[Agenda/Discussion] Scientific Governance and Risk #99 - Thursday, July 2 9AM PST (4:00 PM UTC)

Agenda

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

Governance

MIPs

Oracles

Risk

General Q&A

We’ll open the floor for any questions about Scientific Governance and Risk.

Please join us and help shape the future of the MakerDAO.

Links

Call Summary

Will be provided here after the call as time allows.

1 Like

Episode 99: July 02, 2020

Agenda

  • 00:00: Intro with Rich Brown

  • 01:17: Governance at a Glance with LongForWisdom

  • 02:51: Weekly MIPs Update with LongForWisdom

  • 03:55: Oracles Proposals Update with Niklas Kunkel

  • 07:42: COMP Liquidity Mining Impacts with Cyrus Younessi

  • 49:49: State of the Peg with Vishesh Choudry

  • 1:04:42: Open-ended with LongForWisdom

Video

https://youtu.be/Rn2MGdQ4L7A

Introduction

LongForWisdom

Agenda Summary

00:00

  • Hello, everyone, and welcome to the 99th edition of Scientific Governance and Risk, taking place on Thursday, July 2nd. I’m LongForWisdom, one of the MakerDAO Governance Facilitators, and I’m joined by a bunch of Maker people.

Governance

LongForWisdom

Governance At A Glance

01:17

Governance at a Glance Forum Thread

COMP Liquidity Mining Impact
Seeking Signal: wBTC Debt Ceiling and Risk Premium
Vault Compensation Plan
  • I would just like to highlight as well that the Vault Compensation Plan thread is in the forum because we could still use some feedback. So if anyone has some time, please add comments.
Community Greenlight Polls
  • The second set of community greenlight polls are still on-chain! Please vote to signal your support or opposition for the domain teams spending time on adding: ENJ, MATIC, WKT, LEND, PAXG, LINK, and KNC to the Maker Protocol.

MIPs

LongForWisdom

MIPs Weekly Update

02:51

  • This week we’re in-between two governance cycles, so there is not much of MIP specific content today.
MIP submissions
  • MIP 16 Introduction, The Weekly Governance Cycle, by Charles St. Louis.

    • MIP16 works towards codifying the weekly Governance cycle. It’s now in the RFC phase and will probably be submitted to the July governance cycle during the formal submission period, which starts next Monday, until next Wednesday.

    • Next week, we’ll be able to tell everyone what’s happening in July, hopefully.

Oracles

Niklas Kunkel

Oracles Proposals

03:55

Overview of Last Month
Next Step: Executive Proposal
  • I’ve been assisting those teams with setting up their new feeds. Looks like everyone is ready to go, so we will be batching those into tomorrow’s executive proposal for final ratification from the community(passed.)
Proposal from Argent
  • We also got a proposal from Argent submitted today to become a light feed. That’s going to be something we process in this month of July.

Discussion 1

Marginal Returns

05:00

  • Rich: I have a question for you, Nik. Is there an upper bound on light and dark feeds where it becomes less useful?

    • Nik: It’s implied. I think I spoke about this on a previous call. One of my little submissions is to reach an equilibrium in this balance of power between dark feeds (anonymous identity) with the light feeds (public identity). I’ve heard some people say, “why don’t we go super hard on the light feed ones?”. I think it’s a double-edged sword. You don’t want the community to one day turn against Maker and have the power to completely screw over just based on sentiment. In that sense, the dark feeds can be really useful. The idea is that if there is a sort of an equilibrium between the two, and you set the quorum higher, overall, it makes the oracles more secure. That is why we’ve been adding so many light feeds lately. I think we will continue to add light feeds at an accelerated pace, until we reach that equilibrium. You are absolutely right that there are diminishing returns. Each of these feeds are paid 1,000 Dai a month, currently at least. We could always change that down the road. But it’s a trade-off between adding cost to the system and the extra security of one more feed. The returns do become marginal at some point.

Risk

Cyrus Younessi

COMP Liquidity Mining Impacts

07:42

  • I’m going to start with how I perceive COMP farming and how it has evolved over the last few weeks. If anybody wants to jump in and correct me, more than happy to have that.
Farming COMP mechanics
  • Compound launched their governance token roughly three weeks ago, and they had a very novel distribution mechanism where the token was allocated to users of their protocol. Specifically, it was weighted towards users who were paying a higher percentage of interest on their borrows and earning a higher percentage rate on their assets supplied. As a result, in order to earn the highest amount of COMP, you would want to essentially pay as much interest as you can, which sounds backward until you realize that the amount of COMP earned more than compensates for the amount of interest paid. In addition, you could loop through, or recycle, through these assets by continuously borrowing and resupplying the same asset, with the limit being their collateral factor for that asset.

