[Signal Request] Quantitative easing to fix the peg (aka Manual PSM)

SebVentures · September 6, 2020, 5:21pm

Summary

Creation a USDC-M vault with no stability fees and a liquidation ratio of 100%. Only whitelisted address from Maker can use it. This bring the ability to unlimited leverage if needed to:

Fix the peg
Turn on the revenue stream (separate the SF parameters from fixing the peg)
Make a profit doing so

It is quite similar to the PSM but manual (PSM prototype?).

Motivation

The Maker Protocol is a two sided business. Vault owners want to get DAI loan (producer side), DAI users want to have DAI (consumer side). Both want 1 DAI = 1$. Currently we have more consumers than producers. We set the SF to 0%, DC are not used to their full extends.

With a USDC-M vault with no fees and a liquidation ratio of 100%, we can have unlimited leverage and arbitrage ourselves the DAI price.

The USDC-A lower LR proposal is similar but I think less effective (higher exposure to USDC for the same effect, leaving the profit to the users, having to handle “abandoned” USDC-A vaults that are not easily liquidated).

This proposal is very similar to the quantitative easing programs from central bank that were quite effective at setting rates.

Quantitative easing process

Create a USDC-M vault that only some wallet can use (i.e. only Maker) with no SF and LR @ 100%. No limit or a 100M USDC limit.
Borrow 100K DAI with a flash loan ( as an alternative to take 100k DAI from the surplus buffer)
Sell the DAI to get USDC (around 101k USDC at first iteration, more later)
Use USDC as collateral in the USDC-M vault to generate 101k DAI.
Pay back 100k DAI of the loan, use the 1k DAI remaining to get ETH for transactions costs
Repeat from 3 until the DAI don’t sell at 1.005 USDC/DAI.
Unwind if 1 DAI <= 1 USDC

Pros

We control the whole process
100% of the generated DAI goes to pressure the peg downward. This is unlike the behavior of USDC-A that tend to keep the DAI in a Compound loop (very limited impact on the peg at best).
When we unwind we most likely make a profit
We can liquidate at any time.
We can stop the process anytime if the results are not great
We can say “we do whatever it takes, and believe us it will be enough, don’t fight MakerDAO”
That would enable Different Approach to Rate Setting
The 1 DAI = 1$ of collateral is kept

Cons

A lot of manual work (fake it until you automate it), therefore operational risk
Operational burden (who do it, under which supervision, …)
If 100M DAI from USDC is not enough, it will increase our last accepted exposure to USDC.
May have unwinding issues if we want to unwind when DAI is expensive.
A government may freeze MakerDAO USDC collateral
The strategy may have limits as it would be public and may be front runned (pro or con not sure).

Further possible iterations

Add other stablecoins vaults (TUSD, PAX, GUSD, …) as soon as possible to diversify risk, including cDAI and similar
Automate it and create a whole framework to make it a peg stabilizer (a real PSM).

Signal Request: Should we reduce the USDC-A collateralization ratio? Alternative proposal to lower the USDC-A LR for all customers
MIP20: Target Price Adjustment Module (Vox)
[Poll] Should MakerDAO print unbacked DAI to solve the peg issue? First iteration of this idea
MIP13c3-SP3: Declaration of Intent - Strategic reserves fund (SRF) can be used as source of funds
Different Approach to Rate Setting
[Discussion] cDAI/aDAI/yDAI collaterals as a peg enabler tool

Signal request

Do you want to put forward this quantitative easing plan (Maker leverage a USDC-M LR100% SF0% vault to fix the peg)?

yes, this is the way
no, not a good idea
abstain

0 voters

If you voted no please help us understand:

I voted yes or abstain previously
No, I prefer the USDC-A with lower LR solution
No, I want the PSM
No, I want negative rates
No, I want unbacked DAI
No, the solution is RWA collaterals
No, the solution is adding more collateral
No, too much operational work
No, we can’t use the surplus buffer for that
No, the peg is not such a big issue
No, I don’t like USDC
No, we should do X instead (please explain below)
No, I have another reason (please explain below)

