Onboarding USDC as collateral to mitigate liquidity risk

Given the ongoing liquidity risk of Dai in an uncertain market, both the community and members of the Maker Foundation have been making noise in the official chat about onboarding USDC as collateral as an emergency measure to help mitigate this liquidity risk.

Naturally this would need to be confirmed by governance, and go through an executive vote, but at this stage the Foundation is making the technical preparations to make this possible.

I will be updating this initial post with pros, cons and points to consider presently. Please feel free to share any thoughts in the meantime.

See this previous discussion for an overview of why this is useful: What's the point of adding stablecoins as collateral?

A call is planned in which the Foundation Dev Team will be going through the code on the collateral adaptor for USDC. It’s currently slated for 4pm CET / 11 am ET / 3pm UTC Edit: Call has now concluded. Recording is available here.

Proposal from Cyrus is now live, complete with community polling: Proposal for Collateral Onboarding of USDC


Pros

  1. Helps to create Dai liquidity and push the Dai peg back towards $1. The mechanism of this looks like: Lock USDC -> Mint Dai -> Sell for USDC -> Repeat.

  2. Will allow Vault Holders to close their Vaults without eating the loss on the Dai peg being high against USD.

  3. Will allow keepers that want to participate in the ongoing FLIP auctions to acquire the Dai they will need for those auctions.

  4. Will allow keepers that want to participate in the coming FLOP auctions to acquire the Dai they will need for those auctions.

  5. PR Response to fixing liquidity might be positive.

Cons

  1. Reduces the ‘purity’ of Dai as a decentralized stablecoin - Dai is no longer backed entirely by decentralized assets.

  2. Comes with regulatory risk - US Government could decide that dollar-backed stablecoins are not okay.

  3. Comes with ‘blacklist’ risk - KYC/AMC could result in Circle blacklisting some amount of USDC used as collateral.

  4. PR response due to Dai purity might be negative.

Points to consider

  1. We need to decide on a Stability Fee - Note that this should probably lean towards the higher side, at some point Cyrus will share a risk perspective.

  2. We need to decide on a Liquidation Ratio - Note that this should probably be lower than ETH’s but should not be so low that a single Vault can mint all the USDC, at some point Cyrus will share a risk perspective.

  3. We need to decide on a Debt Ceiling - Debt Ceiling especially is something that is to some extent more of a governance decision than a risk decision. We need something that is high enough to provide the required liquidity, but not so high that we take on additional risk.

  4. Can we get away without oracles and hard-coding the price of USDC to $1? What are the pros and cons of doing so?

  5. Is the collateral adaptor safe? There likely won’t be time for an audit.

  6. This can be a temporary measure, we can lower the debt ceiling to zero once we feel the crisis has passed. (Note that this does require the crisis to actually pass and people to start opening ETH Vaults again.)

  7. We need to decide on a Liquidation Penalty. As ‘debt-from-Stability Fees’ increases, Vaults can still be liquidated even if the underlying asset price is stable!

  8. We need to decide on a Lot Size. If liquidations are going to occur, we need a lot size that makes sense for a stablecoin.

  9. Other miscellaneous parameters? There are several of these that all need settings.

7 Likes

Right, I think that about covers everything I can think of immediately. Please @ me if you think of something else that I’ve missed that should go into the initial post.

1 Like

Thank you for the intro LFW,

Let’s all discuss this today on a community call at 4pm CET / 11 am ET / 3PM GMT, Invite details here:

Meeting ID: 723 226 374

I would also add liquidation penalty to your list of risk variables, as well as lot size for flip auctions, and dust limits.

In terms of planning, if we determine this to be suitable for onboarding, we would consider putting up an executive spell after the current executive has changed the GSM to 4hrs, which occur tonight after 19:15 UTC, then we would need to vote on a new spell - let’s discuss this on the call shortly.

2 Likes

I’m actually not sure how important those parameters are. If we hard-code price then no one is getting liquidated. Even if we don’t the price of USDC is going to be very close to one anyway. Personally I’d be happy to defer to Risk for those parameters.