  • So opportunists went in search of the asset on Compound that would pay the highest rate of interest. These rates are determined by each asset’s individual interest rate curve.

    • For example, if the USDC interest rate curve caps out at 15%, but the BAT interest rate curve caps out at 35%, then you would want to borrow as much BAT as possible instead of USDC.
The BAT Market
  • There’s fairly little price-risk involved because you’re borrowing and resupplying the same asset, more or less, but this does open up users to a bit of liquidity risk, so we did see some distortion on the BAT market over the weekend. The BAT market blew up as a result of these incentives. We saw, I think, as high as 350 million BAT supplied versus 320 million BAT borrowed. Of course, the largest portion of this was recycled BAT. So it’s a little bit hard to quantify the total amount of leverage embedded in the system.
Compound’s Governance Proposal
  • There was was a fair amount of concern on the implications of this large BAT borrow market were, especially with the liquidity risk that I was mentioning before. So their governance community made a proposal to alter the allocation weighting scheme for COMP farming. So the new proposal, which was ratified two days ago, is scheduled to go live on-chain later tonight works as follow:

    • Instead of paying the largest amount of interest, an ideal farmer would want to borrow the highest amount of dollars possible, denominated in dollars.
Impact of Compound’s Governance Proposal
  • This changes the game theory a bit because now you no longer want to pay the highest amount of interest, but the lowest, so your COMP rewards are diluted the least possible. So you would go and search which assets pay the least interest at the highest utilization rate, and that turns out to be Dai and USDC and potentially ETH as well, depending on the utilization rate.

  • At this stage, it becomes a little bit confusing trying to figure out which asset is the ideal. There are a lot of nuances, especially when it comes to evaluating the utilization rate. Also, if you’re recycling farming assets in an asset where there is an enormous amount of excess supply, then potentially, it will dilute your COMP yields. So there’s this little debate whether Dai, ETH, or USDC will dominate.

Mitigation Schemes
  • I think this is a discussion worth having and definitely worth understanding. But the more important thing is that given that there is at least some non-negligible chance that Dai is the ideal asset, this community should start working on mitigation schemes; what we want to do to prepare for market impacts. Over the past few days, there’s been a number of proposals on the forums, so that should be interesting to carry on and explore.
How Comp May Cause Dai Demand
  • From the governance perspective, here is how I think about the whole situation. So imagine if some third party service was offering a very high DSR, then you could see a potentially huge influx of demand for Dai. And at the same time, the same third party is providing a source of very negative stability fees, which makes borrowing very attractive there. We could potentially see a large refinancing activity from the Maker platform to Compound. If we see both large influx of demand to get this external DSR and refinancing from Maker to Compound requires sourcing Dai on the secondary market as well, so that’s where the primary demand concerns come in.
Potential Questions for Governance
  • How would governance respond to such a rate-driven environment?

  • Very simply, if demand goes up, then we want to increase supply. Most of the solutions over the past few days in the forums are variations of:

    • How do we increase Dai supply dramatically, maybe drastically, in a short time?

    • What’s the safest and most sensible way to do that?

    • What concerns do we have on increasing the amount of collateral in the system?

USDC-based solutions
  • For example, from a price standpoint, the easiest way to increase Dai is USDC. Seeing last night’s large mint, and potentially more to come if we increase the debt ceiling, there’s an opportunity for the community to turn this into a very positive outcome if the right SF and the right DC are found. There’s definitely an opportunity to keep the peg in line, and expand Dai supply, even if it would be mostly USDC-backed. I think that’s been the forefront of the discussion. It’s some change to the USDC parameters in addition to raising the debt ceiling and the stability fee.

  • A really common suggestion has been to decrease the liquidation ratio, so smaller market makers would leverage up to even more Dai, putting further downward pressure on the peg.

    • There’s this question of what happens if farmers mint Dai out of USDC from Maker and stuff it into Compound instead of selling it off? That’s what we saw last night: 20 million Dai generated, and I think all went into Compound. It wasn’t necessarily used to push the peg down. This is where governance needs to normalize significantly larger amounts of collateral and Dai supply in the system if they go down that route.
Centralization Risks
  • If you’re worried about centralization risks with USDC, a reasonable assumption is that beyond some small threshold of 10-20 million, it’s my hypothesis that it’s probably too much centralization risk for Maker to handle at this time, then there’s no difference between 100 million and one billion. Just as a thought experiment: if the amount of Dai that maker would need to keep the peg in line and everything in check was a Billion Dai backed by USDC, Would the community be for that or against that? Especially if there was a stability fee put in there for good measure. I think that’s a very important question for Governance to consider.
ETH-based solutions
  • Other supply focused actions are a second ETH collateral type, ETH-B, with a higher SF. Certainly, there will be a lot of ETH-denominated market makers to get into the farming activity, and that could be another source of supply generation and potential SF generation.
Outside of the Box Ideas
  • We’ve also seen outside of the box ideas. I think MakerMan suggested some sort of reserve or open market operation that would definitely be a huge departure from what Maker has been doing until now. There have been quite a few different and diverse proposals out there, and unfortunately, time is of the essence. That proposal goes live tonight. I’m not expecting anything to happen right now, but based on the activity that we saw with BAT a few weeks ago, over the next days, the situation will be much more clear.

