Poll: options to handle COMP farming

Compound Governance Proposal: Update cDAI Interest Rate Model

Also, wondering why @rleshner polled for “Add MKR as a Vault collateral type with high debt ceiling and high stability fee type parameters, something like USDC-B” – highly doubt the Maker Risk Team would agree with such opinion :thinking:

Speaking as a MKR holder, its an efficient route to draw DAI. Nobody knows MakerDAO better than MKR holders, and the “availability” of the asset probably rivals or exceeds ETH. I’m not sure why it’s considered so taboo.

2 Likes

The problem with USDC liquidations is that the current auction system cannot handle them well if they happen at a large scale. The new upcoming liquidation system would do so. But it is not as urgent as with other assets from the risk perspective, since USDC risk is very much all or nothing.

The difference between Forex and USDC as collateral is that there is no market risk - both are USD. To better understand what it is you’re worried about, can you explain what scenario you imagine where having real time USDC liquidations online would be useful in the short run? So not a situation where USDC has been banned/seized/frozen because trying to do a liquidation then wouldn’t really matter.

Keep in mind governance can always manually activate liquidations, or turn on the real time oracle based liquidations.

And let me try to better use examples of how a low USDC LR helps create an upper bound on the peg.

The peg goes above e.g. 1.01 USD because people are buying Dai with USD and are paying more than 1.01 USD for 1 Dai. But if the cheapest way to buy Dai becomes buying 1.01 USDC, and then generating 1 Dai with it, people will stop buying it on the open market, and start generating it instead (thus alleviating open market price pressure). If market participants still pay above the 1.01 on the open market (e.g. 1.02), then arbitrageurs will step in and sell e.g. 1 Dai for 1.02 USDC, then use 1.01 of that USDC to generate the 1 Dai back and then pocket the free 0.01 USDC. This arbitrage will continue until no one is buying Dai on the free market above 1.01, as long as a single arbitrageur is active.

1 Like

I guess the risk with usdc liquidations turned off is more just not being able to stay above minimum collateralization as interest accrues… which is a risk that becomes much greater as the LR moves close to 1. with a 1% SF at 101% CR I believe it only takes a few days for a maximum leverage vault to be underwater.

2 Likes

PSA: USDC-A debt ceiling is already 95% used, ETH debt ceiling 82% used, and DAI is already 1.3% above peg. COMP farming hasn’t even started yet—it will pick up tomorrow as Compound gov proposal is executed. Pointing this out to emphasize that there is some urgency in addressing this issue ASAP so that DAI does not drift too far off peg.

2 Likes

Can we have an onchain vote ASAP about the LR for USDC-A (i.e. 101%, 102%, 105%, 125%…)? It seems that we also might want to vote for the new USDC-A DC.

4 Likes

There are a couple of assumptions implicit in your argument.

  1. There will be 0 interest charged otherwise over time these positions can go underwater relative to surplus due to interest. at low interest this could take a lot of time but then what about abandoned vaults? Another implicit assumption here is that high DAI PEG means low or 0 SF.
  2. Orderly liquid markets - without significant fees and with liquidity to fund them (not just DAI but USDC etc.)
  3. All DAI holders value $1USDC or even 1.01USDC = $1 USD
  4. Implicit in the above is that the DC will never be hit so that DAI always can be minted.

I think there are cases where these assumptions will be violated that lead to different forms of risk either for DAI holders or MKR holders.

I am skeptical that a liquidator would actually pay $1DAI for 1.01USDC if the DAI price is above 1.01USDC since bidding and moving money around involves fees and liquidators want to profit not lose money. I am also skeptical that ALL DAI holders value 1 USDC = 1 USD. We already know markets can get ‘disorderly’. I am skeptical the USDC SF will be 0. In fact it is not 0 now even though at .25% it would take something short of 4 years to put the position underwater due to interest. Also unless I see MKR governance putting the USDC DC at say some effectively infinite or very large number that the DC will be hit and DAI won’t be mintable, or it will be at a higher price.

Right now USDC-A is almost at capacity after being down for a while and we still have a PEG around 1.5 the question then again will be whether to even have a DC. The whole interest thing on USDC-B would also still apply even if one set the 1.01 LR there - in fact it would be worse as it would only take a week to go underwater on interest there. Where does DAI get minted if we are having a collateral asset price crash and possibly a USDC price drop below $1 USD. In this case even if we were liquidating the net DAI yield for 1.01 USDC would likely be below 1 DAI making a system surplus loss even if small.

I grant you - IF the above assumptions hold - then your argument is reasonably sound. I don’t believe we can count on the assumptions above to always hold and in particular to hold for all DAI holders at all times.