0 voters

The poll will be open until 2020-09-07T22:00:00Z

ElProgreso · August 31, 2020, 9:45pm

At this point – with 300,000+ DAI sitting in the sidelines (system surplus) catching dust–I rather explore–than sit around and wait for another Black Swan event–similar to the DAO hack (hopefully not)–or a major sell-off based on World economic, regulatory, & political risk

Let’s not forget that Not only do we have the entire Ethereum community dependent of DAI being Stable – but we have supporters out there wishing us success:

Jtathmann · August 31, 2020, 9:45pm

Trying to wrap my head around this. It sounds like the end result would be the equivalent of minting X million unbacked DAI, then retroactively backing it by purchasing assets (USDC) and locking it in a maker controlled vault that can be liquidated (but not until we have the Dutch auction system live).

While these peg stability tools are important to research in the medium to long term, the yETH launch is imminent, and I’m concerned the +200M curve stable coin buffer is going to get eaten up. I would not be surprised in the least if the 80 million left in the ETH-A vault gets filled on day 1

Aaron_Bartsch · August 31, 2020, 9:45pm

Great work here @SebVentures for taking this initiative. I agree that brave actions are required now and I think this would be a great experiment. Lets hope others agree!

Aaron_Bartsch · August 31, 2020, 9:48pm

We can hope can’t we!!! Oh if all our problems could be solved with this one addition!

SebVentures · August 31, 2020, 9:58pm

Yes, you can say it’s the same, maybe we can do it with a flash loan. No capital needed and it should work or liquidation as well.

I agree with the yETH, but what happen if the underlying strategy change and hurt the peg? Let’s move forward on both and activate quantitative easing only if needed.

For the PSM I didn’t add that this option was only valid if you are working on it

rileyjt · August 31, 2020, 10:22pm

My suggestion would be to increase the debt ceilings. They are not high enough to allow large players to successfully arbitrage the peg. Primarily ETH, but also USDC (doesn’t make sense to me that this was lowered).

Trying to do large scale arbitrage is risky because there is a chance the market moves against you. It needs to be done fast in order to limit your risk, but because doing this arbitrage would hit the debt ceiling cap before the peg was restored, you would get exposed to extraordinary risk market demand pushing against you.

DeFi can create demand of hundreds of millions in the space of a few days. DAI is simply failing to scale to the point where it can meet this demand. So people are making do with inferior alternatives like USDT because it is the only other option.

Current market demand for stable coins is over $10B on Ethereum and we are only supplying 3-4% of the demand. Why is there any confusion at all as to why the peg is off when we are constraining supply?

mario · August 31, 2020, 11:08pm

I don’t know if there are implementation difficulties but if there aren’t I see no harm in trying it out with a 100k from the SB. There should be a designated individual from the dao to execute it.

Edit: It would not be unbacked dai, we would be using our own dai reserves for abitrage trading, the only downside I see is a black swan event on usdc but it’s a contingent event vs a real problem

monet-supply · August 31, 2020, 11:25pm

I think there’s a big difference between Maker accepting USDC as collateral, vs Maker outright owning USDC (or other real world assets). If a government wants to target the Maker protocol, or if a lawsuit receives a judgement against the protocol, it will be much more difficult to freeze/attach USDC collateral assets that belong to a user, and comparably much easier to do that to assets belonging to the protocol itself.

mario · September 1, 2020, 12:10am

I think that if a government wants to target the maker protocol the consecuences go far beyond USDC, how probable are RWA under that scenario?

On the other hand if the government wants to target USDC we still have 11m under custody, doubt we would get any dai back

I believe there isn’t a perfect solution therefore we should act upon those that we believe may have the highest impact with the lowest risk.

Currently with the peg still hovering above 1 I find the approach interesting, it could even be a temporary one while we keep onboarding new collaterals, RWA included, if not we may be facing the fact that we are months away from a peg stabilization (if this idea actually helps the peg), isn’t that a far higher risk?

