-
LongForWisdom: Okay, so we’re going to be moving into the more general Q&A, discussion section. Maybe Primoz, you want to start a discussion around the surplus buffer and stablecoin stuff?
-
Primoz: Yes, sure. So, I made this post today, and I will drop the link here in the side chat. The post concerns around potentially increasing the surplus buffer. We need to think about the day we’re going to be filling the two million surplus buffer and, from then, are we going to have flop
auctions? The issue I’m having is that I believe on a daily term we are accruing about 50.000 DAI of fees from debt on stablecoin collateral. The problem I’m seeing is that we know those fees won’t be collected entirely. We are aware of the issue we’re having with a low collateralization ratio and the outcome via self-unbinding or liquidations, and there are some uncertainties. How effectively can we collect those fees that the system accrued? And, from my standpoint, at least from a Risk’s perspective, it doesn’t seem reasonable to be spending money which we don’t know if we’re going to be able to fully collect. It is a risk to spend it on flop
auctions because if we are to do so, what we are effectively doing is that once the money is spent on flop
auctions, when we have this unwinding, we will realize that just a proportion of the accrued fees, the surplus buffer, can be drained. This is because 50 thousand per day is 1.5 million per month, so if we postpone this and then we do unwinding at some future date, potentially, if we unwind it with a lower collection of fees, we would be draining our surplus buffer. I want to avoid this because the surplus buffer is meant to protect against unexpected losses. However, in this case where we already know there might be some impediment, which is why I was suggesting we already increased the surplus buffer to avoid the situation. Long story short is that the more flop
auctions we’re going to have, unless we don’t increase the surplus buffer, the higher probability of flop
auctions will follow in the future because we would be just wearing the 2 million surplus auction. This is something that I want to avoid, and that’s why I was proposing to increase the surplus buffer.
-
Chris Mooney: You know, the alternative here is that we change the stability fee from stablecoin down to zero. That would be another way to keep the surplus buffer the same. There’s one nice benefit to having the surplus buffer kicking off flap
auctions, which is that you could think of MKR as the sort of buffer in case we end up in undercollateralized territory and we need to mint a bunch of MKR via flop
auctions to re-collateralize our system or to take care of that bad debt, which happened right after black Thursday. That ends up removing this kind of cushion in MKR - like the amount of buffer that we have of value accrued there. So one of the nice things about flap
auctions is that they sort of wind it back up again and provide a cushion if we need to mint MKR. The other nice thing about flap
auctions is that we’re above peg, and these auctions will release some of that debt onto the market. So, I think that flap
auctions are important. Whatever needle we throw right here, I agree that this is a moderately urgent issue that we should figure out. It would be nice that we could keep the flap
auctions in place so that they would have their own even small impact on the peg and rewind that capacity and the cushion up in MKR.
-
Primoz: Yeah, I agree on the positive effects of flap
auctions. In regards to shutting down or disabling stability fees, maybe a downside of that is that for all the new positions that are being potentially established, you wouldn’t be charging anything but, you know, the new positions are mostly the ones where you can collect some fees so you wouldn’t be maximizing your return if you disable those fees prematurely. Therefore, I’m not sure if that’s so wise. Otherwise, I agree. Also, I forgot to mention that the last time we increased the surplus buffer to two million was right after black Thursday, and the DAI supply was around 100 million. I’m not sure, but now that the DAI supply is much more significant. We know that half of the supply is on stablecoin, but the other half is still at least four times bigger, so just from a risk standpoint, we should have a larger surplus buffer. Ideally, you should always aim toward a percentage of the DAI supply. Only from a risk standpoint was I going to propose we increase it. This thing has some trade-offs, but it’s up to the community. From a risk standpoint, I need to mention this.
-
Christopher Mooney: You brought up a great point there. We can collect some fees on these vaults; we have to cut them off at some point. That makes me wonder what if, as we get closer, let’s say within a 10th of a percent, we’re at the one-to-one sort of ratio? Those vaults would be liquidated either way. What if we had some executive that swept them into a new vault type, and then some kept that vault type around for when we get offloaded either through liquidations or something else so we’d be able to collect the specific fees on these things. Therefore, when they’re all tapped out we sweep them off to some other vault?
-
Primoz: I’m not sure that I’m following if I’m honest.
-
Kurt: Imagine how different it would be from zeroing out the DC on the existing collateral type, and shutting off fees and then starting a new one.
-
Christopher Mooney: It’s effectively the same thing.
-
Kurt: How would you implement what you proposed because you’d never actually want to loop over all open vaults or something in an executive, right?
-
Christopher Mooney: Yeah, the only downside there with what I’m suggesting is that if you did end up looping over whichever ones were bad, you wouldn’t do that on-chain, but you could find them and effectively trigger some manual sweep. Facility A(FA) remains the collateral type that everyone uses. We can slide everything over onto another vault type that has liquidations frozen temporarily. We get to collect the maximum fees on everything. Otherwise, we have to shut down FA and then move over to some other type, and there would be some stuff in FA where we don’t collect the maximum fees.