Discussion 2

Are interest rates the cause?

14:30

  • Befitsandpaper: Is Dai right now better collateral for putting on Compound because of the USDC-A interest rate on Maker is lower than on Compound?

    • Cyrus: Not necessarily. One of the ways to disincentivize Dai is to increase Maker as a primary source of Dai. If the USDC rate was to go up, then we wouldn’t have seen that large mint last night. Basically, for that person, it was cheaper to borrow Dai from Maker rather than Compound. However, that is not really at the core of why Dai is a popular farming asset.
USDC-A Stability Fee

22:40

  • Lev Livnev: It’s very hard to predict what would happen over the next few days. But there are also some shorter-term things that are more clear. Yesterday, someone took out 16 million Dai against 19 million USDC, maxing out the USDC-A debt ceiling. This has two important implications:

    • The USDC-A is now exhausted, which is not good already. That collateral type is there to relieve Dai supply, and now that’s gone. So should we try to address that? The obvious thing would be to increase the USDC-A stability fee to a point where that sort of activity wasn’t attractive. The advantage of tackling that problem is that we can also look at the economics of that trade to see what makes it attractive and what level of stability fee would make it unattractive.

    • The second thing is that we could look at this as an example of what people are doing with Dai in Compound. But I thought it’s worth starting a discussion with the USDC-A because it’s at hand now.

USDC versus Dai in COMP farming
  • Cyrus Younessi: Can you briefly walk us through the mechanics of why you think that an increase in the stability fee of USDC-A is helpful? Because I think I could easily argue that it’s hurtful at the same time.

    • Lev Livnev: Sure. I’m speculating about the intentions of this trader, but we would have to ask ourselves why someone would put up USDC into a vault, and take out a bunch of Dai to put it into Compound. I don’t think the answer is straightforward because you can’t rule out reasons such as being worried that in the future, that debt ceiling will be exhausted, so they wanted to take it while they had the chance. There’s this dynamic every time we get close to the debt ceiling. It could have been that. And of course, parking it into Compound is not an unreasonable thing to do, once you got this position. And the interest is very low, so the cost of doing this is pretty low. That is the most naive option that has nothing to do with COMP farming. This is quite unlikely because this user is a huge COMP farmer, and this Dai went straight into Compound, and they have this huge BAT recycling going on, which is actually the main source of income for this COMP farm. The second option is that, for some reason, there is superior economics to putting USDC into Maker and then putting Dai into Compound instead of the more direct route of just putting USDC straight into Compound and then do this BAT recycling. My guess is that currently, these considerations aren’t dominated by the actual interest rate, the stated interest rate on Compound for Dai versus USDC, even if indeed the supply rate for Dai is a little bit higher than the one for USDC right now. I think the reason probably is the ratio of the total amount of USDC supply to the total amount of Dai supply on Compound. There’s way more USDC supply, which is not surprising because it’s easier to get and dump into Compound. On the COMP yield side of things, you’re making much more as a Dai supplier than as a USDC supplier for the same amount. I think the data actually dominates the actual headline interest rate. The practical next step is to decide, “is there a sensible level of SF that would make that trade no longer attractive and make the USDC go to Compound directly?” That would have the advantage of taking all this Dai against USDC and freeing up some debt ceiling, that could be used in the future for potential demand shocks.

    • Cyrus: To sum up; The stability fee should be high enough where it’s still profitable for people who want to arbitrage the peg but not profitable for people who want to do other things with that and waste our DC space.

    • Lev: That’s what I’m thinking about. I haven’t checked the numbers on this, but if it turns out that a 10% SF for USDC-A makes this trade no longer preferable to go into Compound, then I would think that it would be worthwhile because the primary function of USDC is to meet demand. It’s probably going to be less effective when Dai is at 1.01, but it will definitely still be effective when Dai is at 1.05 or something. I think that would be a comfortable middle ground where we would get USDC-A back as a tool.

New COMP distributions

28:50

  • Cyrus Younessi: I think that’s going to be hard to forecast because the distribution is going to change dramatically starting tomorrow. Right now, most everything is in BAT, and the annual yield is around 70%, depending on the COMP price. But with the new distribution, if it splits two or three ways, who knows how much dilution there is within a particular asset then. I don’t think we would know what kind of stability fee we would need to incentivize things one way or another.

