Summary of USDC protocol risk

A question came up in the internal chat asking what, if any, the risk of a USDC stability fee of 0% posed to the system; the following is my response, posted here at the request of others. I think that this is well understood by those who expressed their concerns about adding USDC in March, but I want to share here for those who are participating in governance votes that are determining the collateral fees.

The collateral stability fee is essentially an insurance policy against the risk of the protocol taking losses against a given collateral type. USDC is much riskier than it appears because it is centrally managed by a company that is subject to U.S. regulators. The USDC token has a blacklist mechanism that can be toggled if they are ordered to do so. The current regulatory climate for stablecoins does not look not promising, think of how Zuck got pulled in to testify over suggesting that Facebook create their own coin, and the Federal Reserve and congress have recently been making sweeping statements about their dislike of stablecoins. It is very likely that U.S. regulators could push on Coinbase/Circle to blacklist the Dai adapter to weaken MakerDAO. At that point, all of the USDC would be frozen in place and can’t be transferred in or out, so we can’t liquidate accounts and users can’t withdraw collateral. The protocol would incur a total loss for the balance of USDC on-hand, which could be worth up to $20 million dollars. A USDC stability fee of 20% (the rate of the USDC fee when the collateral was deployed) is basically an insurance policy that the adapter won’t be blacklisted within the next five years, at a 10% fee, 10 years. At 0%, the system is subsidizing all of the risk that USDC will be shut down in the future. I’m personally not comfortable with that. We’re still reeling from an $8 million loss a month and a half ago, I’m not sure that the system can handle the PR backlash and fallout from $20 million in losses and a frozen adapter.

13 Likes

This assessment is excessively alarmist. I’m not sure if you are from the US, but this is not the way the legal system works here. There is close to zero % likelihood (much lower than ETH flash crash) that the USDC custodians would be forced to black list the Maker adapter without giving people a warning and time to unwind their CDPs. People who have open CDPs with USDC have done nothing wrong right now and there is no plausible mechanism for the US govt to seize the funds of innocent US citizens without going through a long legal process that would give people time to unwind their positions.
Furthermore, if a criminal was identified who had deposited USDC into a CDP, the USDC custodians could easily just freeze the depositing account rather than the whole adapter. This would freeze the criminal funds without affecting innocent people.
Finally, keep in mind, that even if there was a temporary freeze on the adapter, the collateral would still have the same value. They would eventually work out a system whereby innocent users were able to close out their CDPs- but in the meantime there would be no reason to liquidate the frozen CDPs as their value would still be the same (the only scenario where it would be a problem is in a emergency shutdown).

2 Likes

So get more collateral released! Where is TrueUSD, PaxUSD? Where is LINK? What about the other MIP6 applications? I am in favor of more conservative parameters for USDC, but only once the peg is fixed.

4 Likes

For sure, this is the worst-case scenario. I came of age during the war on terror, however, and I have very little faith that regulators will follow the rule of law if, for example, darknet or terror funds are determined to be in the adapter. There’s no reason to believe that any warning will be provided, and an analysis of those odds is part of what needs to be factored into governance’s actuarial analysis.

5 Likes

I agree with you. We cannot count on any warnings before all USDC collateral is frozen (the Maker adapter blacklisted). However, the blacklisting will probably kill USDC as a DeFi asset so the probability of such blacklisting is rather low. The actions towards the ban on stablecoins in any form increase that probability.

The problem with the SF is that it is also a means to restore the peg. We got ourselves in the MCD mess before it was necessary and now we suffer.

1 Like

Good to think about worst case scenarios- but don’t kid yourself that the risk of this happening without a warning is higher than an ETH flash crash that would cost MKR holders much more. Even if the DAI adapter was frozen, the collateral would still retain its value -there would be no reason to auction it off except in a scenario where it was frozen without warning and regulators made it clear that innocent USDC holders would have no opportunity to retrieve their holdings and these regulator claims survived lawsuits from the innocent USDC holders.

1 Like

This could very well be, and that’s why it should be part of the risk calculation. The risk of immediate shutdown is not zero and it would be imprudent to factor it as such.

