- Types of Risks
- COMP Farming Simulation
- Proposed Risk Parameters
- Other Considerations
From MIP30 Summary:
“COMP farming offers an attractive yield on USDC with very little risk, and the ability to receive additional leverage from a CDP is likely to present an attractive opportunity for a yield-seeking investor, so this collateral type could be expected to produce very high dai issuance, while generating significantly higher fees for MakerDAO than what is currently collected on MakerDAO’s significant stablecoin exposure. We propose a new USDC based collateral adapter that performs COMP farming on behalf of depositors.”
In essence this MIP30 Farmable cUSDC Adapter proposes a solution for levered COMP farming using USDC as the underlying supplied asset to Compound which is then levered to maximize yield. Leverage is built at Compound (up to 4x), but Vault users could additionally boost their yield by leveraging this collateral position at MakerDAO (depending on liquidation ratio set by governance).
We understand this adapter is general enough to be used at other protocols in order to let LPs of other protocols collateralize their LP position at Maker, borrow DAI and at the same time collect farming rewards. This is not possible in current implementation of UNI LP collateral that Maker supports, although Uniswap rewards are currently still disabled.
We believe this type of adapter will be very beneficial for Maker from a business perspective but risks need to be properly analyzed and parameters carefully chosen since collateral asset behaves differently than what Maker is used to.
Types of Risks
By implementing this cUSDC adapter, Maker will have direct exposure on Compound’s supply and borrow side of USDC market. Because of Compound’s recycling/leverage feature on the same asset, a nominal position maintained at Compound will be in majority hedged, but initially supplied USDC to Compound carries solvency risk among other risks.
Solvency risk is high if borrowed USDC can not be fully repaid during liquidation events. This means that a risk assessment of USDC loan book portfolio and its collateral needs to be performed to assess likelihood of liquidation events.
Analyzing top USDC borrowers on Compound reveals that most of USDC outstanding borrow positions are collateralized by USDC. In other words, the current USDC borrowers at Compound are already pursuing this “recycle” strategy to maximize their COMP yield strategy. This confirms there is definitely demand for usage of Maker enabled cUSDC CropJoin Vaults.
Additionally it also tells us that solvency risks are lower than we would expect. Since the majority of USDC loans at Compound are collateralized by cUSDC ($1bn out of $1.2bn) rather than volatile crypto assets, liquidations and loss events are less likely to happen. Additionally, if a loss event does occur, loss is spread amongst the large USDC supply market, making a haircut smaller in proportional terms.
Finally, USDC at Compound has a fixed oracle of $1, which means that a scenario witnessed on the DAI market a month ago could be avoided.
In our opinion solvency risk carries low to medium risk. This risk can be mitigated by setting debt ceiling limits for cUSDC CropJoin.
As written in MIP30: “Even if the Compound system is solvent, there is no theoretical guarantee that it is possible at any time to withdraw a supplied asset, since the reserves of the supplied asset may be tied up in outstanding borrows.”
Compound has been historically successful at managing liquidity with the technique where rates start increasing very fast after 80% borrow utilization is reached (4% rate at 80% USDC borrow utilization and 30% rate at 100% USDC borrow utilization).
Yet, “run on bank” is theoretically possible and vault users may not be able to withdraw their collateral from Compound, although users could in theory trade cUSDC for cDAI on Curve and withdraw capital. Still, such scenario could create a problem for vaults that maintain a very low collateralization ratio at Maker which would need to unwind exactly during “run on bank” events to avoid liquidation. However, the proposal assumes liquidations are disabled and oracle price is fixed to 1$ having a similar logic as USDC-A.
In our opinion liquidity risk carries low risk. This risk can be mitigated by setting debt ceiling limits for cUSDC CropJoin.
Recycled cUSDC position carries counterparty risks of USDC, which already represents 27% of Maker’s collateral. Community wants to minimize the backing of DAI with USDC and cUSDC MIP30 proposal doesn’t achieve this.