    • Lev Livnev: I agree. Just to clarify, this trade that we were discussing now, it makes sense given the current pre-patch COMP distribution. As soon as the new distribution goes live, I’m certain these users are going to start moving things around. This huge recycled BAT farm is going to become suboptimal, so they’re going to actually switch to another asset. Whether they keep Dai as a collateral base is also unclear. There are multiple phenomena here that we’re discussing.
Liquidation Ratio Versus Debt Ceiling for Creating Dai Supply

30:20

  • Cryptowanderer: I’m not sure why lowering the liquidation ratio, which prior to this call seems to be the most popular option to do first, would help. If we haven’t activated liquidations on USDC, and people can, in theory, go below 120% without being liquidated, what effect does lowering that ratio actually have on the amount of Dai that people can get?

    • Cyrus Younessi: My preference is definitely increasing the Debt Ceiling first. Especially because we know that at the current liquidation ratio, people are willing to take out Dai, as we saw last night. The point of lowering the liquidation ratio is just to accelerate Dai generation. As an extreme example, if the liquidation ratio goes from 120% to 101%, a market maker with 1 Million USDC on hand can generate 100 Million Dai of sell pressure, which is enormous. Of course, the trick is that they take that and try to arbitrage the peg instead of continuing to COMP farm with it. The simple answer is that it gets more Dai out there quicker with less collateral.

    • Lev Livnev: The debt ceiling is important first, and then we can think about the liquidation ratio. If the debt ceiling is maxed out, it doesn’t matter what the liquidation ratio is. I think that in the grand scheme of things, the debt ceiling is what will determine supply rather than liquidation ratio.

Are Polls Going Out Today

32:42

  • Chris_p: So are polls going out today? Is that what’s going to happen?

    • Cyrus: Does the Maker community want to wait and see what happens? 24 hours? A day? Two days? One could argue that doing anything preemptively could be smart and ahead of the game or could be rash. It’s an open question. Additionally, over the next few days, depending on how things turn out, it might be worth it to consider skipping governance polls if we need to, basically ad-hoc, or pseudo-emergency executive votes, especially given the poll turnaround time, and the GSM.

    • Chris_p: And with the holidays coming up too, lots of people will not be at their computers.

    • Cyrus Younessi: Coordination is definitely going to be a huge challenge over the next four or five days.

Next Steps
  • Lev Livnev: I would say that the most important thing would be to discuss what actual steps should be taken. I expect that a lot of assets in Compound will get shuffled around after the patch, which makes some of the possible calculations incorrect.
Stability Fee

34:40

  • Unknownname: If the SF for Dai is increased, but the demand for Dai for people to put into Compound stays the same, basically it transfers the pressure from people who are minting Dai to people buying Dai in the open market and probably pushes up the Dai price maybe significantly if people cannot make more Dai and relieve that pressure.

    • Cyrus: that’s a good point. When you increase SF, that’s typically what’s used to push the Dai price back up. What we want to focus on is do whatever we can to increase supply, so definitely a good point.
Emergency Procedure

35:49

  • LongForWisdom: We have had emergency polls and votes before, so that’s fine. In order to be comfortable doing that outside of the weekly cycle, as a Governance Facilitator, I would want either a clear forum poll asking, “does the community think that we should take emergency action to fix this, outside of the regular cycle?” Ideally, with a clear idea of what we want to do as an emergency action. Either that or, if Cyrus, as the risk domain team, thinks that we should take emergency action. Then it would make sense to do that. If we think we should act now, we could potentially do that. Alternatively, we can wait and put some polls up on Monday. Or wait, and if things get worse in the next few days, then we can do emergency actions then.

    • Chris_p: So does anyone know when the new patch is going to be applied? Is it tonight?

    • Cyrus Younessi: I believe so.

    • LongForWisdom: I think it’s 2h15 from now. I could have the timezone wrong.

    • Chris_p: are people going to be paying attention on July 4th? Could we have another governance meeting potentially if things look bad? We should have somewhat of a plan to at least work over the weekend where the market could be moving really fast, and a lot of people might be away?

    • LongForWisdom: We can definitely do emergency governance meetings; we have had those in the past. Either Cyrus says that we need to do something as an emergency or the community votes, and there’s a consensus to do something immediately.

    • Cyrus Younessi: For what it’s worth, I’ll be monitoring this.

    • LongForWisdom: Likewise.

