Governance and Risk Meeting: Ep. 37 (May 30 - 2019)¶
Video | Audio | Discussion
| References | Person | Text |
|---|---|---|
| # / 00:00:00 | Richard Brown | Is that red light turned on for anyone else? Oh, there it goes. All right everyone, welcome to the May 30th edition of the Scientific Governance and Risk Meeting, with MakerDAO. My name is Richard Brown. I am going to moderate the call today and talk a bit about governance and the recent votes. After that, we will talk to Cyrus Younessi who represents the risk team at MakerDAO and he's going to in turn introduce us to a bunch of really talented members of the community who are also helping us out with risk. And then after that, we're going to hear from Cyrus who will start digging into collateral types and collateral risk, which is an interesting conversation we need to have as an entire community over the course of the next weeks, months, leading up to multi-collateral DAI. |
| # / 00:00:52 | Richard Brown | And then afterwards if there is significant interest, we'll have a Q&A session that could possibly run past the top of the hour. We'll stick around until energy begins to wane. Generally we've been hanging out for a half hour after these calls and just doing a general Q&A and there's been a lot of interesting conversation. So if you're interested, please hang out for that. |
| # / 00:01:14 | Richard Brown | My preamble I am going to jump into again is that we are intensely intensely interested in getting as much feedback and questions, comments and concerns from the community as we possibly can. We have an enormously talented group of people in these calls and we want to know what they're thinking and get as much from them as we possibly can. Sorry [inaudible 00:01:38] for a second there, I'm digging up a link. So I'm going to past this in the chat on the side. If people have questions or comments, please type them out in the chat and we will attempt to get to those and when there's a quiet moment, we'll read them out for you. |
| # / 00:01:50 | Richard Brown | If you have access to a microphone please jump in and just start talking. There's no rules against interruptions here. If you think there's a thread that we picked up or dropped that needs to be explored more completely, please jump into the Reddit thread that I just posted, and let's continue the conversation there because as always, this is a forum for us to discuss and to debate and to explore things and to surface new issues and to discuss memes and/or educate ourselves and the rest of the community, but we don't really make decisions here. We make decisions through the voting portal, and we have deeper debates in Reddit over the course of the week. And so please make sure that your thoughts and concerns are known and recorded in those threads. |
| # / 00:02:42 | Richard Brown | Generally we have Steven come in and give us the thought of the week, but one of the main themes of these calls is that we're iterating every single time trying to improve things, trying to tighten things up a bit. And one of the things we're trying to do is to... one of the ideas we're wrestling with are how much information we pack into this one call, this single call, how many calls do we need to have in the future, what should the agenda of those calls be, who should we advertise them to. There's a lot of questions we need to answer. Do we need a separate risk call, do we need three or four different separate risk calls, do we need one governance call for monetary policy, do we need another governance call for things like the DSR, the stability fee. There's all kinds of questions that need to be addressed. But in the short term, it's my preference that we keep this one primary call as optimized as we possibly can, and as focused as we can. |
| # / 00:03:38 | Richard Brown | And as it becomes absolutely required we can start fragmenting the discussion a bit. But until then, we're going to tighten things up a bit, and that means that Steven's thought of the week is going to be migrated into more of a role where Steven is a stakeholder like everybody else. And so as we discuss interesting topics or there is clarity that he thinks there needs to be addressed, he can jump in as a participant in the calls as opposed to sort of the opener and the closer of the calls, which we've been doing in the past. |
| # / 00:04:12 | Richard Brown | In light of that, the desire for efficiency I'm going to try and tone down my speechifying because I've gone back and listened to a few calls and I've talked for 20 minutes without realizing it. I have a feeling that I'm getting there today as well, so I'm going to tighten things up. |
| # / 00:04:27 | Richard Brown | My discussion today though is around governance and voting, and that's going to be sort of my gig, as been in the past and will continue for the near future. I want to talk about how the vote has gone in the last two weeks because we've seen some interesting behaviors that have surfaced. Some issues, concerns that we knew could happen in the voting system and we've addressed it in a way that we anticipated that we would have to. And that concern is that we have a continuous approval voting system where the idea is that maker holders signal one picture of the system or a new picture of the system. And as they have a chance to view how the ecosystem is behaving and the market is behaving, the peg is behaving, they decide whether they want the new executive to be ratified or whether they're happy with the old executive. |
| # / 00:05:20 | Richard Brown | In the past, we got... our past about two months, we got used to this idea of a... we got wrapped up in this thing which I call sometimes "the new normal", where because something had happened a few times in a row, we assumed that that's the way that the world is going to be from this point forward. And that's a risky thing to fall into. And that new normal assumption was that every Friday, Maker holders would immediately jump into the voting portal, they would pick whichever the new vote was because new is obviously better than old. Then we would proceed to enjoy the success of our governance system and one day we'd have a poll that was based on that very clear picture of the system that we all understood, which was clarify the previous [inaudible 00:06:05]. |
| # / 00:06:06 | Richard Brown | The last two weeks some interesting things happened. A bunch of interesting things happened. The peg largely stabilized, air quotes and caveats apply. I'm not really qualified to figure out whether it has or not, but it seems like it has. A lot of the voters seemed to have shared that same opinion and there wasn't a mad rush to raise and or lower the fee on the Friday two weeks ago. And there wasn't one on the weekend either, and there wasn't one during the course of the week. And we ran into this situation where we began having polls that didn't reflect the current state of the system. They reflected what potentially the future state of the system might look like, irrespective of what the main voters were doing with the actual stability fee itself. |
| # / 00:06:51 | Richard Brown | So we had the stability fee at 19. There was a significant amount of signaling that wanted it to go to 17, so we saw a disconnect between signalers and the polling infrastructure and actual voters... or people staking their weight against the actual executive. That was something that we need to learn about. It also surfaced some issues that we have with the system in general, which is that our requirements are evolving faster than the governance portal is evolving. And one of those things we've run into, which obviously we all knew we would, is that you can only vote for one thing at a time, and that includes a poll. So if you have your MKR staked on a poll that says you want to lower the fee, you need to take it out of a current executive that is lowering the fee or vice versa. And if you want to vote for a lower fee, you need to take it out of the poll and that changes the signal. |
| # / 00:07:49 | Richard Brown | So we have all kinds of issues we're addressing there and we've spoken about in previous calls. Chris from product was in the call last week to give us some... solicit some information and show us how the portal will be evolving, but anyways. We ran into some issues about weak executives and confusing polls. Happily that ambiguity was cleared up this week and there's some very interesting things that happened here that I want to explore a bit, or at least put on people's radars is that the significant amount of data that shows that the peg is probably stable-ish right now. I'll leave Vishesh and the risk team to explain why because I don't understand why it's stable-ish. |
| # / 00:08:33 | Richard Brown | But the interesting thing is that instead of choosing to leave the fee where it was under the understanding that potentially it's just working the way it's supposed to, Maker voters chose to lower the fee a couple days ago. And that to me is very interesting for a few reasons. It sends a signal, beyond whether it should have been lowered or shouldn't have been, I don't know. But the great thing about that change is it signaled to the community that our fees are not marching ever upwards, that the voters are more than willing to lower those fees when they feel that it's required. That takes the steam or the wind out of a lot of people's sails in the contrarian MakerDAO community, which I enjoyed seeing. It also goes towards dispelling the myth that there's a correlation between the stability fee and the price of Maker, which is sort of a pernicious myth, and one that particularly annoys me because it's so immediately and obviously disprovable because if you look at the stability fees and you look at the price of Maker and you overlay the two graphs together, you can see that raising of stability fees is generally deleterious to the price of Maker. |