Dust might be more important, that’s the minimum size of a Vault?

I will add them to the post though.

Yeah, for completeness I was trying to think of all the variables to make sure it’s as transparent as possible. Happy for us to add/remove as you see fit.

I think it’s a good idea to add USDC as collateral, to enable liquidity to be added to the market for emergencies. Currently DAI is trading >1.07, there’s a lot of demand for DAI liquidity.

Vaults will be able to lock USDC and mint DAI. They’ll be able to sell the DAI and provide DAI liquidity to the market.

Why USDC? USDC has a lot of on chain liquidity and is also integrated with many defi other defi products. Also, DAI/USDC is traded a lot too.

Price Oracle will be pegged at $1. They won’t be able to get liquidated from DAI/USDC price movements.

My personal thoughts on Variables.

  • Stability Fees [sf+15%] - So I was initially thinking something lower. Then Cyrus mentioned 10% and made some good points. We generally want this to be mainly used when there is a great need for liquidity, and not used up for slight deviances in the DAI/USDC price or for rate arbitrage in normal circumstances. This could possibly raised higher, in emergencies compound has a 20% sf fee and dydx has a 50% fee. But unlike those, you can actually mint DAI with this maker facility. Vaults should still be able to profit as they shouldn’t be holding for a year. The main thing is that we don’t want the DC to be used on peg arbitrage when the peg is slightly out of sync, also historically it’s been very normal to have a 5% spread between USDC deposit rates and DAI deposit rates.
  • Collateralization Ratio: 120-130% - This requires that vaults use some of their own capital to put on these trades. We want to require capital, so that someone with low capital can’t just utilize the entire facility. Also we don’t want them to get instantly liquidated from sf fees at a loss for maker. At 120-130%, vaults will be able to lever up 3.33-5x.
  • Blacklist risk - This might not be that big of a problem. Compound also uses USDC and they might have an understanding with coinbase. Definitely not a problem in the short term.
  • Debt Ceiling - 10-20m - this is a tradeoff between risk and liquidity. There is probably more than 10m in liquidity demand. The debt ceiling could be greatly lowered later (2m?) when the crisis pases.

I think even with a stability fee of 15%+ and a Collateralization ratio of 120-130%, there will still be a lot of demand for this facility when DAI liquidity is needed (current situation).

3 Likes

Oh! I completely forgot liquidations can be triggered from Stability Fees. We definitely do need to consider liquidation penalties and lot sizes.

1 Like
  • In the short term, I don’t think we have to worry about regulatory/blacklist risk at all. In the longer term, this is just one type of risk among many.
  • Agree on @LongForWisdom’s reasoning on debt ceiling (as required by liquidity, modulated by risk)
  • No strong opinions on SF/CR/penalty/lotsize.
  • I’m very worried about fixing the oracle to $1. Unless USDC/$ convertibility is trivial and has infinite liquidity (I don’t know if that’s the case), if e.g. USDC starts trading below par anyone could prop it back up at the cost of bringing down the DAI price (which might have been perfectly happy at $1).
  • I’d much prefer an actual oracle, if doable quickly. Otherwise we should strongly commit to getting rid of the fixed $1 as soon as possible.
2 Likes

@sawkya How do you suggest finding the value of 1 USDC? Seems difficult.

1 Like

I just want to encourage people to think through the implications of possible parameters when the DSR is greater than 0%.

Or maybe we don’t care about this case because using USCD as collateral is just a temporary measure?

1 Like

For sure, that’s a good point. SF for USDC needs to be kept above the DSR at all times.

1 Like

With 1m USDC of capital they could create 4m DAI backed by 5m of USDC. They’re paying fees on 4m DAI. Someone else could earn the DSR on 4m of DAI.

Yep, @Jiecut is right. The SF to DSR is still 1:1, not 1:5.

