part of SES anymore. However, I was part of helping incubate sidestream. That is to have a CU focused precisely on this problem: to have up-to-date liquidation tools for the community. Whenever a new collateral type is onboarded, we should efficiently liquidate it immediately, which is not the case today. There is always some tail risk because right now, the keepers are not updated immediately. They could also help maintain the Auction-Demo-Keeper and the documentation around how these things work so that the community can run it. They are not going to be like a keeper that is running it. They want to make it as accessible as possible for the community to liquidate these vaults efficiently.

48:35

  • MakerMan: During Black Thursday, we saw evidence of mempool manipulation. And nobody has talked about the lack of network access for all the tools we are talking about right now. It was unclear to me whether EIP 1559 helps us in that respect. I am not entirely sure what to do about it. That is the missing piece I see here. What happens if you lose network access? You will be dead in the water. If we do not have access to a network, we will not have any plan for dealing with or even setting up liquidations.

49:11

  • Christopher Mooney: If you lose access to the memory pool or the nodes, time will continue to tick, and prices will fall in the current liquidation system. We have a catch at a certain cusp or at a certain tau amount of time passing. The catch says that like 60 or 50% price, its forces at the auction will be redone. If we lose access to the mempool, the worst case would be losing access until the collateral right before the cut-off. Then someone would be able to fire through and take them for this cheaper amount. However, we would not see the one-way bidding as we did on Black Thursday. So, there is a bit of protection there.

50:11

  • Makerman: It was unclear whether 1559 would help. There might be a possibility that we could use an operative L2 when the L1 kind is not working well. It is a whole different mechanic that we would have to think about. We do have other networks. If we can source liquidity from them, we can provide a type of promise token. This token essentially promises that you will get it if we are selling and you are buying. This is a way to have a backup network if the primary network no longer functions. We might be able to source liquidity.

50:46

  • Christopher Mooney: Maybe once Wormhole is finished, there could be a way to use Wormhole for liquidation service. That would solve everything.

51:00

  • David Utrobin: I recently listened to a podcast that talked about provable fairness in NFT auctions. Has anybody been thinking about improvements on that side? Would that help with mempool manipulation? And is there a way to take things off-chain? For example, one of the things that smart NFT auctions do is they take orders, but not on-chain. They batch the orders until the time of the auction comes to an end. Then they execute, so the orders do not ever go on-chain. The execution is just determined after process and execution. I can probably post a link.

52:15

  • Christopher Mooney: Gauntlet had an interesting liquidation system that we evaluated when we did Liquidations 2O that sounded a little more like this. I think it was Gnosis.

    • Q: You are thinking of the batched auctions?
  • Christopher Mooney: Yes. We did not go with a batch system yet, or for this iteration, because the protocol would take on so much collateral. At this point, we knew that waiting for any kind of capitalized keeper to handle this would hinder us. We have tried to leverage single block composability, allow our auctions to go around, and source liquidity from as many sources possible on the chain that it is on. This will get interesting in the future as we think most capital will probably end up moving to L2, and L1 is going to be a settlement layer for all the call data happening on L2s. If that world materializes, anything on L1 will need some sort of wormhole liquidation strategy that allows us to handle liquidation and settle it over a week or a similar course of time. There will be innovations in the future.

53:49

  • Kirk: In that world, you must move most of your native Dai issuance to the L2 as well, so the positions are held at the same place. The actual size of the vaults you can support on L1 gets highly constrained by what liquidity is available on L1. I can imagine a possible future in which L1 is 100% roll-up commitments, and everything happens on L2. However, I am worried that the market is currently underestimating the speed at which this could happen. This is a risk. At what point does the 50% of Ethereum block space become all L2 to keep an eye on? Then, how long after that until it is 99% and so forth? L2 has an economy of scale where they can afford to pay higher gas prices since they are compressing so many transactions.