| # / 00:09:44 | Richard Brown | But this also provides some very interesting signals to the community, that the fee does go down, that it's not exclusively the domain of a bunch of money hungry plutarchs, and that ultimately... and this is the most important one, that there is a peg recovery in the future. And we've proven that that is a situation that can occur. And one of the interesting things we might see in the future now is that the next time that cheap DAI does show up on the markets or there's a sufficient supply overshoot, people now have empirical evidence that it will go back to the dollar. So it might be time to enjoy some arbing opportunities. |
| # / 00:10:32 | Richard Brown | So those are some of the things that I wanted to leave people with. Governance has survived some hiccups and some unforeseen, largely unforeseen situations. It's recovered. We have a new governance portal coming very soon that will solve some of these issues. But in the meantime, I just wanted to thank all the voters who did vote and the people that are sitting on the fence, please come out to vote, because it's critically important that even if you don't think that your Maker is going to make a difference, trust me. It does make a difference. And I need to convince myself of that [inaudible 00:11:07] because I'm not a whale either, but I do vote and I hope that everybody else gets out there and does it too. |
| # / 00:11:13 | Richard Brown | I think that's where I'm going to leave it I think. Let's continue to think about some of the edge cases with governance and just think about how that signaling works in the community. And at that... yeah, I'll leave it there. So Cyrus, I'm going to hand that off to you. |
| # / 00:11:27 | Cyrus Younessi | Yeah, just I also think it's interesting to reflect back on all the social media attacks that we were withstanding for about a month, right? Where there was a lot of absolutism with oh, this is never going to work, there's no chance it's going to work. And then when it finally did get fixed, just kind of everything went silent. |
| # / 00:11:52 | Richard Brown | Yeah, it's tempting to... I think that secretly we can pat ourselves on the back a bit, but there is a certain amount of grim satisfaction to watch these alarmist contrarians suddenly just move onto greener pastures. |
| # / 00:12:06 | Cyrus Younessi | Well I'm not saying it just to be petty but more along the lines of there will be times where governance decisions may appear unpopular or there may be other incentives at play for people not wanting policy to go a certain direction. And we just have to be aware of that and trust in the scientific governance process. |
| # / 00:12:29 | Richard Brown | Yeah, actually that's great. And I wish that I'd brought that point up because it's so good. And it's a risk that we ran into in the past where sentiment, just crypto sentiment, I began to be concerned that that was actually moving voters' behavior. And I think that they've largely been resolute in what should and should not be done. So yay, those are great points Cyrus. That's one of our takeaways here, that sure, the subreddit may explode with trolls and people that are synthetically enraged for some reason. But as long as we look at the numbers and really listen to the analysts, I think we have a good data point right now that everything is going to work out as long as we don't get scared. All right, thanks for that, Cyrus. |
| # / 00:13:15 | Cyrus Younessi | Okay, cool. Onto the DAI price. Vishesh do you want to take it away? |
| # / 00:13:22 | Vishesh Choudhry | Yes. Okay. Can you see my screen? |
| # / 00:13:38 | Cyrus Younessi | No, I can't. Looks like I'm going to have to manually pin it from... |
| # / 00:13:43 | Vishesh Choudhry | Yeah. |
| # / 00:13:44 | Cyrus Younessi | So for everyone else on the call, there is a user in the user list called Presentation, Vishesh Choudhry. And you'll have to manually pin it to see the presentation. |
| # / 00:14:03 | Vishesh Choudhry | I guess that's going to be a thing from now on. All right so to just start off here, I'll get her in a second, but to just start off, so this is a visualization we've used for the past couple weeks for the DAI price. It's been interesting to see this semi-normal distribution. It's usually centered around one dollar, or at least it has been for the past I would say couple weeks. The last four or five days it's been pushed up slightly actually, which is very contrary to what we would expect. And I think this kind of speaks to what Rich was alluding to which is DAI price is actually fairly well-supported compared to what we would have expected with what's been going on with ETH price. And we can touch on sort of why we think that's happening. |
| # / 00:14:56 | Vishesh Choudhry | But the volume has been relatively healthy, hovering around 2, 3 million. The price, like I said just slightly above a dollar. There were a few dips but by and large, this is where it's at. |
| # / 00:15:11 | Vishesh Choudhry | The majority of the volume has been on Oasis. There's a slightly higher bump for DDEX right now. That's just a blip. And then Uniswap has actually been holding fairly steady. It's a pretty big place for DAI to be traded as well. So to just touch on what's been going on with the price. So like I said... |
| # / 00:15:56 | David Utrobin | Did Vishesh time out? |
| # / 00:16:00 | Cyrus Younessi | Maybe, I thought it was me who timed out. |
| # / 00:16:02 | David Utrobin | Yeah, I thought it was you too. |
| # / 00:16:03 | Richard Brown | [crosstalk 00:16:03] Yeah, I was waiting to see somebody moving in the sidebar to see whether it was me. |
| # / 00:16:06 | Vishesh Choudhry | In- |
| # / 00:16:07 | David Utrobin | I thought everybody was frozen. |
| # / 00:16:08 | Richard Brown | Vishesh, [crosstalk 00:16:09] we lost you for like a minute. |
| # / 00:16:11 | Vishesh Choudhry | Oh, sorry about that. Yeah so where I just kind of left off was the DAI price as mentioned had been coming up fairly steadily and there was that significant event a couple weeks ago. And since then, things have actually been pretty quiet, which is interesting because things have been all but quiet with ETH price, I feel. So this decision to drop the stability fee was very recent, and usually we don't see the effect of the stability fee change on price and supply until a little bit afterwards. So we do still have to wait and see to make any conclusions from that. But initial signs are actually much quieter than expected. So that's good news I feel. |
| # / 00:17:01 | Vishesh Choudhry | This is a little bit noisy, so I'll just take off a couple of these lines. So the volume weighted average, the simple moving volume weighted average has been kind of consistently rising over time as DAI has now kind of proved that it can punch up above the peg. Similarly, ETH price has been making, basically tracking those exact movements. So what's really interesting is a lot of their relationship that we'd seen in the past between ETH and DAI is actually fairly reversed over the past few weeks. And that's something else that I think we should discuss. |
| # / 00:17:47 | Vishesh Choudhry | In terms of what's been going on with draws, wipes, and supply, [inaudible 00:17:54]. So, supply is actually not significantly changed in the past couple of weeks. So like I said, a lot of quiet. It's come down consistently since mid-April but since the 19.5 percent stability fee increase, there was that little bit of a run-up that we talked about and since then, it's been fairly steady around 81 million. |
| # / 00:18:20 | Vishesh Choudhry | There's been a lot of movement in DAI, but the supply has not changed significantly, so that's something that I want to touch on in a bit. Because just scroll back up for a second, the volume, the sort of daily volume has been significantly higher in the past two, three weeks than it had been for the past two, three months. But the supply has held relatively steady. And I think a lot of the activity that's been on DAI recently has actually been on a lot of these secondary markets. So I think that's a topic for us to explore further, and start to talk about a bit more. |
| # / 00:18:59 | Vishesh Choudhry | The age of debt as mentioned prior has come down significantly in the past couple of weeks. This speaks to an increased circulation in loans. So as we said, more loans being paid back and some newer loans being taken out. But as you can see from the supply, that kind of effect has balanced out between the two. So that's good. |
| # / 00:19:25 | Richard Brown | Vishesh, could we talk about this for a second. Is that healthy, unhealthy, or ambivalent, that behavior of like [crosstalk 00:19:33]- |
| # / 00:19:33 | Vishesh Choudhry | Yeah, I would call it healthy. I think you need... and I've kind of championed this idea a few times, I think you need that circulation of debt to a) it's just good for setting expectations in terms of fees that have been realized vs. not, which is always just a risk to keep in mind. And then b) I think as those kind of costs are realized and effected in the short term, I think that's healthier in the sense that if there's a 19.5 percent stability fee and people are kind of ignoring it and putting it out of mind because they're long term leverage and they're not worrying about it until two, three, four, five months later, you can potentially get into some dangerous economic scenarios. |
| # / 00:20:25 | Matthew Rabinowitz | Well one question on this, do we have... you can go back up, but do we still have an overlay of what the stability fee actually was at that point around May, whatever that looks like, the 10th? When it started to really contract? |
| # / 00:20:40 | Vishesh Choudhry | Yeah [crosstalk 00:20:40]. |
| # / 00:20:40 | Matthew Rabinowitz | Was that [crosstalk 00:20:40], 19? |