USDC Stability Fee Signal Request

38:58

  • Andy Tudhope: Since the USDC debt ceiling is maxed out and in line with Lev’s points, it might be potentially a good idea to get ahead of that. Cyrus, while I agree with you in principle, if we decide that the centralization risk is sitting between 30 to 50 million, and there’s, in principle, no difference between 100 million and 1 billion debt ceiling, I think that in practice there is a difference, and it has to do with the fact that governance is largely about optics. The community is doing everything it can to defend the peg. A preemptive small raise on the USDC-A debt ceiling might not be a bad thing to poll in the forums now after the call, just to prove that this is something that we’re carefully watching and considering for the wider community, people outside this call.

    • Cyrus: That’s a very valid point. I would suggest after the call we put out a signal request. At this point, I’m not even clear on what sort of options we want to put on the table. I would love to get some clarity on that, either on this call or afterward on the forums, and we can start working on a governance poll.

    • LFW: It would be good to have a better idea of what our options are in various situations, which might be good to try to figure out in advance.

    • Chris_p: Immediately, we all agree that there’s an issue with USDC-A vaults. Maybe we just try to kick a poll with some parameters around, raising the debt ceiling, increasing the stability fees, and taking it from there. That’s the type of action that we’re very comfortable doing. We have a lot of history raising stability fees, and lowering and raising debt ceilings. It seems very natural; I guess that’s what I’m saying.

  • Cyrus: anyone else? Lev?

    • Lev: I would probably lean against raising DC without raising the SF because that would be potentially eaten by the same trader since we didn’t figure out at what stability fee it actually breaks even. On the other hand, if we are increasing the stability fee, it could be appropriate to increase the debt ceiling. But just feeding this trade, assuming that this behavior isn’t irrational, doesn’t help.
Stability Fee Raise Tradeoff

42:58

  • Cyrus Younessi: Don’t we risk raising the SF high enough so that it’s no longer profitable to do the intended function, which is arbing the peg down? Don’t you almost need to keep it low enough until all the excess demand from Compound is diluted away, and then they start arbing the peg?

    • Lev: I agree we’re trading off, that setting it too high would be counterproductive. But you have to keep in mind that most of the debt ceiling is currently being used on this COMP trade. This means that by clearing that up, you could potentially get more room than from scaring the so-called legitimate part of that Dai generated from USDC. Secondly, I think it’s acceptable to raise the DC if we come to the conclusion that a good number would be somewhere in the 3% - 13% range. You could still imagine people using that for peg arbitrage at Dai price of 1.01 and higher. I think that heuristically that seems reasonable. And basically, we’re in a much worse situation if we have no USDC rope at all. Imagine if another exogenous situation were to happen right now, like ETH getting bearish. We have no rope at all in the USDC-A debt ceiling. It’s not a good situation.
Signal Request Threads

44:55

  • Cyrus: What about a signal request thread for the debt ceiling and a signal request for the stability fee? Maybe we can put those out today, and we put out a governance poll maybe tomorrow so that people have a chance to see what happens during the course of today and take it from there?

    • LongForWisdwom: If you want to make sure that both pass or none, you would want to make a combined poll.

    • Cyrus: Makes sense.

  • [Signal Request] ETH-A Stability Fee and Debt Ceiling adjustments

Analysis

Vishesh

Relevant Links

State of the Peg

49:49

  • Dai peg has come up in the last 48 hours. COMP activity and ETH prices looking good. The combination of both things creates a liquidity crunch on Dai. Prior to the shift in secondary-lending-interest-rate-models for Dai, there was a fair amount of buffer in the amount of available supply for borrowing. From that perspective, the impact on the liquidity crunch was mitigated. But I do think it will continue to have a potential price impact.

  • It shouldn’t be a surprise to see Dai prices at $1.01 or $1.02. There is often an impact on price from secondary markets since, very often, it’s a circular relationship. In this particular case, there is a rapid use case for Dai on secondary lending markets, a fact that Cyrus touched on. The impact COMP farming has on a particular underlying asset is elevating supply rate and depressing borrow rate; because both sides are compensated by the underlying yield token (which in this case is COMP.)

  • We’ve seen this in a different context. When you have a low borrow rate and a high supply rate, it incentivizes rapid refinancing and secondary market re-wrapping and basically pursuing a large degree of leverage very quickly. In the past, we have seen that the yield for those rates doesn’t necessarily last very long.

  • That’s the scenario you want to try to consider and manage against. For the peg, that price could potentially continue to go up. That’s something to be wary of.

  • Total Dai at $146 Million.

  • Dai from ETH at $114 Million.

  • Dai from BAT has shrunk.

  • ETH-A supply, you see short term ups and downs, while the last 48 hours has been flat.

  • There are other more interesting things to be doing with your capital at the moment, rather than putting it into Maker and generating Dai.