However and importantly, increased usage of cUSDC will have a negative impact on DAI price which means that current stablecoin vault positions could be unwind easier. In that sense, we might achieve a swap from USDC or other stablecoin vaults to cUSDC, but importantly the interest rate collected will be much higher and risks compensated better.
Finally, having USDC deposited at Compound does protect Maker from blacklisting risks.
In our opinion counterparty risk carries low risk. This risk can be mitigated by setting debt ceiling limits for cUSDC CropJoin.
MIP30 proposes disabled liquidations and we may assume that we Maker could end up in a similar scenario as with USDC-A vaults, where vaults could get abandoned. However this mostly depends on how high liquidation ratio is set for these kinds of vaults. Users will be able to collect COMP rewards without closing their position, but the remaining overcollateralized capital will remain locked. If interest rate accrual lowers collateralization ratio below 100%, users will not have incentive to unwind such vaults and Maker would need to manually liquidate them. This means that also stability fee accrual plays an important role here.
If Maker would be forced to manually liquidate these positions, this creates a problem since a lot of coordination and potentially even adjustments are needed under the current liquidation system. We may end up in a scenario as with USDC-A vaults where we needed to disable fees because a large amount of vaults would end up underwater. Ideally if and when this issue becomes a problem, Maker would already have Liquidations 2.0 readily available and controlled liquidation could be performed much easier here.
In our opinion abandoned vaults risk is high if liquidation ratio and stability fee parameters are not selected carefully.
Compound Governance / Integration Risks
Just recently Compound’s governance changed the “speed” of COMP rewards that are distributed to particular markets. This creates a problem for Maker’s governance that might need to react before their upcoming proposed changes are effective.
For instance, if a large amount of cUSDC is deposited by Maker to Compound and Compound’s governance potentially lowers COMP rewards to the USDC market, yield for Maker cUSDC vaults will fall. If this yield falls below SF being charged by Maker, vaults would unwind instantly and lead to unintended events. This means Maker’s governance always needs to keep an eye on dynamics at Compound and react before any changes that might hurt Maker’s cUSDC recycled positions.
Other parameter changes at Compound that Maker needs to monitor:
- Rates Curve
- Collateral Factor
- Reserve Factor
- COMP vesting
To our knowledge, Compound’s governance usually needs at least a week before proposal is discussed on forums and being effective on-chain, so this should give Maker enough time to react and potentially change parameters if needed (mostly SF).
In our opinion Governance risks are medium.
COMP Farming Simulation using cUSDC
Link to simulations can be found here.
Below are simulated expected yields for Maker vault owner after SF deduction and includes COMP yield (incl. Compound’s net interest rate). Leverage is maximized based on simulated Maker’s liquidation ratio and Compound’s Collateral Factor used for recycling (CF).
Proposed Risk Parameters
- Liquidation Ratio:
Proposed liquidation ratio of 110% represents 11x leverage that could be maintained by users. Note that liquidations are disabled and collateral price is stable, therefore users will all the time be able to maintain this high leverage. From a risk standpoint, low LR for stable collateral shouldn’t be an issue, but it still needs to be sufficiently above 100% so that 1) vault owners have incentive to close it out (abandoned vaults problem) and 2) any fees can be collected by Maker at all.
If stability fee on average will be lower than 10%, Maker should have at least a year before rates start accruing below 100% collateralization ratio forcing governance to disable them (past stablecoin vaults case). On the other hand 110% LR will be very tempting for users as COMP yield will be very high. It is likely that few users would dominate this vault type, especially if we roll out a debt ceiling increase in steps.
- Stability Fee:
Stability Fee should always be lower than net yield from COMP farming. If it is equal or higher, vault owners would unwind their position instantly, causing some intended dynamics on DAI markets. This is why cUSDC COMP farming yield needs to be simulated by conservatively assuming few equilibrium states that might lower farming yield. In the last few months cUSDC recycling yield was between 15% and 20%. Currently it is around 24% due to COMP price increase (at max CF of 0.75).