-
LongForWisdom: You mean the USDCA, right?
-
Christopher Mooney: Yes, that’s what I meant.
-
LongForWisdom: Yes, just double-checking.
-
SamM: So, I don’t think I quite understand. What would be the point of pushing them over to this new Vault facility?
-
Christopher Mooney: You get to collect the maximum amount of fees until they’re at a one-to-one ratio, but you know you can’t. You know that you’re going to accrue bad debt if you liquidate them, so you move them over to some other vault, and they just become Maker reserves at that point until we’re below peg. Then we unleash auctions in them and create some DAI demand that brings things back up.
-
Sam M: So, would the owner still be able to close them?
-
Christopher Mooney: No, I think this would be we like the vault itself. It has gone out to liquidations. This would be the equivalent to saying you’ve lost your collateral because it’s under-collateralized and it would move over to this other vault type so…Anyways, this might be too complicated. As I said, I literally just thought about it.
-
brianmcmichael: And the result here is that the protocol ends up owning these stablecoins, so it seems kind of similar to the PSM idea in that we end up holding these stablecoin bags.
-
Kurt: Pretty much. So long as they never get to a point where it’s profitable to repay them, then the protocol will end up owning them one way or another. So I don’t think it’s technically feasible to try to do mass seizures and execs or something of that sort. You could change the way liquidations happen to these vaults and then give some incentive for keepers to do it for you, but I’m not sure it’s worth the hassle of trying to isolate the ones that are underwater versus the ones that aren’t. I propose in the future we introduce new collaterals with a fixed window for taking advantage of them so that they’re all on the same schedule and every three months or something, we introduce a new USDC collateral. I mean, give the people a week or two weeks to join it, and then basically they’re all on the same fee collection schedule, and you could shut them off as they get to that under-collateralized point.
-
Matthew Rabinowitz: But at that point you’re basically just gating. Let’s say you have a waterfall of USDC-A, B, C, all the way to F and, as A fills up, you shut off the debt ceiling, drop it down to zero and now you’re blocking the entrance if you will, which will force it into letter “B.” No?
-
Kurt: Yeah, exactly.
-
Gregory di Prisco: Can I chime in quickly? I’ve been thinking about this a lot for the last month, and I want to clarify a couple of points. There is nothing wrong with letting these vaults go under 100%. It doesn’t harm the protocol unless we start burning MKR with that DAI. I would advocate that instead of us spending our resources on something that’s not a problem, we monitor the surplus buffer like Primoz initially suggested. First of all, this isn’t even a problem until any vault in USDC-A is under 100%, which isn’t going to happen for at least 45-50 days at this SF, and even from there, you watch that debt and every week if you need to increase the surplus buffer to make sure that that debt doesn’t burn MKR, then that’s all you have to do. I don’t think that this is a major problem yet.
-
LongForWisdom: So, Primoz made a point with me the other day, that it’s all well and good, but if there would be a bad debt situation which happens, then it will be canceled by the surplus buffer which doesn’t exist, and DAI will not longer be 100% backed. That is also something to consider.
-
Primoz: Yes, that’s a bit of an edge case. Well, maybe not an edge case, guys, but yeah, basically, if you have a loss event and the surplus gets drained, you would develop an ambiguous situation.
-
Kurt: Yeah, from a risk perspective, if you know that you have a large amount under-collateralized or underwater, specifically vaults, which have actual bad debt and you know that’s in the surplus buffer, then you need a little extra cushion on top of your surplus buffer in case of emergencies. However, that’s a risk related discussion in order to evaluate the pros and cons.
-
Gregory Di Prisco: Yeah, I think that’s a risk related discussion. Also, keep in mind, if the worst-case scenario happened and there was unbacked DAI, the Maker holders could manually trigger an option and raise funds. It’s no different; it’s just a different sequence.
-
LongForWisdom: Yes, that’s true. We would then be forced to potentially liquidate them when we’re above peg.
-
Gregory Di Prisco: Now we do the flop
auction.
-
LongForWisdom: No, but this is the point; flop auctions wouldn’t trigger because the protocol would think we had enough money, but we actually wouldn’t have enough because the USDC vaults would be underwater.
-
Gregory Di Prisco: Yes, but I am saying is you will have to do it manually.
-
LongForWisdom: Yeah, I guess you probably could do that.
-
Gregory Di Prisco: And this is such an edge case that if we get to that point, I would rather deal with it manually anyway.
-
Christopher Mooney: How do we envision performing the increases regarding this? Would we do this every week, or go up by a little by the amount of bad debt we believe we’ve accrued to keep the buffer accounting for that?
- Gregory Di Prisco: I would do it monthly, but I’m lazy.