Turn your argument around. IF USDC is so valuable why not allow liquidations to be turned on since based on your assumptions the system would never lose money. Right now the answer is because the liquidation system can’t handle the massive load. Why is this? Because without the fear of liquidations there is no real structure to the collateralization of vaults. Everyone can lever up to max at the same LR without fear of liquidation where as there at least would be some skew of collateralizations if liquidation was turned on meaning not everyone is going to lever up and do it at the 101LR level because they would be concerned about either interest or DAI USD price rising to cause them to be liquidated and ofc they would get nothing back.

Your example basically incentivizes the worst behavior without any control or possible scatter to allevaite a massive liquidation scenario and just heaps risk of some uncovered loss into the system…

EDIT ADD: I think this idea has some merit but I think a LR of 101 really doesn’t build in any system safety margins from a risk perspective and liquidations off means there is nothing to temper borrowing behaviors to provide system Collateralization variance so the vaults don’t clump at the same CR levels.

Even with a 120% CR, the USDC-A Vault is now at 95.74% utilization.

4 Likes

One thing I want to remind people of. The Maker system ran with DAI PEG > 1 with a suplus buffer -5.3M with something like 70M DAI outstanding for over a week.

I for one was surprised that the markets did NOT immediately punish the DAI PEG by dropping it to like <.95 because of this until MKR was auctioned to fill the gap. Normal and efficient markets would have done this. It was one reason I suggested and suggest simply minting DAI out of thin air (the threat of this should drop the DAI PEG by itself), doing it but not buying anything should actually drop the PEG, and direct injection would/should be the final nail to actually lowering the PEG via directly injecting this unbacked DAI into the markets to lower the PEG and buy assets that are in a seperate Maker reserve account (not available if system ES’s to DAI holders except if Maker governance specifically approved it).

I honestly would like to see other DAI creation mechanisms - something more along lines of @cyrus suggestions and possibly @rune except not with a 101LR but I think given the size of the compound driver here we could bloody well need to mint .25-.5B DAI to satisfy possible demand. I just am concerned if we play into this this way vs. first doing something of the opposite to measure what price demand exists at we simply are going to whipsaw the markets when compound makes the next move.

I keep hearing this DAI > 1.0 crushes people. It hurts vault holders but with SF fees on the most used vaults running .25% I don’t see a high PEG hurting them here. It simply is a market based SF based on the DAI premium and this is good for DAI holders. If vault owners are well collateralized this has no impact on them since they can simply wait to pay down debt because the additional PEG cost is literally 4 years of current SF. If anything these vault holders could add more collateral and take advantage of the DAI short opportunity. If their collateral is going up and they want to get access to collateral all the higher PEG does is eat into trade profits to buy DAI to release capital to capture slightly less gains.

Literally a high DAI PEG incentivizes DAI minting via users opening vaults here there and elsewhere. The real issue generally is liquidity and this is where the rub always seems to come and what we are learning is that a decentralized approach trying to incentivize people to take this on may simply not be enough. Going off the wheels to incentivize this by tossing system risk out the window pushing system parameters to the edge or into unlimited catagories doesn’t have to be the only tool in Makers arsenal. But right now IT APPEARS to be the only TOOL IN the arsenal due to the current structure of the system determined by the current smart contract structures.

3 Likes

My feeling is that we’ve been too shy/risk averse to increase debt ceilings/lower CR. The market has determined what it wants and by us not giving it to them (more USDC/WBTC) we’re only shooting ourselves in the foot. At this point it seems like we have to balance systemic collateral risk with DAI peg risk and determine which is more likely in the short/medium term.

I still have faith that market forces will play out properly to keep the peg closer to $1 than $1.10 but we should definitely keep driving forward with new collateral types and other tweaks to existing ones. It’s important to stay pro-active instead of reactive.

4 Likes

I appreciate this view Aaron I think I would just like to see more of a measured approach to loosening with some eye to measurement mechanics and some application of science to optimization than a whole hog - throw everything we got at it at once and leaving ourselves without any options if it looks like we are getting trapped thats all.

Perhaps we should list the actual available options within the system now that we can do and then list ones that we might want to have in our arsenal that would take work and see if we can get risk teams to kind of rank what they think are the least risky and most impact on this list so we all can hone in on the best options because we may find ourselves needing to act and I completely agree with you we want to be ready to ACT strong and confidently but not RE-ACT in panic and fear.