55:08

  • David Utrobin: We talked a bit about what Maker can do fundamentally to improve liquidation at its home base. But I am curious about a different kind of approach. How can we educate our vault users to use risk mitigation tools like deep buy more safely? Also, is it a good idea for us to push third-party risk mitigation tools as a protocol? What is our kind of duty and comfort level there?

55:43

  • Christopher Mooney: That is a great idea unless we want to rent safe liquidations. Currently, we have a liquidation penalty intended to stop auction grinding attacks. This is a way of a collateral crash; a bunch of unbacked Dai is minted. We have this 13% liquidation penalty to stop this. But it turns out in practice that the liquidation penalty ends up creating an enormous amount of income for the protocol, which is a side effect of it. We should continue to think of it as a side effect of liquidations. Otherwise, we are too likely to try to rent seek that particular mechanism. As a result, I think it makes sense. Defi Saver is doing a ton of work ahead of time. This is a major feature for users as well. This delayed Oracle price allows them to see into the future and stop themselves from liquidating these services. That is a major selling point over other protocols.
  • We should lean heavily on that or think about alternative liquidation schemes like B.protocol. However, we should leave the liquidation system that we have as the defense in debt. This is the Swiss cheese approach. You can make it through the first layer of Swiss cheese. Hopefully, we will catch you with our liquidation system after that. The stats say that most stETH growth is happening through Defi Saver and has automation turned on. This means that our stETH growth presently is being protected by Defi Saver despite not having a liquidator to service it.

57:37

  • David Utrobin: Good point. Nikolai also puts up a really interesting point. Vault protection makes us more money because users are more comfortable getting close to the liquidation ratio. I never considered that. On a side note, based on my reading of the forums over the last couple of years, there is high consensus that DAO members eventually want MakerDAO to have zero liquidations. As you mentioned, that 13% penalty is a function of a game-theoretical necessity to prevent auction grinding attacks. I have always seen support for third-party risk mitigation. I am less concerned about the rent-seeking part based on the views of the collective over the last couple of years.

58:38

  • Christopher Mooney: This is maybe debatable, but on Wild Wednesday, we ended up having so many profits that the burner kicked on hard. Multi-millions of dollars went to the burner. When I originally was looking at the graphs of MKR, it appeared to be supporting the MKR price. However, I was less convinced, so I think the jury is still out. If that was a happy accident, that is almost exactly the type of virtuous thing that you would want to happen to MKR. As the markets were crashing, this would support MKR if it needed to flop MKR later.

59:18

  • Kirk: I must jump in and give an alternate perspective. In the happy case, prices dip, the system buys the dip, everything goes back up, and there is no bad debt. That is great. But let us say the market dips, and we get some surplus from penalties. The surplus is used to buy MKR, and then the market dips a lot more. Something then goes wrong, such as network congestion or liquidity constraints. Now we end with a lot of bad debt. The MKR price is now a lot lower than what the system bought it at. If we must reprint it MKR to what was burned previously, a lot more will be printed than what was burned. We would have been a lot better off just saving that surplus for the crash. Thus, it is an asymmetric risk in which the downside is worse than the upside. I would just counter against the thought that buying and burning is a perfect capital battery.

1:00:32

  • Makerman: I talked with AES about gas cost and trying to figure out our emergency buffer. We were honestly thinking that the time you wanted to buy ETH is when it’s crashing because that is when your gas prices will be highest for Oracles. We might want to reconsider what we buy with our excess Dai when we are crashing.

1:00:56

  • Kirk: There is a big debate to be had about how well the protocol should diversify its treasury because that is essentially what is happening. If you are taking the surplus buffer and buying, let’s say, ETH, now you are diversifying your treasury into other assets. What mix of assets does not make sense for the DAO to hold? This is an interesting discussion with a lot of room for creative improvements.

1:01:33

  • Prose11: We will encourage everyone to bring up any topics that we have not gotten to yet. This meeting is nearing an end, but I appreciate all the people participating in the discussion today.