If governance votes for 7% SF and 0.7 CF, this would currently represent about 11.6% APY on a single unleveraged position at Maker. Levering this up to 11x, 127.6% APY yield can be achieved.
However as explained, Compound governance could reduce COMP rewards to cUSDC market in the future. Furthermore, once Maker starts increasing cUSDC exposure at Compound, COMP yield will fall as it gets shared between larger pools of capital. We also have to keep in mind that the COMP price is currently very high and we are in the middle of a bull market.
We do think that 7% SF would currently still be low enough for users to generate high yield and attract them. We should always aim users to make between 5% and 10% spread on unlevered vault positions and at 7% SF we are confident this can be achieved. If the situation starts becoming worse, governance will need to react fast and decrease SF. We wouldn’t want vault owners to close out their positions instantly and come back in a week because this would increase DAI price volatility. Especially in case debt exposure towards this asset is high.
- Debt Ceiling:
No debt ceiling for MIP30 was proposed, but we believe we should treat it similarly as we treated PSM-USDC-A or UNI-LP-ETHDAI and increase the debt ceiling from 3m in steps. This adapter is complex and it needs to be tested before larger scale exposure can be achieved.
The maximum debt ceiling depends on governance risk appetite. If the idea is to reduce stablecoin vault exposure and increase more yielding cUSDC exposure, the debt ceiling could eventually be lifted towards the 100m range or more (depending also on general community’s acceptance of increased Compound’s platform risk). However at those DC levels and including leverage, total expected yields starts falling from 123% to 83%.
- Collateral Factor:
When a user levers USDC at Compound, net negative rate can potentially be paid. If users’ supplied cUSDC position size is 133% of borrowed USDC size (using max. collateral factor of 0.75) but borrow rate at Compound is more than 133% of supply rate, your position accrues negative rates and vault can be liquidated once 0.75 CF is reached (in this case instantly).
Borrow to supply rate ratio at Compound can reach a ratio of 5x or higher at lower utilization levels. However the net rate accrued is smaller because rates are nominally lower. Also keep in mind that rate accrual is a slow process and governance will always have enough time to react.
Still, we may want to lower the Collateral Factor below the maximum value of 0.75 and keep it in the bands between 90% and 93% of maximum value, which is from 0.675 to 0.7 . This will enable between 3.1x and 3.3x leverage at Compound and currently about 18% APY on single unlevered position at Maker before SF. Net interest accrual would also need to be unlikely high to increase CF from 0.7 to 0.75 in short time, a 7% increase from accrued interests on borrowed value would need years.
Having liquidations disabled simplifies the vault since the position in essence behaves similarly as USDC, with additional platform risk from Compound. Both USDC and Compound risks are very binary and in such cases fixed oracle and liquidations disabled is preferred.
Proposed Risk Parameters:
Stability Fee = 7%
Liquidation Ratio = 110%
Starting Debt Ceiling = 3m
Eventual Debt Ceiling = 100m (depends on community’s risk appetite)
Collateral Factor = min 0.675, max 0.7
Auction parameters = none, liquidations disabled
- More USDC on Compound - cannibalizing MakerDAO market (not as significant because of recycling feature)
- More integration risks - only Uniswap LP tokens currently represent such asset at Maker
- A lot of monitoring of compound governance needed - we may expect reaction from Compound
- Current Compound’s proposal for COMP cooldown period may affect how adapter claims rewards
- Calculated final APY for vault users can be achieved only if the user knows how to fully leverage via Maker or if a tool for this exists. Otherwise yields will be lower, around 10% or less.
- We may expect few whales to utilize this vault and fill the available DC. It is also possible that current USDC recyclers on Compound migrate to cUSDC CropJoin Vault.
Lead Researchers: Primož Kordež & Marko Štemberger