| # / 00:20:41 | Vishesh Choudhry | Yeah. Yeah, that black line is the- |
| # / 00:20:42 | Matthew Rabinowitz | Okay, that was the number. Thank you. |
| # / 00:20:45 | Vishesh Choudhry | Yeah. So, there is the potential sign of this starting to slow and level out now, which I think is to be expected. So I don't want to make too many conclusions off of tiny data windows. But what I would start to expect and I think we might see, is that this age will kind of level out into roughly an equilibrium point. And so I think there will be this moving equilibrium point for a given supply level, right? So for 81 million DAI, we would expect over time there shouldn't be significant movements in the average age of DAI. There should be some sort of rough equilibrium for how often people draw out loans and pay loans back. |
| # / 00:21:33 | Vishesh Choudhry | I think what we've seen is a lot of... I'll just expand the scale, a lot of choppy movements in the age of debt and a lot of increases and sudden drops. That to me is what I would consider unhealthy behavior, because that's kind of the system constantly in flux. And a lot of major state changes. [crosstalk 00:21:56] Yeah? |
| # / 00:21:56 | Richard Brown | This is purely market sentiment, those peaks and valleys, right? |
| # / 00:21:59 | Vishesh Choudhry | Right [crosstalk 00:21:59]. |
| # / 00:21:59 | Richard Brown | [crosstalk 00:21:59] the stability fee was appropriate? |
| # / 00:22:01 | Vishesh Choudhry | Yeah, and I think that's a good point to consider and probably more of a decision point and less of a data point is what's the right relationship between price movements and the state of MakerDAO? Because in the past I think a lot of the state has been kind of bullied by price movements. But what's happening recently, which is odd but potentially healthy, is a little bit more immunity to some of those price movements in terms of what these equilibrium points are for DAI supply, for DAI price, for average age of debt. I think we're just seeing more stability all around, which is a good sign. |
| # / 00:22:39 | Vishesh Choudhry | So I'm going to skip over the Compound section here for a second because I'll talk about that in a different context. But the percentage of new vs. old DAI, I mean we can talk about this but I think just the amount of DAI being drawn out has been relatively low lately. And this line has kind of dropped down, but to a fairly stable level. So effectively, there's kind of a more stable equilibrium point again of what I've been talking about between the amount of loans that are taken out on preexisting CDPs, vs. new CDPs being created and new loans being taken out on those. |
| # / 00:23:27 | Vishesh Choudhry | So collateral is another really interesting topic, especially with regards to a lot of the ETH price movements, because there have been a few large sort of collateral removal events that have coincided with a few large increases in ETH price, which is kind of an interesting relationship if you think about it, because what we would generally expect is this collateralization ratio to hold relatively steady. In the past, it has held relatively steady. But with the May crypto spring, whatever the proper term is, that collateralization ratio shot up and has actually kind of wavered around this 530 percent line, despite a few large events in terms of ETH being taken out. |
| # / 00:24:22 | Vishesh Choudhry | So, as ETH price appreciates, people decide, "Hey, I don't need to keep as much collateral in the system, I feel fairly safe, so I can back out of some of that capital cost." As opposed to what had happened in the past a bit, which is as ETH price had run up, people had kind of kept that buffer, and then taken that as a signal to draw more DAI out of the system. So what I'm just saying is that's a slightly different response to the same stimuli. So that suggests that the nature of the system has changed a bit. And so what I think this actually speaks to a lot more is the actual effect of the stability fee, and some of the economics that have been discovered, I guess I would say, with a lot of these secondary lending markets. I mean, you've heard probably anecdotally that dYdX has experienced significant volume. And there's been a lot more refinancing to some of these secondary platforms. And I think people are starting to see them as more of a viable place to get DAI, and I think as the demand for DAI increases, those are basically an efficiency function. |
| # / 00:25:40 | Vishesh Choudhry | So let's say the demand for DAI is roughly 70 million, and there's sort of this 10 million buffer of oversupply, if that can get eaten up through a lot of lending on secondary markets, then what you could actually see is a more close tracking between DAI supply and DAI price. So if the demand for DAI significantly increased, you would have less of this buffer living on Maker and I think a lot more on the secondary lending platforms, and so you could see a more immediate effect from changes in demand. So to me, that's just kind of the market maturing a little bit, I would say. And so I do want to touch on these secondary lending platforms in a second. |
| # / 00:26:29 | Vishesh Choudhry | So, loan [crosstalk 00:26:30]- |
| # / 00:26:30 | Cyrus Younessi | I'm sorry, hold on one sec I have a question with that. So, sorry did you say that the high collateralization ratio is a sign of the market maturing? I mean, I don't expect people to just perpetually or continuously have five to one [crosstalk 00:26:49]- |
| # / 00:26:49 | Vishesh Choudhry | No. |
| # / 00:26:50 | Cyrus Younessi | Ratio locked in voluntarily. |
| # / 00:26:51 | Vishesh Choudhry | Yeah, not the high collateralization ratio, but the increased activity of people moving a lot of these loans to secondary markets when the rates are cheaper. Because that's sensible behavior, that's efficiency. And so it doesn't make sense to me why you would draw DAI from Maker when the rate is higher. And so there's always been kind of this buffer between the Maker rate and the rate on these secondary lending platforms. What we're just starting to see now is basically more utilization of that rate. And so to me, that's just kind of increased efficiency realizing in the market, and so I would call that maturity. |
| # / 00:27:34 | Cyrus Younessi | So what are some rationalizations for the high collateralization ratio? And do we know what it was last year on the way down in the bear market? Does your chart go back? |
| # / 00:27:48 | Vishesh Choudhry | It does. So what time range are you particularly curious about? |
| # / 00:27:53 | Cyrus Younessi | The second half of last year. |
| # / 00:27:55 | Vishesh Choudhry | So what we saw was the collateralization ratio, as I said, was a lot lower. And I would say more volatile. |
| # / 00:28:08 | Cyrus Younessi | So to me it says that people are... because of the bear market they're just perpetually scared of being liquidated, especially because of the penalty associated and just general volatility that even at four, five hundred percent collateralization, they would prefer to keep it in. |
| # / 00:28:28 | Vishesh Choudhry | Currently yes. So what I was saying was in the past there was kind of a) more risk-seeking behavior, and then b) I think this was just kind of a once spurned, twice shy type of situation, where there were... and I had shown this in the past but I took it off of here so the site would load faster, a fair amount of liquidations. And so, when ETH price volatility is significant and then there is a large quantity of liquidations, naturally you would see a rise in collateralization ratio because people sort of smarten up or get scared of getting liquidated. |
| # / 00:29:09 | Cyrus Younessi | Right. |
| # / 00:29:09 | Vishesh Choudhry | What's interesting is more recently, this kind of relationship between the stability fee increases, the DAI supply coming down a bit, and the collateralization ratio increasing, and ETH price running up only in the last month or so, right? So this kind of rise has predated the most recent bull run in ETH prices. Now, what I'm saying is in part, there is a bit of a risk aversion in the sense that as ETH price comes up, people imagine the potential energy is higher, there's a greater height to fall from, and so they do want to keep to some extent that buffer. But what we'd kind of been scared of I think, or what had been a potential concern in the past couple weeks was with this significant buffer of collateralization ratio, and a large quantity of DAI that we know was purchased in exchange for ETH, that people were then going to take that DAI, wait a little bit, and then dump it back into ETH. |
| # / 00:30:14 | Vishesh Choudhry | What's interesting is, and it may be a function of the fact that ETH price has been increasing and the other shoe has not yet dropped, that's entirely possible, or it may be a function of these loan refinancings. It's very hard to say. But we have not seen that large quantity of DAI getting flooded into the market and DAI price dropping, which was definitely a concern. And the other piece is that this sort of buffer in collateral has been maintained, but like I said, as ETH price has come up, instead of choosing to draw out more DAI, people have chosen to take out some of the ETH to balance out the capital cost. |
| # / 00:30:53 | Cyrus Younessi | So another interesting aspect is that the lending market for ETH |
| # / 00:31:00 | Cyrus Younessi | ETH is not nearly as active or profitable as it is for DAI. So all this surplus ETH that CDP owners have locked into the system, it's not like they can take it out and earn that tentative 15% that you could get for DAI on other platforms. |
| # / 00:31:20 | Vishesh Choudhry | Well- |
| # / 00:31:20 | Cyrus Younessi | You could borrow DAI against your extra ETH. |
| # / 00:31:24 | Vishesh Choudhry | Yeah. |