1:01:51

  • Seth: I can speak on what our team has been discussing in educating vault users. The best policies are agnostic towards what solutions we encourage people to use and focus on educating people on what solutions are out there with their pros and cons. We have considered making educational videos but decided it was not a good idea because it is harder to do. An evergreen video would be one specifically about how to use Oasis, but not Defi Saver. Then obviously, that looks like we are favoring one over the other. At least for now, we have focused on getting this information into their community portal. Fairly basic so far. I look forward to adding more information. If anybody has other thoughts, please reach out.

1:03:07

  • Makerman: I want to go back to the PSM. One of the things that had changed since Black Thursday is that the impetus to mint Dai before was a high Dai price. We do not have that impetus anymore. How is the PSM going to act in different situations? There were discussions previously saying that we might have different kinds of activity concerning Dai demand and minting. Given that the price is effectively locked, it is important to consider the unclarity of what will happen.
  • There is a general tendency to think that the PSM will just fill up. Nobody discussed what happens if USDC or stable coins do not want to move in for the price. The peg then goes high because people value their stablecoins, like USDC, over Dai at that moment. If the peg runs high, somebody then might decide to take advantage of this. The real price that makes people move here. When you lock the price, you are left with risks and rates. At that time, your rates are not going to do much. Then, you are just left with dealing with the risk as the markets deal with it.

1:04:57

  • Kirk: I am going to jump in here again. When we did have that so-called natural incentive, we saw that it was extremely weak. There was this natural incentive after Black Thursday. Dai prices were high, but people were either too scared of putting money in Maker or just did not appetite for leverage to take advantage of it. I agree that if you waited long enough to let the price get high enough, it would have perhaps eventually motivated people to mint. That is an empirical question. The thing that we saw was that Dai adoption was being abandoned. For better or worse, Dai’s integrators had come to rely on having a very tight peg. They could not tolerate from a business perspective a four or five percent persistent deviation. So, the market told us it does not want a soft page currency. It is not good for us. It is very hard to go back and reverse that expectation once it gets locked in.

1:06:19

  • Christopher Mooney: To the second part of Makerman’s question, if the market wants to hold on to USDC or Dai breaks peg, high, there will be an interesting situation with liquidity providers. Liquidity providers are going to allow there to be an arbitrage because they are asleep at the wheel. If people prefer, let us say Dai over USDC, or USDC over Dai, or whatever, it will get worked out through sources. It is not until we tap those pools that we are in danger.

1:06:58

  • MakerMan: In general, I agree. That was why I wanted to propose, with others, to push for USDC during Black Thursday because we were stuck. We had no collateral that wanted to move into mint DAI. We also needed to fill a 6 million dollar hole. There is this real question of what is good collateral that is not going to liquidate? We put up one that absolutely would not liquidate. We set it to one and said we were not going to liquidate this collateral. Everybody was like hallelujah, and then we made it into a vault.
  • I was looking at the gUNI V3, which has a very tight band that is supposed to move around with the bots that will post a new transaction to float that liquidity into a new band. But what if it cannot? Then you are stuck. The key liquidity that will source here is V2 because it will never go away, or it must pull. If it cannot access the network, then it cannot pull. If anybody can access the network, they cannot drive it to zero. Whereas for V3, they can pass right through that band and can drive that V3 price rate to zero-rate with that liquidity straight up.
  • When you ask where the liquidity is, I keep insisting that we need to have a real price band. When you have a price band, you have fees, and fees will find liquidity. That liquidity is going to be there potentially to be sourced if it’s in the right configuration. If you have to do a bot and post a transaction on the network, everybody and his brother will be fighting for those transactions. The worst time to try and manage your network and be prepared with liquidity is when you do not have it, and there is impaired network access. You want to have things set up to be ready when needed and not struggle to find them.
  • In principle, I agree with you, Kurt, or Brian. During that time, all your basic liquidity is bad except for the stablecoins. You are stuck knowing that the PSM is all that is left, and it’s pretty gas inefficient. And then you have a vault, maybe. There are few places where we can deposit collateral and find liquidity to help the price drive in a direction. It is a standing issue, and we will see what happens if it ever happens. So let us hope not.