Planet_X · September 1, 2020, 5:44am

This step. You deposit 101K USDC and want 101K DAI in return. This is called printing DAI.

In this plan Maker is set up as a trader in its own currency with more privileges (a special USDC pool and exchange mechanism) than no other market maker has access to.
If the community uses such a solution it will cause a massive blow to credibility. You will probably be able to fix the peg in the short run but at the cost of sinking Maker karma below that of Tether.

bit · September 1, 2020, 7:22am

We should pause the ETH liquidations until the peg is restored but I’ll support any solution that is not status quo.

SebVentures · September 1, 2020, 7:54am

I would argue that printing mean issuing DAI without having a same value collateral. Here we only do what we do every day, making credit. Take collateral and offer a credit on it. the only difference is the SF and LR ratio. SF is just and accounting trick (we don’t pay ourselves), the LR doesn’t change the risk profile. The USDC risk is in main part binary (backlisting or default). It’s still MKR that hold the risk in any case.

I understand that a special poll can mean more privileges. In any case, it seems less privileges than running a PSM which would do the same with more contract risk, more market risk and less operational risk. In fact, I’m just suggesting a manual PSM.

The current best alternative on proposal is to set USDC-A at a low LR. Something I’m not sure we can unwind easily.

@monet-supply I fully agree but is really the US government our biggest problem? And do they really have only this option? The PSM would have the same issue (and I’m not talking about RWA). If we keep having the peg issue, the US government will be the last of our worries.

There is 80M on ETH-A and there was 130M on USDC-A. MakerDAO is not at ease to rush increasing WBTC-A. Therefore I’m not sure debt ceiling are the issue but would suggest you propose a signal request to increase the DC as you see fit.

You are nicely summarizing our problem. I would say that we are not constraining the supply but that the supply is not here. Like any marketplace, if the supply is not there, you make it. I would love to spend time figuring out why the supply is not following. Let’s solve the peg and work on that problem.

The add more collateral argument

As a lot of people think adding more collateral is a better way, I would say that I’m 100% in and we can/should do both. Quantitative easing is a problem fix, not a business model.

I’m super excited that we get PAXG this month but I’m not betting Maker life on it. It’s a long way and I would prefer the peg to be fixed during this journey.

On the next collaterals for inclusion, I see nothing that will solve the peg with an high probability. Are we okay to keep the peg broken for at least the next 3-6 months?

The flash loan option

The key issue (that isn’t often cited, interestingly) is the operational risk. But the flash loan option seems to me working, so we don’t even have to touch the surplus buffer. We just need the vault. The whitelisted wallet would be known by everyone for scrutiny and the vault can be disabled at any time.

Scale of the quantitative easing

The amount needed to bring DAI under 1.005$ is 40M on Curve (most likely some will frontrun us if we announce that, meaning we may need way less than 40M). No one can tell if that’s durable after that or if it will revert to 1.015. What is sure is that no one is doing it. Obviously, we will have to test it with a lower amount at the beginning and understand the market dynamics.

Replenish2030 · September 1, 2020, 2:42pm

This is a great idea, and the kind of bold move we need right now to keep our position in the DeFi economy. My suggestion is that instead of having the LR all the way at 100%, it could be set at e.g. 100.5% with a small stability fee, in order to capture a decent profit for Maker. It would still fix the peg within a reasonable spread of 1 USD, but would be a better trade for the DAO.

SebVentures · September 1, 2020, 3:09pm

@Replenish2030 thanks for your support. The stability fee is not needed as we pay it to ourselves. Both ways it end up in the surplus buffer.

The LR can be set at 100.5% but it increases the complexity of management as the spread is thinner. It’s our vault so we can limit ourselves to any leverage we want, I see no need to set that in the vault.

befitsandpiper · September 1, 2020, 8:54pm

I would like to see this exact same thing done in the long term with uniswap v2 tokens, balancer tokens, other crypto derivatives with lower volatility than bare crypto assets, no custodial risk, and yield.