| # / 00:31:25 | Cyrus Younessi | So why aren't we seeing that? And I guess we should talk, after this we should talk about this DAI supply as well, you referenced it. |
| # / 00:31:36 | Vishesh Choudhry | But I think that's exactly what I was going to say, was yes, you're not going to get a huge yield for ETH, but you could totally dump it into another platform, use it as collateral, and draw out USDC or DAI or something else and then try to construct another trade off that to get some yield. So A, why aren't people doing that? Great question. I think to some extent, there's a lot of uncertainty in what that potential yield might be, especially when people have to go through multiple steps of exchanging currency A to currency B to currency C, and then B, I think to some extent, people want to maintain, and it's kind of hard to explain demand mechanics, especially when we don't have a lot of spotlight on the demand sources at the moment, but like why would you maintain 80 million DAI in Maker at a 17.5 now percent interest rate when you could maintain roughly the same thing in Compound for 14 or 15%? That's an important question for us to consider, is why is there even the DAI supply that currently exists? What is it about Maker that is the differentiator that people prefer DAI from Maker versus DAI from another source? |
| # / 00:33:02 | Vishesh Choudhry | And I think whatever that reason is, which I can't just simply answer for you here, will kind of be the upstream answer to why people are maintaining such a significant collateralization percentage on Maker, because they're not ready to let go of their CDP positions for some reason. |
| # / 00:33:23 | Vishesh Choudhry | So just wanted, and I don't want to eat up too much time, but I do want to touch on some of the secondary lending stats. So I played around with LoanScan's new resources, their API, their updated UI, which are all great resources. And what's been interesting to see is leading in terms of new loans being drawn out is roughly a balance between Maker and Compound, so it seems that Compound v2 is one source where, and they don't have dYdX in here, which is another platform that's been on the rise in the past week or two, but what we've seen is roughly the same amount of DAI being pulled out of Maker versus Compound, so it seems that some of the new DAI that would normally have been drawn out in this particular type of market scenario is being drawn out from other platforms and not Maker, which contributes to part of the reason why some of that supply has held steady despite this ETH price rally. |
| # / 00:34:28 | Richard Brown | Bishesh, at what point does that become a concern for us, that Compound is a more effective source of DAI than we are? |
| # / 00:34:37 | Vishesh Choudhry | I think what I was trying to say earlier is that it should always be a concern in the sense that we need to constantly wrestle with the question of what is Maker's relationship in the ecosystem? |
| # / 00:34:53 | Matthew Rabinowitz | Talk through that for a second. Unless I'm totally out of whack on this one, the only way DAI can be minted is via Maker. So it is created and destroyed via Maker. The fact that it's cheaper in a secondary lending platform, at least from my perspective, implies the Maker rate is overpriced for where it should be today, and we see it on a synthetic number that we're trying to get to the peg, we're at the peg, we're at 17 and a half on one hand, and there are secondary markets that are cheaper. |
| # / 00:35:25 | Matthew Rabinowitz | So that delta, that $10 million delta, the theoretical $10 million gap, it's a question of when, not if, but when people decide to refinance their CDPs with more cost cheaper loans, per se, but it still can only be created and destroyed with one system. It's a natural arbitrage. We will get to that harmony of 70 million if that's actually the demand number, and once we do- |
| # / 00:35:53 | Matthew Rabinowitz | Yeah, and once we do, the Maker rates will be on par with the secondary lending platform. It's kind of the corollary of having algorithmic trades where they in effect put each other out of business because you have everybody trying to fight to capture a ten-thousandth of a penny and then it turns into one-one-hundred-thousandth of a penny. It's the same model, except this is so slow, at least to my eyes, I could be totally wrong, but it's a slow system where if there's a $10 million gap, it's just going to take time for people to refinance those numbers down. |
| # / 00:36:31 | Richard Brown | It's also alternately a question of supply, right? DAI only comes out of a CDP and the secondary lending, the DAI available to be dropped into Compound will dry up at some point and the only place to get more DAI would be a CDP rent. So we're going to hit that lower bound where the premium on pulling it out of MakerDAO is going to be unavoidable. |
| # / 00:36:54 | Vishesh Choudhry | Yeah, so that's actually a point I wanted to touch on. I agree with what Matthew's saying in terms of the relationship is always going to be that kind of generator point. And the secondary lending, what I was saying earlier, do create this efficiency buffer in the sense that if DAI is overpriced on Maker, there's that arbitrage opportunity on the secondary lending platforms. And so it helps people obtain DAI at the rate that they really want to obtain it, but there is definitely a structural question of is it really a bad thing for that natural arbitrage to exist? |
| # / 00:37:33 | Vishesh Choudhry | And I would ... I'm not going to answer one way or the other, but I'm going to suggest the potential that it is more stable to have those secondary lending platforms at that discount to Maker rather than trying to chase, which has been suggested before, the rate that those secondary lending platforms offer as a new point for the stability fee, because a stability fee of 14% with, versus a stability fee of 17% with a 13, 14% lending rate on secondary platforms are not the same thing. And in the event that there is a large amount of DAI that's dumped onto the market or flipping back to the previous market economics where DAI price suffers and DAI supply increases with an increase in ETH price, in those scenarios, there's I think an additional stability benefit to having those separate nodes in the ecosystem in terms of MakerDAO being the generator of DAI versus the secondary lending platforms being this almost UI layer to consumers where they go and get that cheaper rate. |
| # / 00:38:53 | Vishesh Choudhry | I don't think those two scenarios are the same thing, and in the event that you just have Maker and you don't have these significant buffer volumes on these secondary lending platforms, you could run into the potential scenario where you're exposed to the price movements again, and a significant change in ETH price bullies the DAI supply on Maker again, and that could have serious ramifications for DAI price. |
| # / 00:39:19 | Matthew Rabinowitz | Yeah, I look at it as being a very basic version of the DAI savings rate that's coming out. It ultimately destroys the DAI supply because it allows people to refinance their outstanding DAI. It makes the system more efficient, but it's not in the purview or the control of the Maker system. |
| # / 00:39:38 | Vishesh Choudhry | Yep, I totally agree with that. So there's a question of, and I agree, I think it's a really good point to point out that it's a makeshift MacGyver version of the DSR, which is I think a very interesting take. But yeah, there's a question of whether or not that's a bad thing. I don't think it necessarily is. I think it, in the absence of a savings rate, does lend an additional stability layer to Maker, and I think it is, from a biological standpoint, I think it's cropped up for a reason. A lot of these platforms have popped up, they've gained significant traction because there is a real place for them in the market and there is a need for it, especially in the absence of the stability fee ... Savings rate. Sorry. |
| # / 00:40:26 | Vishesh Choudhry | Sorry, and I don't want to eat up too much time, so I just want to punch through the secondary lending stats real quick. So what's interesting to also note is repayments, so this speaks to a lot of what I've been talking about. There are more repayments ... So if there are similar amounts of draws from Maker versus Compound and there are more repayments on Maker versus Compound, then that tells you what's going in is coming out the other end. And there is this measurable refinancing from Maker to Compound at the very least and this is not updated to include dYdX, but when we look at dYdX, I think you'll see a similar effect there that's really been gaining a lot of movement in the past week or so. |
| # / 00:41:15 | Vishesh Choudhry | So sorry, yeah, I think that's everything that I wanted to run through. I know that was a ton of time. Happy to answer any followup questions if there are. |
| # / 00:41:23 | Richard Brown | I think that we have Matthew coming in for his thought of the week, but there was a significant number of questions for you, Vishesh, maybe we should tackle those at the top of the hour? And Cyrus, how do you actually want to play out the last 15 minutes we have left? Do you want to do the risk, or the collateral stuff after 10? |
| # / 00:41:45 | Cyrus Younessi | Let's do Matthew's thoughts and then answer some questions and then we'll just do the collateral stuff at the top of the hour, or afterwards I guess. I don't know. I hate to break flow and go back and forth. |
| # / 00:42:01 | Richard Brown | I'll leave it up to you. What do you want to do? |
| # / 00:42:04 | Cyrus Younessi | Yeah, that's fine. We can go ahead with ... We can continue on the same. |
| # / 00:42:09 | Richard Brown | All right, Matthew, you want to give us your thoughts of the week? |