  • WBTC continues to stay level since it’s maxed out.

  • USDC-B not in use right now.

  • KNC is a new thing, so small amounts.

  • BAT, we saw on a large impact in March, and then it continued to rise through May. Basically, in the last couple of weeks, the remaining amount of BAT borrowed has shrunk, probably as a result of yield farming on Maker assets.

  • Zooming in a little on just the last three months; Once there was a more attractive yield for BAT, it was refinanced away. The amount of Dai minted from BAT went down.

  • Looking at USDC, which has been sporadic since May. And now, in the last 24 hours, it maxes out utilization very quickly. This isn’t a surprise.

  • Just looking at Compound markets, BAT became the asset for yield farming. With better opportunities, the supply in Maker shrank. The amount of Dai minted from ETH has slowed, and the collateral amounts are smaller.

    • On USDC, we saw a large amount supplied and a great amount borrowed. Pretty comparable numbers, in nominal value, to Compound.

    • More on USDC; We saw a large amount supplied and borrowed on Compound, and correspondingly, a large amount of Dai minted from USDC on Maker. We can expect this pattern to continue, given the attractiveness of the rates on Maker. There is a lot of risk math. Given a low collateral requirement and that it’s relatively safe and stable trade, that makes it a potentially very attractive option. That’s an important conversation for what kind of debt ceiling you want to manage on USDC.

  • A decent amount of Dai has been both supplied and borrowed on Compound too. It’s important to keep an eye on it. We’re talking about behavior, peg, supply, demand, etc. What is the underlying risk that’s being created from this behavior? You have 61 million Dai out of 140 million as supply on Compound. That’s a large chunk. And the larger that chunk, the larger part of the conversation of what is the total risk of the system, for Maker. That share could potentially grow larger, so it’s undeniably an important part of the conversation.

  • For ETH, we didn’t see a large movement in the amount of Dai minting, but there has been a movement in the amount of collateral. Since it’s attractive to supply ETH to Compound, that changes the mental calculus for certain users around how much ETH to leave sitting on Maker. Though they aren’t making large changes on the amount of Dai that they have minted from that ETH, I think there have been some shifts in individual users in the amount of ETH sitting as excess collateral. We’ve seen the 200-250% and 250-300% collateralization buckets grow. 300-350% has shrunk as well as 500+%. I think this has gone from being a trimodal graph to being a bimodal or primarily dominated by this 200-350% collateral range. I think it’s very clear why.

  • If we look at some of the largest USDC positions on Maker, the biggest one is at roughly 16 million at around 120% collateralization ratio. This has been the conversation with USDC since it was added, there was never a discussion around price volatility risk. But there are risks when the collateral type becomes more and more prominent month by month. It’s important to note that it’s a non-negligible share of the total amount of Dai that exists. What are the risks you assume as it grows?

  • In the last seven days, BAT was predominant in the loans originated on Compound.

  • In the last 24 hours, there’s a creeping share of that Dai and USDC interest. WBTC is a different conversation.

Discussion 3

54:28

  • Cyrus Younessi: I wanted to highlight that it is definitely not a foregone conclusion that Dai will be the highest utilized asset for COMP farming. We want to focus on the oversupply of ETH and USDC. So right now, ETH is 242 million versus three million. And USDC is 227 million versus 34 million. So the gross borrow for ETH is up 500%, and USDC is up 21%. If those numbers continue to increase, all that ETH supply that’s sitting there starts to become borrowed by its owners, reducing the spread. That would be beneficial for us. It’s also the rational course of events over the long run. This goes back to Lev’s forum post from yesterday. If those large ETH suppliers don’t borrow their own ETH, the impact might come back to Dai.

    • Vishesh: Two important stats there. You can see the excess supply for ETH. It’s huge. The attractiveness of the rates is the name of the game. Wherever is the most attractive yield is where the capital moves. In this case, how the attractiveness of the yield is determined, I think it depends on the share that the borrowing markets have on the market and the share of the total interest in Compound. There could be a whole conversation around the attractiveness of the different assets on Compound. If those rates aren’t as attractive as borrowing your own ETH on Compound, you could see the scenario where that interest shifts to a different asset. That was exemplified with BAT as the attractiveness shifts, so does the interest. So it could potentially be a moving target.

Ending Discussion

LongForWisdom

Discussion 4

  • We are passed the hour now, so the Q&A session is officially on if anyone has questions about anything.
Dai Interest Model on Compound

01:04:50

  • Akash: It seems from the comments that Dai is right now the most attractive because of the differences between the borrow rate and the deposit rate. There is currently a Dai interest model vote going on on Compound that is pretty contentious. It’s the first time I’ve seen not 100% go one way. Does anybody have any opinions on: first, the vote going on on Compound and how it would affect Dai? Second, if the interest rate difference on Compound was larger than USDC, between the borrow rate and the deposit rate, wouldn’t that solve our problem of why Dai is going to Compound? Because of the recycling effect? I’m hoping everyone understands my questions.