But in the intermediate time period this seems like a good compromise since it achieves largely the same thing, only involves taking on temporary custodial risk, and these other assets which I think would be a better fit are not yet onboarded as collateral in the maker protocol.

I prefer this to the PSM since we have direct control over how much USDC we take on, and when we buy or sell dai on the market.

MakerMan · September 2, 2020, 2:05pm

Actually if you look at this carefully from an ES perspective all DAI holders basically own the collateral in the MCD system. Maker DOES NOT own the collateral in this protocol the DAO only manages this facility like all the other facilities for DAI holders.

While I am a strong advocate for direct intervention in the form of the above one thing no-one has considered is what the heck happens if the markets goes against this facility. When I look at what Central Bankers do there is something that as far as I can tell CAN’T be done by Maker DAO.

Stealth and power. Central Banks when interviening in currency markets don’t advertise what they are doing (price limits), nor how far they will go (liquidity they will throw at this). While they privately have a price target they generally DO NOT publicaly advertise if/when or how much they will move. The point here is facilities and activities like these are best accomplished with secrecy, and some stealth as well as real unknown financial power and then once they begin often times an announcement is made to try to influence the markets. These types of direct market interventions are tricky and require significant exposure as well as coordinated efforts not just in markets but with news releases. Meaning if one is going to mess with the Yuan say - you better have brokers that have direct access to all the markets at once because the power of these moves is that they happen across all markets simultaneously limiting arbitrage opportunities. You don’t want to be driving down DAI-USDC on DYDX and not also selling on Coinbase for example. I think there are real legal and tax hurdles in doing this kind of direct market interventions for this reason. It was why I wanted to advocate a more dynamic PSM that moved its pricing brackets when utilized and only allowed so much liquidity at a time to be pushed into or removed from these markets so that people could ‘expect’ there would be intervention - but market participants could NOT anticipate how much or at what prices the PSM would act. Hence in a way I wanted not just moving price bracketing based on the facility usage and OSM prices but also to use the actual decentralized users of the facility to drive an element of randomness in how such a facility priced and how much volume it would provide. (I never posted details of this as the PSM appeared to already have too much controversy and was not moved forward as far as I could tell)

I also felt it was important to couple an auction that constantly drove auctions when this facility was actively filled with assets on one side or another.

Example lets say this facility has like 80M DAI of 100M DAI utilized and basically sold at an average price of 1.01. There is no reason NOT to do constant reverse auctions for buyers between 1.005 and 1.01 while having the PSM decide to set the DAI buy back price at say 1.005 for users until the next market fill happens and then these would be dynamically adjusted again.

For such a facility to be useful it should appear pretty unlimited and should act somewhat randomly so as not to be easily gamed or gamable.

My growing problem with a lot of these tools as well as the increasing complexity not just of Maker but the entire DeFI system is that it is becoming more and more difficult to understand what the real ramifications of such a beast will be. Also we have governance decidedly unwilling to take on significant USDC/stablecoin risk. It has become pretty apparrent to me that governance really needs to sit back and take a look at what the future of Maker is. I started constructing a post but life events have side swiped my ability to contribute much lately.

Put simply when network tx fees are growing at the current rate we basically are pricing out many smaller decentralized players in Maker and pretty much destroying any ability for Maker to serve the underserved except by another third party who can in-expensively take on collateral and make micro-loans (i.e. loans in the 100-5000 DAI levels). Unless somehow this entire system gets the massive tx capacity boost we are rapidly going to see DeFI on Ethereum become the playground for the rich and so some level of discussion within governance on what the future looks like for Maker and decentralized users at not just 400-500gwei/tx but 4000-5000gwei/tx really is in order.