-
Sam M: There is a practical limit, as well. If we go below 100% collateralization, we may begin treading into a territory where we wont be able to gain anything on that. Therefore, there should be a practical set for lower limit. We can’t just run it forever.
- Primoz: Yes, I suggest we go month by month, but still do some monitoring between the months in order to see how the protocol behaves through supply and demand. We will have a better view of this over time. Personally, that’s how I would do it.
-
Kurt: I’ll emphasize again that as long as you keep the collective fees and the surplus buffer, or assert that you maintain enough surplus buffer to cover those unbacked fees, you are never actually in a situation where you’re losing money because you can always liquidate, or otherwise sell, the vaults in some way. As long as you got the collected fees there and you haven’t spent them, you’re okay, and you can hold onto it with the hope that eventually DAI goes below a dollar. You can recover some portion of the value of those fees. An emergency shutdown isn’t affected either as long as there’s enough surplus buffer to cover the unbacked DAI, whereas there will still be dollar parity. Finally, yes, it does become a little bit of a thing to manage if you have to watch it closely.
-
Sam M: Regarding liquidations, though, if we were to turn those on, some inefficiencies would lead to an overall loss that’s greater than the fees collected so far. That’s supposing the price is stable.
-
Kurt: Even if you do turn on liquidations and DAI is below peg; it depends on whether you’re going to use the current auction system or if you want to wait for the new auction system, which is probably going to be more efficient. I think you would get very close to the value of the collateral back if you did it carefully. As you said, there is a possibility for some inefficiency there. How big do we estimate that inefficiency to be and factor that into the risk discussion?
-
Matthew Rabinowitz: If you fast forward the video, let’s say an entire calendar year to do a gross exaggeration if I understand it correctly, it’s a declining return. We’re capturing value today because of an RP that’s associated with it, and over the time of one entire calendar year, if we got into that scenario, we would in effect be giving back those return where we wouldn’t earn anything from it. We would have “stability.” DAI would be a proxy wrapper at that point for USDC, which would be the same as if we set the stability fee to zero percent if we were to wait an entire calendar year. Is that fair?
-
Sam M: If we waited an entire calendar year, the leading vaults would be at 97% collateralization with USDC-DAI, and that would lead to overall losses. I don’t know, but we would have to get the price drop below 97 cents for those vaults to close themselves out; it would probably very likely go into liquidations at that point, in which case we’re getting some loss due to auction efficiency. This is why I would say there should be a lower bound of approximately 100%, where we should turn off SF because they’re no longer giving us anything. We might even take on losses in the long run.
-
Gregory Di Prisco: Wait, Sam, I don’t understand that at all. If they’re at 97 cents, but that extra three cents is sitting in the surplus buffer, we need to liquidate them at 97 cents. Why would we forgo the fees on all of the overcollateralized vaults to prevent that from happening?
-
Sam M: Sorry, when I said 97 cents, I meant 97% collateralization. If DAI returns to a dollar, they’re still under-collateralized by three percent.
-
Gregory Di Prisco: Yeah, but we already collected that money; it’s sitting in the surplus buffer. We only need to get those vaults out of the system at their current collateralization ratio.
-
Juan: Yes, but if those go to auction, there’s going to be 3% in bad debt that’s going to have to be…
-
Kurt: Well, I don’t think we’d liquidate them that stupidly. We’d set the peg to a lower value or use a different auction mechanism for them at that point. That’s something we have control over as the DAO.
-
LongForWisdom: The three percent difference between the 100% CR and the 97 percent ratio is that it exists in the buffer, which is kind of what Greg is saying. Therefore, it’s not lost; it’s just in the buffer, and when the auction happens it then subtracts three percent from the buffer, which thereafter goes back to what was.
-
Sam M: Yeah, but that’s a neutral operation. I don’t understand why we would run stability fees past 100 percent collateralization.
-
Matthew Rabinowitz: The point is that if everybody were at 100%, then you’re right. Some that will be at 100% and some at 102%. As an example, for the ones that are at 102, you would continue to charge a fee and either manually liquidate the ones that are at 100 or wait for them to lower. This is for when the blended number is below 100, but even then, the people above 100 are still generating value for the protocol.
-
Sam M: Yeah, sure, that’s a fair point.
-
Kurt: Basically you could either come very close to breaking even or if you get the situation where DAI goes below 100, then you could get positive income.
-
LongForWisdom: Ok, we’re getting a little late, so we’re going to wrap up here. I feel like we’re slowly converging on some shared understanding of how this works because we’ve discussed throughout a couple of meetings. I’d encourage everyone to join in on the forum threads and try and figure out what the various options are, and what the benefits and downsides to each option are. Once we have that figured out, we can then move towards figuring out options and getting an agreement on what we want to do. Thank you, everyone, for their points on this. Are there any final reminders or anything before we close out the call?