Due to the on-going liquidity situation and Dai price instability, we need to make a quick decision on risk parameters such that the executive can go on-chain as quickly as possible. We are targeting getting the executive vote on-chain as soon as the GSM delay is changed to 4 hours. The contents of said executive vote will be influenced by the outcome of these polls.
First, the Interim Risk Team presents a note on oracles and liquidation given the nature of USDC as collateral. Then we follow with a brief summary of the tradeoffs for each of the major risk parameters (Stability Fee, Debt Ceiling and Liquidation Ratio) with respect to the emergency onboarding of USDC collateral as part of the effort to increase Dai liquidity.
Following a description of the tradeoffs, the Interim Risk Team presents a number of polls that will be used to gather community sentiment and feed into the risk parameters included in the executive vote. Initially, the plan was to present combined risk parameter packages, however, after consideration, separate polls for each of the key risk parameters were determined to be more appropriate.
The main goals that MakerDAO should aim to achieve with the introduction of USDC as an emergency collateral type are as follows:
- Increase Dai liquidity specifically for keepers taking part in both FLIP and FLOP auctions.
- Increase Dai liquidity such that the Dai peg moves closer towards $1.
Oracles and Liquidation
Due to the nature of the collateral and the time constraints, it is proposed that the oracle price of USDC be set to $1 for the purposes of this Vault. The governance community can consider migrating to another oracle solution in the future.
Under normal circumstances setting the oracle price to $1 would still allow liquidations due to accrual of stability fees. However, the collateral type will launch with the new “collateral liquidation freeze” component switched on. Vault users will not be subject to any liquidations in the short term, even with the accrual of stability fees. Please note, however, that Vault users may be at risk when the liquidation freeze is removed in the future. As always in a non-custodial, permissionless and decentralized system, users alone are responsible for their Vaults.
There are three main drivers to consider for the USDC debt ceiling. The primary use case is to provide keepers with access to Dai liquidity should they need it to participate in ETH
flip auctions. Based on an analysis of the current structure of the Vault portfolio, if Eth were to fall into the $60-$80 range, the set of keepers would collectively need to raise ~20 million Dai amongst themselves.
The secondary use case is to provide liquidity for Dai markets, where the Dai price has been trading as high as $1.10. A high debt ceiling would allow market makers to absorb the excess Dai demand seen on centralized exchanges as well as on supply shortages seen on secondary lending platforms such as Compound and dYdX.
Thirdly, the community may want to allow additional debt ceiling room to facilitate DeFi users to refinance their outstanding positions from secondaries back to Maker Vaults. The benefit of this is that Maker governance would be able to more directly influence the liquidity situation for the broader ecosystem.
A final consideration for debt ceiling is the risk of “blacklisting” by the issuer of USDC or another entity in control of this function. If such a situation were to arise, MKR holders could conceivably need to mint additional MKR via the debt auction in order to cover any potential shortfall from a blacklisting event.
Depending on how aggressively the community wants to address these three issues, the debt ceiling could conceivably range anywhere from $10 million to $40 million. Please keep in mind that a more conservative or more aggressive debt ceiling can always be increased / decreased through governance (subject to the 4 hour GSM delay). This will allow the Interim Risk Team and MKR Holders to reevaluate a highly evolving situation, at a tradeoff of governance overhead.
Liquidation ratios are typically expected to be fairly low for fiat-backed stablecoins due to their low volatility. However, in this particular instance, governance is targeting highly specific use cases, and the community may be susceptible to a rationing risk. For example, a 102% LR allows for 50x recursive leverage. This means that a user with only 1 million USDC could deliberately soak up the entire debt ceiling through recursive leverage. Spreading the USDC over multiple users should mean that on average more of it will go towards the intended use-cases.
Similarly, too high of a liquidation ratio (such as the 150% LR for ETH) is a potentially inefficient use of collateral. As a comparison, secondary lending platforms typically target a 125%-135% range. As a result, a 125% LR is suggested.
The stability fee should be set with the community’s specific use cases in mind (auction keeper liquidity and Dai peg management). The primary consideration is that the stability fee should be set high enough so that the limited debt ceiling is not utilized for any alternative use cases that don’t alleviate keeper liquidity or the Dai peg. Too low of a stability fee could cause just a few Vault owners to utilize the entire debt ceiling. Therefore, a high stability fee can price out these Vault owners.
Another consideration is secondary lending platforms, which currently carry a 17%-45% lending rate for Dai due to the high utilization rate. If the stability fee is below these amounts, then a significant amount of Dai will be generated to immediately lend to these other platforms. This is not necessarily undesirable, but it is not one of our primary goals, and the debt ceiling could be hit very quickly.
At the same time, too high of a stability fee may cause fewer Vault owners to take advantage of this collateral type, reducing the effectiveness of the collateral type.
Due to the competing dynamics, governance should be vigilant in adjusting the stability fee appropriately based on changing market environments. The community should be aware, however, that it will be easier to start high and lower the stability rather than the opposite.
Liquidations would be disabled at the launch of the collateral type. However, the following are proposed auction parameters subject to the removal of the liquidation freeze.
The liquidation penalty is suggested to remain at 13% for similar reasoning as other Vault types. The liquidation penalty cannot be too low due to the auction grinding attack vector.
Auction Lot Size
An auction lot size of 50,000 USDC is proposed for similar reasoning as other Vault types. Lot sizes that are too low cause significant friction with respect to network congestion and usage. Lot sizes that are too high price out many individual smaller keepers. To strike a balance between network congestion and participation, a 50,000 USDC lot size is suggested.
Auction Bid Duration
A bid duration of six hours is suggested for similar reasoning as other Vault types. Too low of a bid duration can be dangerous due to potential network congestion issues. Too long of a bid duration means the keeper will incur additional price volatility risk
Minimum Bid Increment
A bid increment of 3% is suggested for similar reasoning as other Vault types.
Maximum Auction Duration
A maximum auction duration of 3 days is suggested since liquidations might occur under uncertain circumstances. For example, if the market price of USDC is trading at $0.50 (yet the fixed oracle is still displaying $1), then Vault owners are incentivized to max out their Dai generation, which would lead to mass liquidations soon after. In such a scenario, the community might want to keep auctions running for several days in order to facilitate the liquidation process.
A dust limit of 20 is suggested for similar reasoning as other Vault types. This decreases the risk of spam Vaults being created.