| # / 00:42:12 | Matthew Rabinowitz | Yeah, so basically this is just a continuation from the ... Or actually from last weekend, it relates back to how we can start putting together a machine learning and PID pieces of this, so starting to assemble a community-driven movement or a team about what we can do to put together a working group to optimize this. And in many ways, it's very similar, and I wrote about this, it's whenever Tesla put together their concept of an autopilot or driving a car within the lane, and that lane, for the time being, let's just say is a targeted, stabilized peg, and the question is what are the market forces that cause it? Whether or not it's a stability fee that causes or destroys supply, and the same thing with the DSR. Both of them have corresponding effects and how that can be treated to the corollary of wind and how that, whatever pieces cause a correction to cause an algorithm to keep the car more or less in the lane. |
| # / 00:43:13 | Matthew Rabinowitz | And we have all sorts of I guess data points, and there's no shortage of the data, the question really is what is the correlation on the data and how we can pull those in on both the DAI price, the ETH price, anything that could be corollary that could cause wind, per se, and that can include Bitcoin, and this notion of how many inputs you put in there allow us to abstract away not necessarily predict, 'cause I don't want to get in the mode of forecasting. It's more of historic associations that ... |
| # / 00:43:50 | Matthew Rabinowitz | Let's just, as an example, if we can find a way, and this is hypothetical, but if we can get the data from Fiat on ramps that are pulling data, excuse me, pulling fiat into an exchange, let's say Coinbase or send Wyre, just of the gross dollar amounts, it's kind of the corollary of saying, "Is there a storm building?" Doesn't mean anything's going to happen, but then we can start to see if there's wind and what kind of impact that would have on keeping the car in the lane. All of those things, if we optimize them, and that stuff takes time, so this is a full-on shout out to any AWS engineer that understands AWS inside and out that does machine learning, please contact us, put it in Reddit, and let's start the discussion of how we can implement that. That's it. |
| # / 00:44:46 | Richard Brown | Okay, cool, thanks, Matthew. And let's continue that discussion in the link that I've posted in the side, too and the discussion thread for the call, because this is part of a larger initiative that we're exploring and that's ... Well, Cyrus, I'm not going to speak for you. Well, I just said I'm not going to, but I'm going to. We're intensely interested in the [inaudible 00:45:08] talent from the community. It's been pre-filtered and pre-selected and pre-sorted and we have the talent here, so if you have ideas about how to contribute or if you have data models or you understand machine learning, then ping us and let us know what you can do. |
| # / 00:45:28 | Richard Brown | All right, Cyrus, do you want to ... How do you want to take the rest of the call? So we have the risk stuff? |
| # / 00:45:34 | Cyrus Younessi | Yeah, let's just spend a couple minutes on any important questions that we had from the presentation. |
| # / 00:45:42 | Richard Brown | Yeah, Vishesh, if you want to, take a look at the side. |
| # / 00:45:46 | Cyrus Younessi | I think just maybe the main takeaway for me was the static DAI supply that's been stuck at 80 million for quite some time now, despite the huge ETH rally, and so I think it warrants some revisiting of our earlier hypothesis from a few months ago about the negative correlation between DAI price and ETH price and potentially looking at some alternative indicator, such as ETH price sentiment, which is more of a relative metric as opposed to absolute number. And I think that some of this stuff that Vishesh and Primoz have been working on in the background, is trying to gauge the sentiment of the current price, or essentially what the demand for leverage is. |
| # / 00:46:41 | Joe Quintillian | What the ... Sorry to interrupt, David, just one thing I want to put out there is that on the OTC angle, which we don't have a lot of stuff on here, is it's getting a little harder to source big blocks of DAI. And this is coming from some of the major OTC desks, where you want to buy two million, three million DAI, it's getting a little more difficult. So I thought that was something you guys should know, just 'cause the stability fee is so high, maybe you have to get more people to, DAI prices go higher for people who issue more DAI at these prices, but thought the community should know. |
| # / 00:47:23 | Cyrus Younessi | Yeah, I feel like there was ... There's a healthy level of increased speculative demand as the ETH price, or as all the crypto markets go up, right? It's almost concerning to see no new DAI generation, given the recent rally. So I feel like they're- |
| # / 00:47:43 | Vishesh Choudhry | Sorry, go ahead. |
| # / 00:47:44 | Cyrus Younessi | Go ahead. No, go ahead. |
| # / 00:47:46 | Vishesh Choudhry | I'm not 100% sure that that's concerning. So what's I think is a good data point, it's helpful to understand what's going on in the OTC side. I think it lines up with what I've been observing on Uniswap, for example, where, and this is a bit anecdotal, I don't have the data on this just yet, but I think smaller trades than prior, affect more changes in the exchange rate on Uniswap, so I think there's less liquidity and that, I could go look at the liquidity pool and check on, but what my sense is is that liquidity for DAI at current demand levels is less than before because demand levels are actually higher for the use case of leverage. |
| # / 00:48:34 | Vishesh Choudhry | And that's something that we could go and confirm and actually measure, but I think what's going on is there's an increase in demand but not enough to where demand dries up liquidity to an extend that DAI price goes up enough to effectively cause people to draw out more DAI, because people are able to currently source it from these other areas that we're talking about. And so if you think about that, there's this margin or this buffer that exists between where the supply is and where the demand is. And I think what's been going on lately is that the demand has been able to rise and that margin has been able to narrow with the same supply levels, whereas in the past, there was a larger margin between those two and so changes in demand levels effected greater changes in supply. |
| # / 00:49:27 | Vishesh Choudhry | And so that's my sense of what's going on at the moment, so if this continues, and I think that's not necessarily bad, is my point. Because what that means is you can have a theoretically more efficient supply that tracks to demand a bit better. Now as demand, if demand continues to increase, then you would see an increase in supply if that hypothesis is correct. So I would just say I don't think it's necessarily a bad thing. |
| # / 00:50:00 | Cyrus Younessi | Okay. Cool. |
| # / 00:50:02 | Matthew Rabinowitz | Especially in a world where the secondary lending rates are lower than the stability fee. If we get to be the other way around, then it is concerning. If we have, for whatever reason, the secondary rates would be higher, which I can't imagine that, but imagine they were, and we had an underlying price rally in the collateral and we didn't see a DAI outstanding increase, then yeah, there would be an issue. |
| # / 00:50:24 | Vishesh Choudhry | Yeah, that would be a very strange economic situation. And I think we really want to be cognizant of that buffer, not for the purposes of setting the stability fee, but of just understanding the dynamics of what's going on, being aware of how thin that margin is between secondary lending rates and the stability fee. Because I think it probably and it's very hard for me to say from a scientific standpoint, but it probably tracks fairly close between the buffer between demand and supply. |
| # / 00:50:59 | Cyrus Younessi | Okay, cool. Let's move on for the last five minutes so I can give a little spiel on collateral risk. So as you guys all know, collateral risk will at least slowly start to filter its way into these discussions for a variety of reasons, primarily obviously because we are in the ... we're on the cusp of Multi Collateral DAI and there is a lot to unpack there, and monetary policy, at least for the very current time, is doing okay. |
| # / 00:51:43 | Cyrus Younessi | So I'm just going to give a brief introduction of how I'm thinking about it or how we're all thinking about it internally, and I want to emphasize that just like the monetary policy discussions, we're hoping that these primarily take place offline, Reddit, social media, and so forth. We want to just try to seed ideas here, and in some respects, it's actually more important because the monetary policy is somewhat of a reactionary process where we are just looking at the DAI price and adjusting the stability fee one way or another. |
| # / 00:52:26 | Cyrus Younessi | But with the collateral risk, there is definitely some more forward-thinking that's required. We definitely want to jump ahead of some of the risks involved. It's not as pleasant to be reactionary to some of these, the collateral moves. |
| # / 00:52:45 | Cyrus Younessi | So to start, there's going to be a lot of groundwork that we're going to have to lay here on this call and basically over the coming weeks. There's a lot of operational aspects to discuss. We're going to have to start exploring questions such as what kind of risks specifically are we looking for and how do we quantity them? How do we blend fundamental analysis and the quantitative modeling and all manners of how to build up the collateral portfolio and maintain it? |