I saw wouter posting that fees just to run Maker are now running 50K/month. That is 600K DAI/yr which could rapidly grow to 1-2M if fees keep running like this. I don’t know about the rest of you but fees are pretty much locking up most of my positions in places and forcing me to do less not more in terms of market making activities. It is a prime reason that low cap collateral is somewhat pointless and why Maker is going to see risk clumping in virtually ALL of the MCD classes - particularly the small ones which generally is bad for bankers. We don’t want to have like 5 MANA holders or 3 wBTC holders taking up the entire DC because then there we have a huge and growing liquidation risk in these things. It is my hope the liquidation system upgrade will be up to this task - but this will be new and itself may have bugs or attack surfaces that may only appear when we have real clumping in liquidation risk.

befitsandpiper · September 2, 2020, 1:45pm

I’m not very convinced that transparency about what Maker is doing will cause issues. Sure in theory there’s some way gamification can happen. But I think in most cases the market front running us actually helps us because it creates a situation where the peg stabilizes without the need for MakerDAO to take on any assets.

Doing random things and losing the transparency we have which is one of our core tenants would be terrible. I believe that clear decisive action that market actors can count on creates a better dynamic. It’s possible that things can be gamed and for example the peg does not return to $1. However we have full control over these funds and can stop and find new solutions any time. I doubt there’s any real risk here besides the prior mentioned custodial risk, and the risk that dai will not return to a $1 peg.

The tax considerations are interesting and I’m not sure what they would be. But taxes in the crypto space are terribly nasty and convoluted already and probably should not greatly impact what kinds of decisions we make for government policy. If there are ways to do this and optimize taxation, we maybe should look into that, but there are a number of significant differences between this and PSM which are unrelated to taxation.

MakerMan · September 2, 2020, 2:09pm

If this was true one could just make a news release and markets would move to the new price. Typically the move is made and the news about it is releases so that market participants are ‘guided’ to stay in the new range. My point here is that typically these types of operations are tricky at best and one key component to any chance of success is stealth with power. Maker misses two things here, stealth, and with a fixed DC uncertain power.

One can still have things happen programmatically but using random elements. In PID control loops one normally doesn’t use such features as there is no outside agents acting to game the PID controller.
I think we will have to disagree. One can have clear decisive actions but do them unpredictably. We know that IF a system CAN be gamed - IT WILL and doing something like the above basically will reduce the effective power of such actions because as the DC fills up the gamers will know the price can be pushed over until ‘governance’ does something else. If this is programmatic it ofc will be frontrun. This on first glance looks like it will help but there is a reason currency interventions don’t do things this way and it is because the markets often themselves will do the unexpeted. Think of it this way. Do you think you would ever win at poker if your opponents could always see your hand?

As to tax implications. My thought was not dydx but trying to do these kinds of open market operations on CEX’s. Who is something like coinbase going to send a tax bill to? Which means they would be done outside of the CEX’s and arbitragers would make bank doing the arb on these - reducing the overall financial power of such activities.

I want to add one last thing. I really want to see something like this as a tool in the arsenal but some really clear thinking needs to be made as to how it could work and my primary issue here is that there is no stealth possible (AFAIK) and no uncertainty in ‘power’ that would be applied and so becomes a new system for outside agents to find ways to game for profit.

befitsandpiper · September 2, 2020, 2:36pm

Doing operations on CEX’s? Can you even interact with coinbase via smart contract? Sorry but I find this idea completely wild, and I’m not even sure how it helps. The entire point of decentralized finance is we believe it will eventually be more efficient/cost effective and overtake centralized financial markets, at least within the crypto space itself, right? Why would we want to bake interactions with CEXs into the protocol even if this was possible?

I still don’t buy the stealth idea, or that we need to use random number generators to make decisions. Can you provide some evidence of this claim? It seems like one of those things that maybe feels intuitive, but it’s also a significant claim which should require data. Even the US fed nowadays tries to be predictable and reliable for markets. There were examples in the past of the Fed trying to be unpredictable and outsmart markets and it ended poorly. Here is an article about federal reserve policy over time.

It seems to me the thing we need to do if we want to emulate the more successful aspects of the US federal reserve is provide forward guidance and MORE predictable long term policy decisions, not less predictable. Market actors will make the trades that are profitable. We just have to make sure that what is profitable provides incentives that align with government policy. And if new issues crop up, we have the capacity to find new solutions to those problems.