1 Like

To me it sounds like a band-aid to a deeper underlying liquidity problem which ought to be solved with careful consideration. Yes, supply from the side of the vault owners was limited at times, but the main issue was extremely low participation in the liquidation auctions themselves causing valuations that were too low. If you talk to the liquidated owners, you will find that they are mainly frustrated about the amount of lost collateral (for various reasons…), not the liquidation itself.

Contrary to what seems to be the current belief, I’d say that the problem is therefore not that we didn’t have enough DAI, it is that DAI users were in fact trying to participate but unable to due to various reasons (no UX available, limited knowledge about the keeper system, 500 ETH fragments are simply too high a barrier for many people, etc.). People were begging for ways to get to participate and add liquidity! I am aware of the current efforts to solve this, but in my view we ought to first explore removals of these obvious barriers to the otherwise well-designed auction system first.

As to the USDC proposal itself, there are new economical risk that can sneak in with such a solution and mixing it with an otherwise decentralized system … Lastly, I echo the sentiment that I am in this project for it’s pure decentralized nature and would not want to see it becoming inter-mixed with USDC. If it were, I would rather just want to stick with USDC altogether.

2 Likes

Concretely there are at least 2 exchanges with USDC/USD markets and more price feeds can be synthesized by going through other assets. That’s not great I know :confused: It does seem that USDC convertibility is straightforward at the moment.

So as far as I’m aware coinbase allows convertibility at 1:1. I don’t believe there are limits on this either. I think we could get away without oracles in the short term. Long term it may be better to put some in, or it may not. I know dydx just prices USDC to $1.

1 Like

I’m not sure USDC is a good way to help liquidity and the peg (decentralization). I think the current parameters (0.5% SF) will bring down the DAI price but I’m not sure how long it will take (maybe days maybe weeks).

I’d take a different approach and allow negative DSR until almost all funds are withdrawn. Ideally, DSR should be baked into the protocol but I’m not sure how much work is needed (and is there a consensus about that?).

It is not a bad idea though (USDC collateral) - it might be worth to have it ready.

This assumes coinbase always will convert without limits, won’t have reserve issues, won’t have delays that naturally lower the price of USDC just because of varying time preferences, etc. I’d rather us be distanced from these risks using a price mechanism than one day becoming a free money pump for USDC holders.

That being said, I’m OK with a fixed $1 in the very short term, but we should do it with our eyes open: this goes against a lot of reasonable principles & is like sweeping some tail risk under the rug.

4 Likes

Given the context of this collateral onboarding: ensuring there is adequate liquidity for Dai in the upcoming MKR auction, I support fixing the price feed from Dai to USDC. The price feed for USDC can always be changed later on (taking the governance delay into account) and this offloads a huge amount of complexity for on boarding USDC today. The community can set a target to replace USDC’s oracles in say 21 days. At this point a governance vote could either remove the collateral from MCD or replace the oracles.

Maker is trying to set up an auction for ~5.76M DAI in the coming days. For a sufficient amount of participants in an auction, this will lockup a multiple of that amount. However every DeFi protocol that encourages locking up Dai is reducing liquidity in the market. There are currently 37M Dai in the DSR (2M in Chai) and the uniswap liquidity pool has over 1.5M Dai. All of these won’t necessarily become liquidity to help with Dai demand even if Dai trades at an even higher premium than right now. We could very well see Dai going up to $1.1 again during the auction which puts an additional exchange risk on all of the keepers and will likely get priced into the MKR price (an additional dillution of MKR).

@jiecut’s point I think is important:

  • Collateralization Ratio: 120-125% - This requires that vaults use some of their own capital to put on these trades. We want to require capital, so that someone with low capital can’t just utilize the entire facility. Also we don’t want them to get instantly liquidated from sf fees at a loss for maker. At 120-125%, vaults will be able to lever up 4-5x.

Without overcollateralizing USDC backed vaults, someone borrowing right at the liquidation threshold and subsequently triggering a liquidation will put additional strain on keepers.

1 Like