| # / 00:53:18 | Cyrus Younessi | Additionally there's going to be a lot of governance issues at play. The risk teams, their incentives, what is the governance process by which we add collateral? As a primer, I would suggest everybody rereads Steven's three risk documents, they were phenomenal, and they really set the tone for the scientific approach that we have been emphasizing for the past year. A lot of our risk discussions will be framed in that context. We're also working on some followup documentation as well. |
| # / 00:53:58 | Cyrus Younessi | And yeah, I think in general, we're going to try to keep this as an open discussion, maybe have a light structure in place. I'm going to pause here for questions, if anybody has anything to comment on right now. Okay, cool. So basically I think the overarching picture here is how do we get from inputs to outputs? How do we take a collateral and then figure out what its risk parameters are, debt-sealing stability, all that stuff? And there's issues of factoring in the qualitative aspects of the collateral, essentially what is it, what does it represent, as well as its trading profile, it's market presence, its liquidity. |
| # / 00:54:51 | Cyrus Younessi | A lot of moving pieces and a lot of analysis to conduct. But I think there's very elegant answers to almost all the questions, and it's just a matter of iterating and building it out piece by piece. And yeah, that's it. I don't really have a ton of information today. I think I'm just trying to set the tone for how the ensuing collateral discussions are going to take place. |
| # / 00:55:32 | Matthew Rabinowitz | I have a question. Is it planned for the DAI savings rate to be uniform across ... I know you mentioned the stability fee will be different depending on the collateral type, if I understand it correctly. Is the savings rate planning to be uniform across all assets, all collateral types? |
| # / 00:55:50 | Cyrus Younessi | Yes. It is in the time being going to be uniform. |
| # / 00:55:55 | Matthew Rabinowitz | Okay. |
| # / 00:55:58 | Vishesh Choudhry | I also have a question. So you're talking about evaluation of individual assets, but what type of work are we planning around looking at intercorrelations between these assets and looking at the basket of assets that exists on Maker as a portfolio analysis or combines liquidity risk? What's the sense of how we want to start thinking about assets in a network and not just individual buckets? |
| # / 00:56:31 | Cyrus Younessi | Right, so I think we're going to have to start releasing some high-level overviews and outlines of an end-to-end life cycle of these risk models, and then just approach it step by step. I think it makes sense to actually just redo an analysis on ETH from the ground up, maybe optimize some of the current parameters that we have in place. I think for example, the collateralization ratio in debt ceilings and so forth were all somewhat defensive, just to protect the system in the early days, but there's definitely room for optimization there. |
| # / 00:57:14 | Cyrus Younessi | Yeah, and then I think the proper approach is to just look at a single asset approach and then factor in, add in an additional asset and then look at their interplay, their dynamics, and continue from there. |
| # / 00:57:31 | Josh Wisniewski | So what about assets like the difference between bringing in something like TUSD versus a Uniswap liquidity pool? Like one's trustless and we can see all the contracts, one is totally opaque, unless you start getting lawyers involved and looking at bank accounts and stuff like that. |
| # / 00:57:54 | Cyrus Younessi | Right. There's going to be discussions and methodologies in place for basically all manners of risk, including one of the primary differentiators between a lot of Crypto assets is whether or not they're bare assets or not. Obviously ETH is extremely trustless and extremely easy to facilitate through the keeper options and what not, and there's going to be some extra steps involved for security tokens or regulated, registered assets. |
| # / 00:58:33 | Josh Wisniewski | Are we talking about legal entities being set up for some of those more real-world assets? [crosstalk 00:58:42] |
| # / 00:58:42 | Cyrus Younessi | So security tokens is probably something that we should discuss a little bit further down the line, just given its added complexity and given a lot of the legal uncertainty. So in terms of implement operationally how it's done, it's actually a bit of an open question. There's obviously a lot of security token issuers out there and I believe they all have their own unique way of handling these situations, so we're just going to have to pour through it carefully and understand it and take it from there. One instructive example for evaluating collateral types is along those lines is looking at something like Bitcoin or Ether versus something like Tether, very different risk profiles, so for example, if you were to look at Tether purely based on its trading history, its past profile, you'd actually think it's probably pretty safe and obviously we know that to not be the case and looking at something like Bitcoin or Ether seems to be extremely volatile but in a lot of ways much safer than something like Tether. |
| # / 01:00:18 | Cyrus Younessi | And so it will be interesting to unpack these different situations. Personally it's something I'm pretty excited about. This is my favorite part of Crypto. But it can be a bit complex at times, for sure. |
| # / 01:00:35 | Vishesh Choudhry | So quick question, you're talking about a fundamental analysis of different assets, but how do you think about not like, an asset as a single static monolith, but also as a moving sum of pieces? So in particular I'm thinking about order books, trades, liquidity, like the value or the price of an asset is one thing, and then what's realized on the market through actual trades and through real life, real time liquidity on the order books is a very different thing often. I know for ETH, it's a very big asset they may track, but especially for a lot of longer tail assets, it's going to become very tenuous to consider what's the long-term stability of that asset versus what's the long-term liquidity of that asset? |
| # / 01:01:29 | Cyrus Younessi | Right, so what we're ... In particular what we're concerned there is our expected liquidation value, which is where if we have to, if MakerDAO has to dump the asset on the open market, how much value could it realize relative to spot? Basically some estimate of the slippage involved, and yeah, if you have that long tail of very illiquid assets where you try to sell $10,000 of XYZ coin and |
| # / 01:02:00 | Cyrus Younessi | ... you only end up with $2,000, then essentially what that builds into is the collateralization ratio, which is actually primarily a function of how much slippage you will incur during the liquidation. Which is exactly why I think, in certain cases, the ETH parameters can be optimized. We've had hundreds of liquidations and they never once incurred bad debt. So, for the given stability fee, you could argue that the collateralization ratio was way too restrictive. |
| # / 01:02:47 | Vishesh Choudhry | That's a good point. So, this kind of was the reason that I originally brought up some of the asset interactions because liquidity of an asset is not specific to that asset. It's specific to pairs. And so, DAI ETH liquidity may be fairly strong, but DAI USD liquidity may be lower. And so there's kind of that question of how do you then start to think about what are the pathways for exit for a given asset. And do you just do a blended weighted average? |
| # / 01:03:22 | Cyrus Younessi | Obviously, all of our options are against the DAI pair. But, I think for practical purposes, you can assume that arbitrage situation, you can use the USD markets as a proxy for liquidity. Whereas, you probably wouldn't use the bitcoin-based pairs as a proxy. |
| # / 01:03:41 | Cyrus Younessi | Although, if you really wanted to get deep enough into the analysis, you could. But, I just suspect that for most assets, the USD pair should suffice. If you were going to factor in the Bitcoin one, you'd also have to factor in the Bitcoin DAI pair. Because that's essentially the path it would get from start to finish. |
| # / 01:04:10 | Vishesh Choudhry | [crosstalk 01:04:10]. Yeah, it becomes a little bit of a transition matrix of just what are the intermediate steps. They all just kind of contribute to the ultimate path. |
| # / 01:04:20 | Cyrus Younessi | Right. |
| # / 01:04:30 | Primoz Kordez | Cyrus? |
| # / 01:04:31 | Cyrus Younessi | Yeah? |
| # / 01:04:32 | Primoz Kordez | Do you think we could use some TCRs or should we develop our own one? How do you see? |
| # / 01:04:37 | Cyrus Younessi | TCRs? |
| # / 01:04:38 | Primoz Kordez | Token Curated Registries. |
| # / 01:04:41 | Cyrus Younessi | We are a TCR. |
| # / 01:04:46 | Cyrus Younessi | MakerDAO through its governance and its assessment of the risk parameters actually is a live TCR. Yeah, MCR. Let's actually talk about that for a bit. As we evaluate from a fundamental perspective these collateral assets, it's actually very similar to what the ratings agency such as Moody's or S&P does. And we're actually filling that dual role of being the lender as well as being the evaluator of the asset. That's because there really is no official Moody's for crypto right now. And so, who better than the ones taking on the risk anyways? You can envision kind of in a distant future when crypto markets are more mature, and there are more collateral assets than MakerDAO or the Maker risk teams can reasonably evaluate. We can outsource that to professional services. Or actually, those professional services essentially are the future vision of risk teams. |
| # / 01:06:12 | Josh Wisniewski | How would those professional services be compensated under a system with MKR? [crosstalk 01:06:21]. |