    • Lev: I am not actually convinced that post-patch the spreads between the displayed supply and borrow interest rates are actually the dominant factor, and it is because post-patch the interest rate per se doesn’t go into the COMP yield calculation anymore as it does now. So the interest rate is an additional cost you have to pay on your COMP farm, but the COMP yields themselves are huge. Unless you are really paying a lot of interest, and generally, when you are recycling, you’re making back most of what you’re paying on the borrow on your supply. So keep in mind that, just for numbers that I’ve looked at, you end up getting something like 80% APY COMP yield, obviously at current prices, etc., while paying 5% APY on the actual spread between the supply and borrow. Essentially, the supply and borrow ends up being less important, and what dominates the COMP yield is the relative share of lending and borrowing on that market, which can be a much bigger difference between different assets. I just wanted to point that out for you to keep in mind.
Are Large Yield Farmers selling their COMP

01:07:21

  • Akash: Right. Another question that might be unrelated; I don’t know if anybody has done any research. All of the guys that are recycling on Compound, are they completing the arbitrage by selling the COMP token, or is this a way for them to accumulate COMP tokens? Because that would help some of the understanding of what is driving this. Does anybody know that?

    • Cyrus: I don’t think that a lot of people are speculating too much on the price of COMP, to be honest. I think that it just adds an additional variable to the overall business model. Sure some are because they have opinions, but I don’t think it’s a core part of the entire operation. Of course, the entire concept of COMP farming has blown up because of the sustained high prices, that’s definitely true.
The Source of the COMP Subsidy

01:08:35

  • Chris_p: When do you think the music stops? You know, it’s going to happen eventually. The only question that I have is, “where exactly does this subsidy come from?”

    • Cyrus Younessi: It comes from retail speculators putting up bids on the secondary exchanges. The amount of collateral that is flowing to Compound is a result of the dollar amount of COMP you can get out of it based on.
The COMP Arbitrage

01:09:23

  • Akash: I think we have discussed the arbitrage very cleanly, right? You deposit funds on Compound to earn COMP tokens to sell COMP tokens to complete the arbitrage.

    • Cyrus: It’s an arbitrage, but it’s like a rebate. A centralized liquidity rebate and the rebates are very high.

    • Akash: Yes, I guess a true arbitrage would be if you could borrow USD in the real world for let’s say 10%, and then essentially, you would get a 20% return in the DeFi world.

Nuclear Option QE for Dai

01:10:15

  • Lev: I just thought I’d mention an idea for a nuclear option if the situation gets really out of control. In principle, MakerDAO can do a type of open market operation where it mints a large amount of Dai, puts it into Compound as supply, and then takes the cDAI essentially as collateral onto its balance sheet. By doing this, you could essentially flood Compound with an arbitrarily large amount of Dai; it could be like one trillion Dai or something. [laughs] Essentially, have all this cDAI on your balance sheet instead. That would essentially wash out everyone’s COMP yield.

    • So firstly, it would make the supply side of the Dai market so big that no one is going to be making any money on the supply side. At that point, I think that could really tip the scales in favor of literally any other asset; and it will also mean that Maker would be collecting the COMP for the Dai part so there would be no way of making money involving Dai on Compound anymore. It would all just become collected by MakerDAO. It’s just worth thinking about that possibility, and it’s relatively safe because we are taking the cDAI onto our balance sheet as collateral, so we are collateralized by the other Compound assets.
Lowering the Collateralization Ratio

01:11:35

  • Cyrus Younessi: You can achieve something similar without the social implications. I mean, if the liquidation ratio goes low enough to 101%. I think that at 100X leverage, that would be incentivizing the market makers to do pretty much the same thing, right?

    • Lev: Yes. I just like the component where we can take the cDAI on our balance sheet to avoid taking additional risk. Obviously, if you flood the Dai supply too much and it actually leaves Compound, then you have too much supply. But I think the difference between what I suggested and what you suggested is that we can actually take the cDAI as collateral, like 1 to 1 basically. It is literally QE for Dai, right?
Interest Rate of QE

01:12:31

  • Akash: And if you were to do is, what is the interest rate for the cDAI collateral?

    • Lev: This isn’t a CDP at all; this is an open market operation, where MakerDAO itself mints unbacked Dai, well it’s backed by the cDAI that we get in return. So MakerDAO itself acts as an agent to put a large amount of Dai into Compound. And the reason it’s not completely whack, it’s because we are actually getting cDAI in return. Just like when you do QE, it’s not completely whack because you are getting bonds return. So we can unwind this later by giving the cDAI back to Compound in exchange for Dai.