1:09:26

  • Someone: We do have liquidity in that type of emergency crisis. That is quite literally what trust, cash, and Treasuries are there for. There is mitigation for that type of epic crisis.

1:09:45

  • David Utrobin: Is certain liquidity available but has no precedent to be used? For example, we have liquidity in our treasury, but what is the precedent for our auctions keepers? Should Maker protocol be one of the participants?

1:10:05

  • Christopher Mooney: Our treasury will naturally cover bad debt. Whenever we liquidate, it immediately comes out of the surplus buffer. The first thing we will see is a giant liquidation wave. Everyone is going to see the surplus buffer drop by insane numbers. Your first inclination is going to be Black Thursday, but you do not need to panic. That is what happens. We will then reclaim Dai from those auctions put it back into the surplus buffer. It does so naturally.

1:10:46

  • David Utrobin: I see that is going to trigger coverage and indications.

1:10:50

  • Christopher Mooney: As I said, do not panic if you see the surplus buffer drop in a wave of liquidations. This does not necessarily mean taking on bad debt; that surplus buffer was just hit first. Then, it will recapitalize if the auctions are performing well. If they are not, then that is going to become a really bad debt.
  • So on to Matt’s point on the third slide. We can roll our sort of PSM with real-world assets as well. There is this capability we have now that we have real-world assets. You could even think of a successful facility as a real-world asset and emergency. You could especially set one up independently. If Dai is going high, you can mint effectively unbacked Dai and immediately drop it on the market, which drops the peg.
  • When you drop that Dai on the market, it moves through the broker-dealer that will sell it on the market. That then turns into US dollars in a bank account. It could also turn into a bond. This would allow for more downward pressure on the peg, even if the USDC PSM did not exist. Essentially, we have our way to do a PSM now. In the opposite direction, it is the same thing. You buy Dai off the market and repay the debt amount. You just got to set the loan facility for that to zero. We do not need USDC anymore to do a PSM. We can do it ourselves.

1:12:28

  • David Utrobin: Just to clarify, would that be empowering a particular actor that is already in our system to be able to remedy a market liquidity crisis? We would still need collateral, but we would be able to print Dai and then collateralize it by using that Dai to buy things like treasuries in US dollars. Is that the right way of thinking?

1:13:31

  • Christopher Mooney: Yes, exactly. It is our USD bank deposits if we use a facility like that.

Open Discussion

Questions

1:14:58

  • Prose11: I appreciate that reflection and opportunity for different ways to keep the protocol safe. Keeping an eye on the time here, we are nearing half past the hour, usually when we tend to wind things down. I do want to give people one more chance for questions or comments.

1:15:30

  • Kirk: I will leave people with one last shower of thought. We have discussed different types of risks and have focused on the more important ones for sure. That was the correct one to focus on. But there is a risk in the other direction. Imagine that we do not have a major global macro deflationary event, and the US keeps printing dollars to fuel this bull run and risk assets. Imagine getting to a point where all the PSMS are empty, and Dai is trading at 98 cents. At that point, our solution is raising rates and raising the DSR. How quickly could we do that? What if we get into the situation where we are raising, and it is not working? What does that world look like? I am not saying we will necessarily get into it, but you should keep in mind the possibility of meet-ups and meltdowns. We may find ourselves in that world at some point.

1:16:40

  • A: Let me comment on that from the real-world trust side of this equation in addition to lending out capital. There is another piece to this called the next evolution of the phases. It is indifferent to the scale of what I am trying to achieve and then ultimately changes the scope and scale. There is an entirely new level of bringing on a full-on bank revolver, engaging with the same collateral while they’re segregated, and having a facility set up. In that scenario, if Dai went down to 95 in the natural market that Maker is inside of crypto and the external side, the exact reverse equation would occur. We would be refinancing the existing debt structures we have outstanding with another revolver, paying back 100 cents on the dollar to turn right back around and repay the loan that we had in Dai, buying it at 95 cents, pushing it back up. The whole benefit of using real-world assets is that there is an entire capital market to push that—there is a natural arbitrage that is a Market Maker.