| # / 01:06:20 | Cyrus Younessi | Right, so in the same sense that you would, in the traditional world, pay these ratings services for their analysis. MakerDAO likely will have the ability to compensate analysts for their services as well. So, it becomes a bit tricky with do you just want an objective evaluation of a collateral asset, or do you also want a risk evaluation. And incentives definitely come into play here. Because you want everybody to be on the same page, for sure. But yeah, that's something we're going to have to see how it evolves as time goes on. |
| # / 01:07:13 | Vishesh Choudhry | That kind of begs the next natural question, which is how are those analyzes and how is that information presented to the community? How does the information disseminate? Especially with regards to presumably a very long-term project around building these collateral asset models. To what extent is the community kept looped in and how is that information shared? |
| # / 01:07:38 | Cyrus Younessi | The way I'm looking at it is kind of just some sort of step by step evaluation process. Typically just starts with the classification of the asset. Basically, what is it. Is it a bare asset or security token? Is it a work token, governance token, utility token? |
| # / 01:08:01 | Cyrus Younessi | There's some pretty decent established theory around various token models and the value that they objectively have. First step is just kind of pegging it down on what kind of asset it is. |
| # / 01:08:17 | Cyrus Younessi | I think the next step is creating some sort of due diligence model. Fundamental evaluation. Maybe even create a numerical evaluation model to give MKR holders some notion of what this asset is actually worth. Just think back to the height of the bull market in 2017. Would you accept the collateral asset that's market price is 3 billion dollars, when according to evaluation theory, it should pretty much be worth zero. No risk parameter rule protecting against situations like that. Yeah, first few steps are definitely just gain a very deep understanding of what the asset is, what it's worth. And then you can actually start digging into what are the risks associated with it. Technical risks, legal risks, operational risk from the team, contract risk, bugs, back doors. All sorts of things. And these kind of edge-case risks will basically build into the risk premium for that collateral asset. Just think about if someone took a loan from you and they put up Tether as collateral. You'd probably charge them a pretty high interest rate given the risks associated with it. That's just purely qualitative risks versus giving a loan out on much safer collateral. |
| # / 01:10:05 | Cyrus Younessi | So, essentially, a lot of these fundamental evaluations, this whole TCR ratings agency aspect will primarily factor into the collateral specific risk premium. |
| # / 01:10:23 | Vishesh Choudhry | Maybe more of a thought, less of a question. If we consider the different types of price movements that lead to different types of risks in the system, for example, if an asset is hacked, there's a sudden immediate black swan event. That's a different risk. If an asset, we know is something happens to the dev team, etcetera, it's fundamental value drops and the price drops slowly and consistently over time. That's a different type of risk. Versus if it's fairly stable over time, but it's very volatile. That's again, yet another different type of risk. So, how are we thinking about sort of building a taxonomy of these different types of risks and communicating that and organizing that information? |
| # / 01:11:12 | Cyrus Younessi | That's actually the meat of it. That's basically the crux of the entire analysis. It boils down to trying to find out some estimation of the probability of a CDP being liquidated. Because that's really the building block for everything, is CDP analysis. And thinking about CDPs as loans, regardless of what it's collateralized by. Ether, some other coin, or even a hypothetical undercollateralized CDP that has a reputation or credit score attached to it. And essentially, everything comes down to trying to model out the probability that this CDP is liquidated. And then a conditional loss distribution given that the CDP has been liquidated. Then that will kind of factor in that tail risk that you were just mentioning. |
| # / 01:12:25 | Vishesh Choudhry | So, what we're effectively saying is how likely the CDP is to be liquidated, plus how severely that CDPs liquidation? |
| # / 01:12:31 | Cyrus Younessi | Exactly. Those are the two most crucial aspects of the entire collateral [inaudible 01:12:38] model. |
| # / 01:12:38 | Vishesh Choudhry | In that sense, I would suggest maybe also we think about the entire pool. So, it's not just for an individual CDP, what's the likelihood of getting liquidated. And how severely it's liquidated. But then also, within that distribution, for all the CDPs in the existing pool, how many of them are liquidated in what way? And so, kind of then thinking about one big CDP that is liquidated to varying degrees. |
| # / 01:13:09 | Matthew Rabinowitz | And one of the metrics you need to include in all of that is any type of correlation between the two. What's the probability that if you end up having some type of liquidation in token structure A, that that would have some secondary, tertiary effect on others. |
| # / 01:13:24 | Vishesh Choudhry | Yeah, and that's where a lot of the order books and the trades and liquidity come in. Because I was thinking about if asset A, 50% of the CDPs are liquidated, and there's a huge sale in asset A, denominated in DAI or in asset B. And then asset B is also liquidated to like a 30% extent, how much does that change [crosstalk 01:13:45] that you're able to get. |
| # / 01:13:48 | Cyrus Younessi | Definitely. This has all been specified. There's a ton of precedent for this. Just to cut to the chase, basically no great way to do it, at least in the space of crypto markets because, well crypto's new and there isn't a ton of historical information. And in fact, there isn't even a ton of current information. And by that, I mean any kind of market-implied metrics for some of these risk parameters that we're interested in. |
| # / 01:14:25 | Cyrus Younessi | Typically, if you were looking at a portfolio of traditional assets, there are kind of several different ways of doing it. One is, you could just look at historical data. If you're looking at loans, you could look at historical default frequencies. If you're looking at collateral assets, look at the price history, and so forth. Or, you could even separately approach it from kind of an options framework and look at market-implied CDS spreads, market-implied volatilities and build out models that way. |
| # / 01:15:05 | Cyrus Younessi | The bad news for crypto is that we don't have any of that stuff. But the good news is that all of crypto correlated anyways. So, for as long as we're just adding kind of the standard crypto ERC20 assets, we can actually extract a lot of that out and just assume extremely high or even perfect correlation. Anyone who's ever looked at Coinmarketcap can tell you that. So then, really, at least for the short run, the model converges back down to, okay, what is the probability that any of these CDPs are liquidated, whether it be ETH or any other crypto asset. And then also trying to figure out how bad the liquidations will be. |
| # / 01:15:48 | Matthew Rabinowitz | I suppose the general request, I think, we're over an hour into this call and we haven't even gone over any one of those collateral types because they don't need to because they don't exist yet. But probably when we actually start adding them, just in general, I'd recommend that we try and bifurcate this call into having general governance and then collateral calls to go over those. Or some other structure. I don't know how we're going to officially [crosstalk 01:16:13]. |
| # / 01:16:12 | Cyrus Younessi | Yeah, we've talked about that at length, and it's something we are still working out, the structure of this call. Especially now that it tends to run minimum an hour and a half. Yeah, that's a good point because there's definitely a lot to discuss. |
| # / 01:16:29 | Akiva | Yeah, I have a question. Someone posted on Reddit asking if it's true that whether or not the bullishness or the bearishness or the directionality of the collateral token has any impact on stability fees. The example would be, let's say there's a token that we think is going to have very low volatility and very high liquidity, but at the same time, we're very bearish on it. Does that give it a higher stability fee, or does that keep it the same, we think we'll be able to liquidate it properly? |
| # / 01:17:01 | Cyrus Younessi | Yeah. Somewhat counterintuitively, it really would have a negligible effect on the stability fee. And this is what I was trying to allude to in my preamble about specifically what kinds of risks we are looking for. It's actually not just small market movements or anything ... I'm not great at explaining it, but essentially we want to look at tail risks and highly unexpected events. If you are bearish on the token, that's something you can plan for. And if you can plan for it, there are ways to handle that. |
| # / 01:17:52 | Cyrus Younessi | Sure, you could increase the stability fee. You could also increase the collateralization ratio. But, really just standard market volatility or downward bias isn't necessarily what we care about. Although, it's correlated with what we care about, though, for sure. |
| # / 01:18:10 | Vishesh Choudhry | I don't think you'd have to do much of anything at all. This was an analysis that I initially ran on ETH when first getting introduced to Maker was, okay, what happens in the even that ETH price just consistently deflates? And let's say it doesn't drop 30% overnight, but it drops 30% over the course of a year. |