    • LongForWisdom: If we did that we would get a huge amount of COMP, because the open borrow on the Dai market would be like a trillion, right? So we essentially take all the COMP then, which is not a great look.

    • Lev: Yes, that is why I said, “nuclear option.” This sort of situation looks adversarial or drastic. I don’t want this idea myself, but I’m just saying that it is an escape. And before anyone makes fun of this, it doesn’t need to be a trillion Dai, that would just throw things off too much, but it could be a sizeable amount that we thought would have the necessary effect.

Ideal Proposal for COMP Governance

01:14:05

  • Akash: Let me ask the reverse question: What would you propose in an ideal world, what does COMP Governance do to mitigate risk for the MakerDAO system?

    • Lev: Post-patch, I think we need to figure out whether or not it is actually the case that Dai is getting sucked up, and if so, what parameter makes that trade attractive, what is so special about Dai. Because some have suggested that it’s the interest rate curve, which I think is valid because Dai is this freak asset on Compound that has this absurdly low interest-rate curve. All the other assets have this kind of linear curve, and the Dai one is essentially flat and also pegged to the ETH stability fee. But I think that might not be the main reason initially; it might actually have a little bit more to do with the relative size of the supply.

    • Akash: I think they might have done that to help the MakerDAO system when the peg was off, and you could easily borrow Dai from there at lower rates to hopefully maintain the peg, right?

    • Lev Livnev: I didn’t mean to speculate about the intentions there. I’m just saying that it has had the effect that it can be abnormally cheap to borrow Dai on Compound. Just to give a succinct answer to your question, I think that if it’s the rates, then we need to change the curve. If it’s a question of relative supply because the COMP yield dominates, then COMP supply scheme, I would recommend be patched again in order to not put this absurd pressure on the smallest market, which is what it’s doing.

    • Akash: I feel that we are going to figure out a lot more tomorrow.

Interest Rate Curve Calculation

01:16:54

  • Sam MacPherson: Is the interest rate curve for Dai on Compound set via the ETH-A stability fee?

    • Lev: It’s a somewhat complicated formula. You can see what it is at any point in time, on the front end, you can see this curve and just roll on it. But if you want to actually work out how it depends on the stability fee, you need to look at the code. I looked at it this morning. I couldn’t describe it in one sentence what it is in some sense pegged to the stability fee.

    • Sam MacPherson: So if we increase the ETH-A stability fee, we can increase the interest rate on Compound?

    • Lev Livnev: Exactly, we would be increasing both the supply and the borrow rates, though.

ETH-B Option

01:17:35

  • Sam MacPherson: Right, I don’t know how feasible this is but we could set up an ETH-B, move people over to that one, and the time frame is probably not reasonable but just as another sort of crazy option, and then up the ETH-A stability fee to whatever we want to, to make it unattractive.

    • Lev Livnev: Yes, I’ve also thought about that. I think that is also one of these backdoor hacks that might work, especially in the extreme. It’s not even inherently obvious that if you just manage to jack up Compound supply borrow versus using the stability fee as a backdoor, then you would actually make the situation better because you might also just incentivize a bunch of supply into it; Because imagine that we have this ETH-B which is very cheap and we trick Compound into making their supply and borrow rate for Dai really high then we might just end up incentivizing more of it getting sucked in. I mean, they also have the right to upgrade their interest rate model again so that they don’t have this backdoor as well. So it’s a little bit adversarial obviously.
Signal Request for USDC Debt Ceiling and Conditional Stability Fee

01:19:15

  • Cyrus: Just to confirm, after the call today, I’m going to put a signal request for the USDC debt ceiling and a conditional stability fee.

  • LongForWisdom: Yes, that was my understanding.

WBTC Stability Fee

01:19:42

  • Akash: When does the WBTC debt ceiling increase? When does it go into effect?

    • Cyrus: I guess, along with the executive vote.

    • LongForWisdom: The GSM delay is currently set to 12 hours. The executive goes out on Friday at the usual time, some time passes until it gets enough MKR, and then it’s 12 hours until GSM.

Abbreviated Terms

MCD: The Multi-Collateral Dai system

CR: Collateralization Ratio

DC: Debt Ceiling

SF: Stability Fee

DSR: Dai Savings Rate

MIP: Maker Improvement Proposal

GSM: Governance Security Module

Credits

  • Tim Black produced this summary.

  • David Utrobin produced this summary.

  • Gala Guillen produced this summary.

  • Juan Guillen produced this summary.

  • Everyone who spoke and presented on the call (listed in the headers.)

Bump for visibility.