I think some comments from @cyrus and also @vishesh on how we are going to take measurements to asses results of actions during a possible upcoming mess might be warranted so we have a handle on what we will use datawise to drive the results of previous decisions to funnel into the next decisions as this may be a fast moving issue. I hope this all is a non-event but you all have a worst case planner here so yeah. I will almost always err on the more conservative sides of things, except to toss out what I feel are off the wall ideas just to vet them. Which is why I liked @rune tossing his LR 101 USDC idea out as it is pretty off the wall so we can analyze extreme ideas to see maybe if we can find one that goes in a good direction and allows for whole hog but has a means to go there in a stepwise methodical fashion even if the steps end up being somewhat logrithmic.

Definitely yes on USDC-A DC unless one wants to create a new USDC-C facility for this.

Given other comments do we really want to poll on 101, 102? If we do that add 103, 104, 105, 110, and same 120, as well as 133, 150. basically the approx 100x, 50x, 33x, 25x, 20x, 10x, 5x, 3x, 2x leverage options.

I am mixed on whether we want to make another ETH facility with lower LR and higher fee. I think we should poll cleanly on the USDC LR, and then do the DC since higher leverage folks will probably want a higher DC than low leverage folks (hard to say without polling). The real question is going to be what SF or really the RP change if any on this as I think a higher rate due to risk would be a prudent step here.

Is the above USDC the preferred option or should we look at other options?
Onboarding more collateral?
Expanding ETH either with a new facility or using the existing one?
What else is realistic and doable now with existing system?

I definitely agree that with any sort of positive fee, and without reliable auctions, the USDC LR needs to be significantly higher. But I don’t think the value of getting a bit of MKR buy and burn is worth much when the cost is a worse peg (in a situation where the peg is already broken and has been broken for a long time). I also don’t think having a lets say 5% SF on USDC for 6 months would help make any difference if e.g. coinbase failed.

On the other hand I think that having shown user growth and better momentum for the project by delivering a more acceptable user experience through a less broken peg, would probably make a big difference, because it would impact whether the market would be ready to buy a lot diluted MKR if some sort of big catastrophe happened. <-- thats really the crux of my argument - I think people who want to be conservative with risk parameters believe they are reducing risk to the system, when IMO it is potentially doing the opposite if it’s making the UX and growth prospects significantly worse.

Obviously the ideal situation is a world where there is plenty of collateral and a decent DSR, and diversified stablecoins available as collateral at low LR’s with high enough SFs to properly account for their risk from a risk management perspective.

I also think the idea of a direct liquidity/stabilization mechanism built into the DAO is very interesting. It’s something that I’ve considered too radical in the past but now we’re post black thursday and getting into the defi free money bonanza I think it should be considered more in depth by the community.

But back to the topic of what to do in the short run: Ultimately it’s not my decision what gets up for a vote - For risk decisions like this I’m just another voice in the community, so ill support any measure that can help. The most important thing is just that we do something, and that we always take a broken peg seriously.

The biggest risk IMO, is the market losing confidence in the commitment of governance to maintaining the peg.

5 Likes

Just a quick note to those arguing for immediate action in some form. Given the polls in the OP, the only action there is consensus for currently is decreasing the CR for stablecoins. Given that the debt ceiling has already been fully utilised for USDC, there doesn’t appear to be a large amount of urgency to make that change.

Emergency action can take place, but we need to have a solid justification for doing so. In my opinion, this would take the form of either:

  • Getting a majority on a poll specifically calling for some sort of emergency action outside of the weekly poll cycle. Ideally with a majority on what that emergency action should be. The more support a poll has, the harder it is to justify ignoring it. So aim for more than a simple majority.
  • A clear statement from @Risk that emergency action is warranted. Again, ideally with a clear plan of what that emergency action should be.

There is a segment planned in the Governance and Risk meeting later today in which everyone can discuss the impact of COMP. I encourage those participating in this thread to attend the call and take part in that discussion.

2 Likes

I do not think any immediate action is required, but what I would very much like to happen is that one or more possible emergency responses are planned for in detail. So if things becomes too hot we can respond quickly without lengthy discussions.

3 Likes

Any ad hoc (using that word now instead of ‘emergency’) polls would definitely include a debt ceiling change as well. In fact, debt ceiling change may be first, with liquidation ratio changes to come later, since those are a bit more aggressive. We just saw 20 mil get taken at 120% LR. We could consider lowering the LR as a policy tool to keep in the back pocket.

The most important near term change is the increase in debt ceilings. Can discuss on the governance call today.

5 Likes

There was no option in the poll for “doing nothing in the short term (days)” so I guess there is no consensus for that either. In that case, we should do something (anything) i.e. let MKR holders decide.

You can’t assume consensus (or lack of it) for an option that people were never given a chance to vote on. I suspect that we will end up doing something. Whether that happens this week or next week largely depends on the criteria I shared above about taking emergency action.