| # / 01:18:33 | Cyrus Younessi | As we saw last year. |
| # / 01:18:33 | Vishesh Choudhry | It's actually kind of good for Maker, or at least there's an argument to be made that it's good, in the sense that when there's a lot of long-term slow price depreciation, you actually have a healthy balance of liquidations. And liquidations are not bad for the system. It's undercollateralized liquidations that are bad for the system. And so, when liquidations occur, they actually just keep that debt circulation up. They keep realized revenues up. And they're not necessarily a bad thing. |
| # / 01:19:03 | Vishesh Choudhry | What's the concern is exactly what, Cyrus, you're talking about. When there's sudden events. But, especially with ETH price, we saw that as the price kind of deflated over time, it was actually fairly positive for revenues and fairly neutral for pretty much every other metric. |
| # / 01:19:20 | Cyrus Younessi | Right. Exactly. |
| # / 01:19:29 | Akiva | All right. Thanks, guys. |
| # / 01:19:36 | Cyrus Younessi | That kind of definitely opens up the conversation about why it's so difficult in the first place. In that you can envision a scenario where ETH does drop by 30% overnight. Protocol hack, you could come up with some pretty grim scenarios of why ETH price would crater. But, it's actually pretty much impossible to estimate with those if that probability actually is. Is it 10% or 20% or 5%. |
| # / 01:20:16 | Cyrus Younessi | Essentially what all these numbers come back to is, once you have some estimate of your expected worst case scenarios, your expected losses, you want to kind of take a look of that overall picture and see, okay, is this an exposure that we are comfortable with as a DAO. So, for example, if we have billions in Ether versus billions in a bunch of really risky loans, how do we define how much exposure MakerDAO can actually hold up to, which is another interesting question. [crosstalk 01:21:05]. |
| # / 01:21:05 | Josh Wisniewski | Is there any thoughts about how you scale up governance to manage including collateral types? Because I feel like even with the adjusting stability fees, there's lots of meetings, there's lots of drama on Reddit, all this other stuff. And the risk team has limited bandwidth. How do we add, not just the first five collateral types, but 100. |
| # / 01:21:33 | Cyrus Younessi | Right. Well, there are some ideas and plans in place. Just one idea, and this is just purely brainstorming, is you could have, for example, just one executive vote or one governance vote, say every month or every two months. And that one vote would add in, 20 different collateral types plus all the risk parameters. And you would spend kind of the prior two months discussing each asset on the calls. On social media. And then you just put them all in at once. And then you would adjust them. |
| # / 01:22:16 | Cyrus Younessi | The hope is that the tinkering that needs to be done on the collateral-specific side is sufficiently low. That we can kind of match it up with the timeline of governance. Of course, there will be plenty of emergency situations where we need to adjust things on the fly. And that might take a little bit of work. But, in general, we can add collateral in batches. And hopefully, just kind of plug and play a lot of the models. |
| # / 01:22:51 | Richard Brown | Well, I think that's kind of the key here, though, is there's a model that applies to ERC20s in general. We don't need to reevaluate that model for each one that comes down the pipe, each new potential collateral type. We expand or augment the base model and iterate that every time and then it should be a fairly straightforward process of just determining what the caps are on each collateral type, right? |
| # / 01:23:21 | Cyrus Younessi | Yeah. Although what will be interesting is that the models are sufficiently open for peer review that we should see several entirely different architectures for the same collateral asset. |
| # / 01:23:40 | Richard Brown | It's an interesting question. And, Josh, thanks for asking it. The short answer is, we have no idea what the level of engagement is from a governance and risk perspective when we're introducing these new collateral types. It's entirely possible that, like Cyrus said, we can drop a package and everybody says, yeah that looks great, and we can just move ahead. I think that might be slightly optimistic. I have feeling that we're going to need to figure out where the sweet spot is between decoupling this large set of data into manageable, bite-size, votable pieces. |
| # / 01:24:12 | Richard Brown | And then moving them through the system. Which doesn't take into account the level of interaction with evaluating the models and debating the individual classes and having these meetings and all the rest of it. I think we can probably be prepared to have weeks and/or months of discussions leading up to the first major collateral type vote. And then we'll have to figure out how well that went and then iterate again. |
| # / 01:24:48 | Richard Brown | Just to finish that thought is an example we've seen the level of coordination and effort and community interaction and voting portal interaction it took to move the stability fee number up and down. With that level of complexity, it's not hard to extrapolate what might happen when we start adding 10, 15, 20 different collateral types in the future. |
| # / 01:25:11 | Cyrus Younessi | Yeah. I actually think one of the more contentious aspects will be debt ceilings. Because debt ceiling's the ultimate metric for scale and growth. There will be this divide between short-term and long-term health. Where people want to see the system scale as fast as possible. And a lot of the risks associated are kind of latent or not immediately visible. People say, oh, well that's never going to happen, this is totally fine. I think that's just one of the human fallacies that we just see over and over in history. |
| # / 01:25:57 | Richard Brown | Yeah, we're also kind of blessed in the fact that we're ETH and DAI maximalists and Maker maximalists. And these discussions leading up to this point, [inaudible 01:26:09] name some of the more rabid projects, halfway I stop myself, but once shitcoin shills discover that we're considering their religiously devoted shitcoin is being considered for a collateral type, we're going to see an entirely new segment of people getting involved in this process that's going to muddy the waters a bit, I think. |
| # / 01:26:31 | Cyrus Younessi | Right. |
| # / 01:26:33 | Guilherme Remor | Cyrus? |
| # / 01:26:36 | Cyrus Younessi | Yeah, go ahead. |
| # / 01:26:38 | Guilherme Remor | I've got a couple of questions. So, initially, do we have any estimates in terms of many new collaterals we're talking about? |
| # / 01:26:53 | Cyrus Younessi | It's not up to us. This is definitely a governance MKR-wide decision. And they can ultimately choose to add in as many as they want. On our end, all that we can control on the foundation risk team is how many assets we can properly evaluate in the limited time span given the natural overhead that goes into some of these evaluation processes and quantitative modeling. If an asset will generate growth [inaudible 01:27:35] collateral portfolio, and we'll have to kind of define what it means to make sense, then anything can be added, in theory. And so, you could have multiple different risk teams tackling different assets, and then they all get submitted and voted on all at once by the executive vote. Akiva, I was just about to answer your question. Yeah, I do know about the SPV model. And kind of a brief tldr, essentially, certain security tokens are obviously quite illiquid, and they're even more illiquid on crypto markets. Imagine trying to find a liquid market for tokens that represent some obscure real-world asset. So, the SPV model goes like this. So, the SPV is created and there are ERC20 tokens minted that represent a claim on the liquidated value of the assets in the real world. |
| # / 01:28:56 | Cyrus Younessi | That was really confusing. If there's a liquidation occurs, the asset gets sold in the real world and then the SPV is legally obligated to convert the proceeds into DAI. And then the DAI would get sent to the ERC20 holders. The tokens represent a claim on that DAI. In the keeper auction, you would bid on $ 1 worth of DAI at hopefully a discount. And you would get the proceeds of the auction. I'm not sure if that entirely made sense. |
| # / 01:29:43 | Cyrus Younessi | The goal is to source liquidity off chain and then flow the proceeds back to token holders enchain. |
| # / 01:29:52 | Akiva | I think Greg said something in his speech where, something like that it actually doesn't represent the legal claim. [inaudible 01:29:59] clarifying that. |
| # / 01:30:00 | Cyrus Younessi | Yeah, it doesn't represent a legal claim on the assets. It represents a legal claim on the liquidated proceeds of the assets. |
| # / 01:30:11 | Matthew Rabinowitz | Think about basically that mechanism as being a hyper hyper-efficient version of a bankruptcy court. |
| # / 01:30:18 | Cyrus Younessi | Right, exactly. Yes, Josh, to your question. |
| # / 01:30:42 | Richard Brown | I have to jump in again because I have a significant amount of work I need to do, so I can't record this anymore. So, I need to stop the recording and move on to my next meetings. If anybody wants to hang out and continue the discussion, they're more than welcome to do so. But it won't be recorded for posterity. Once again, thanks everybody for the amazing discussion. There will be a recap summary videos and audio and transcripts will be posted to the subreddit and to Twitter. Thanks for coming, everyone. Thanks, Cyrus, thanks, Matthew, and thanks, Vishesh for your presentations. It was great. |