1:17:52

  • Christopher Mooney: Besides that, vault holders naturally want to repay their debt because they got their interest rate for free. Vault holders end up having a natural incentive to repay. We can also start to raise rates that will disincentivize creating new Dai and incentivize closing positions. It is not a rent-seeking thing. What we did not have in 2019 is the DSR. We can then also stimulate demand for Dai on the reverse side. We can increase the DSR, and then people want to hold Dai. We have a lot of tools on the downside. An interesting idea about the PSM has more automated governance around the tout. The idea would be to create this buffer. There was a discussion the other day in; I think, governance or speculation where you create a buffer of like 500 million. Suddenly, the tout starts to move up a little bit, giving us more time to react while preventing us from hard Dai market shocks that drop below 98 cents.

1:19:30

  • B: Yeah, I just wanted to ask because we work on that with promotions. But that currently, we don’t belong Matteucci assets. We have a few hydrolysate sets very small currently. But if we get some Term Lending and institutional work is part of it. That means that we cannot raise the yield on those. Or if we do, it will be kind of breaching the document agreement we have with the institution And then you can be in a case where you have, let’s say, 8 billion of loans at 1%, which are term loans, we are not allowed to move them for the next six months. And maybe in the future, it could be directly in the smart contract. Then you have the GSR. That is, you do not have anything else in the PSM; you increase stability fees on ECE a and others, but it says there is nothing left, and they continue to fall. We can increase the DSR. If you increase the DSR above 1%, you will bond cash because you spend more on Dai than what you get from the bonds or the loans. So it can be tricky. It’s not a problem for this year, not for next year. But it will become more and more tricky going forward because you can have a material mismatch between the assets and liabilities.

1:20:58

  • Kirk: The PSM we currently have is instant relief. Once the PSM is empty, and Dai starts deviating below the peg, you can be in this situation where you theoretically have tools to deal with it. But the timescale of that response might be much longer than would be ideal. If we have large real-world asset loans that can nimbly move to lower their cost of capital, that could be one mitigation.

1:21:49

  • David Utrobin: Another important thing is something that Joshua just brought up. IAMs are a tool that governance can use to implement a fast change somewhere through specific permission. My open-ended question for the DAO is, are there parameters that should have an IAM?

1:22:23

  • Christopher Mooney: The short answer is yes. The long answer is that every sort of automation comes at the risk of doing it incorrectly or with some peril. We should be heavily thinking about how to governance minimize what we can, mostly because the cognitive load to just manage all of this is going to far outpace our capacity as humans to handle it. For parameter changes, it would be good to think about these things ahead of time and set up an instant access module that behaves correctly in all these different scenarios. But again, you got to be careful. We cannot easily set rates on a chain right now; the formulas are complicated.

Conclusion

Payton Rose

1:23:37

  • Prose11: Well, that’s our time. Let us keep the discussion going on the forums and Discord. Thank you, everyone, for attending.

Suggestion Box

Common Abbreviated Terms

CR: Collateralization Ratio

DC: Debt Ceiling

ES: Emergency Shutdown

SF: Stability Fee

DSR: Dai Savings Rate

MIP: Maker Improvement Proposal

OSM: Oracle Security Module

LR: Liquidation Ratio

RWA: Real-World Asset

RWF: Real-World Finance

SC: Smart Contracts

Liq: Liquidations

CU: Core Unit

Credits

GovComms Core Unit

Summary Team:

  • Alefcripto produced this summary.
  • Artem Gordon produced this summary.
  • Kunfu-Po produced this summary.
  • Larry Wu produced this summary.
  • Everyone who spoke and presented on the call.

This full call is now available for review on the MakerDAO Youtube channel:

2 Likes