We should talk about this other point though. The risk of another ETH flash crash is probably much more likely, and is very much in the back of many of our minds. That’s why it’s so important to leave headroom in this adapter to provide quick liquidity for auction keepers. We noticed on Black Thursday that many keepers had difficulty sourcing Dai, which is why the USDC adapter was included. The idea was that participants could move highly liquid USDC from an exchange to a vault, where they would be able to generate Dai to be used for an auction. When the auction is complete, the keeper can sell the discounted collateral on the market and close the USDC position. A high stability fee incentivizes the keeper to close their position quickly. This option is not available if the collateral adapter is at capacity due to attempts to manage the peg. The USDC adapter is currently at 60% of capacity, which is very risky if it means that keepers will not be able to source enough Dai in the event of another ETH collapse.

USDC is a terrible tool for peg management. It may temporarily drop the price of Dai, but because there is no speculative value in holding USDC it creates immediate pressure to repurchase and repay the vault. It is at best a way to kick the can down the road when Dai is trading at a premium, and at worst, the overutilization of the collateral debt ceiling can create dangerous bottlenecks in the event of another precipitous decline in ETH. Governance should work to keep the adapter empty for the times when we really need that Dai. In my opinion, I think that the 20% stability fee that USDC launched at was too low by an order of magnitude.

8 Likes

The thing your missing here is that we don’t have any collateral that is uncorrelated with ETH price right now except USDC. When ETH is in a bear market and Dai is above the peg, the only collateral type that can restore the peg is something like USDC where you can mint Dai and sell it for peg arbitrage. I have been doing this as much as possible and will buy back the Dai once the peg is reached at $1. When the SF was about 4-6%, I estimated it would be a profitable arb as long as peg was reached in about 3 months. Why would you take away this only reliable mechanism to restore the peg when Dai is trading above it? Your solution of 20%SF would be a disaster right now and we could see Dai prices at $1.05-1.10, which could result in emergency shutdown. Also- it makes not sense that you want MKR holders to take the relatively larger risk of an ETH flash crash with 0% SF on ETH, but you don’t want them to take the much lower and more easily mitigated risk of government shutting down USDC. Once we restore the peg and find some uncorrelated collateral types, then we can talk about removing USDC, but for now we need it to maintain price stability for Dai. Sourcing liquidity for keepers is a secondary issue and if we need this again, we should raise the USDC ceiling.

2 Likes

How about moving 5 million DAI from the USDC-A DC to a new USDC-B collateral package with 20% SF to maintain availability for keepers? This was mentioned a while back and seems useful

8 Likes

Yes, exactly this. Governance should signal to:

  • cap USDC-A to something ($3MM - $5MM)
  • reduce the LR for USDC-A to 115% so Vault holders don’t get instant liquidated
  • turn on USDC-A liquidations
  • create a USDC-B for keeper liquidity
  • set DC to $20MM for USDC-B
  • set LR to 125% for USDC-B
  • set SF to 50% APY for USDC-B or more

Governance is ignoring the risk of a repeat of March 12th. Not only was USDC-A created to provide this liquidity, but we staged some emergency DEFCON spells as levers we could pull in just such an event. I’m glad everyone is excited to have USDC, but what we all voted in was not intended to be used this way.

Keepers need the emergency facility back, can someone please start a discussion or take action to signal for support? I defer to the interim risk team for exact values.

7 Likes

I’ve been advocating for roughly the same plan since at least 15 days ago.

Who really makes decisions around here? Who is this mythical governance?

3 Likes

Yes, I think we are all in agreement for the dual USDC collateral types. We’ll try to put out a proposal in the next couple weeks.

3 Likes

I forgot to mention the lot size for USDC is 50k. Whatever risk comes up with is fine, but we would get more participation with 10k, and there should be minimal risk with the longer ttl.

3 Likes

Another big concern I have is the OSM risk. Suppose we have an unused 20mil debt ceiling with a 50% stability fee. If something bad were to ever happen to USDC, that would be an extra 20mil hole to be plugged. There needs to be a way to be able to turn the debt ceiling on/off when needed (in the midst of massive drop in ETH price), and it needs to happen a) outside of MKR governance, and b) outside of the GSM. Are there any workarounds for this @cmooney @BrianMcMakerDAO

2 Likes

If we had an emergency facility for USDC, we could create an instant access module that would do the DC increase without a governance action, but also rate limit its growth. Perhaps even scale the SF with the DC increase. We’ve only been playing around with instant access modules so far, with scaling the DC for ETH or other standard ilks being our original application. Perhaps we do that, then once we test/audit it, add the same sort of thing for the USDC emergency facility.

This is absolutely doable in the medium term, but we still need to think about the short-term. I agree with you, a 20MM hole would be extremely undesirable to patch with MKR.

3 Likes

Build us an IAM and let’s get it done!

1 Like