Research: cTokens as Collateral

cToken Introduction

  • cTokens represent tokenized deposits supplied to Compound and accrue interest via an increase in the exchange rate of the cToken relative to the underlying asset.
  • If a user has an outstanding borrow against the cToken collateral, then they are unable to send the tokens (they are encumbered by Compound).
  • Full cToken documentation is available here.
  • cDAI: a tokenized representation / claim of DAI deposits in Compound
  • cUSDC: a tokenized representation / claim of USDC deposits in Compound

Goal: MakerDAO accepts cTokens such as cDAI or cUSDC as collateral given their market caps and desirability as farming assets on Compound. Ideally, Maker can align incentives with users to achieve the following:

  1. a significant amount of Dai is generated (thus improving the Dai price)
  2. an appropriate stability fee is charged, generating revenue for the protocol
  3. farmers earn higher yields, driving TVL back to MakerDAO

Takeaway: cUSDC is much more likely to achieve the above goals than cDai, although the latter may still have some benefits for overall Dai liquidity.

cDAI:

  • First, we differentiate between two classes of cDai holders on Compound: ‘single dippers’ and ‘recyclers.’ Single dippers have merely deposited Dai into Compound, and have no outstanding borrows against the Dai deposit. Recyclers, on the other hand, are typically at max borrow against their cDAI (supply, borrow, resupply and so on).
  • Recyclers’ cDai is encumbered by Compound (due to the borrow) and are thus unable to utilize Maker vaults. Currently, only single dippers can utilize Maker vaults. However, these users are probably unaware of recent DeFi developments and are unlikely to utilize Maker vaults (otherwise they may as well have been recycling on Compound all this time). The fact that they are currently leaving money on the table implies that they will not take advantage of any opportunities MakerDAO has to offer.
  • The critical analysis revolves around Maker’s ability to entice recyclers away from Compound into a Maker vault. Technically this is possible, but difficult to analyze.
    • Currently, Compound can offer Dai holders 4x leverage, and farmers will earn COMP from both the supply side and borrow side of Compound markets.
    • A farmer using Maker would earn COMP only from the supply side (as the borrow has now shifted away from Compound into Maker). This essentially converts recyclers into single dippers.
    • The only way to compensate farmers for the loss of COMP from the borrow side is to allow sufficiently high leverage such that single dipping COMP (while using Maker vaults to recursively borrow) is more profitable.
      • Example: Currently, 100 Dai on Compound can be transformed into 400 Dai supplied (4x leverage) and 300 Dai borrowed. This package earns COMP from the 400 supply and the 300 borrow for an all-in yield of 57%.
      • If Maker accepts cDai at 120% CR, then a farmer can transform 100 Dai into 600 Dai supplied and 0 Dai borrowed on Compound. This will roughly achieve a breakeven with the original strategy.
    • However, the analysis breaks down due to an unstable equilibrium. If a significant number of farmers switch from Compound to Maker, then this skews the COMP rewards and ultimate yield downwards. Additionally, for example, Compound governance is currently debating whether to lower COMP rewards or not across all assets.
  • cDai Vaults will improve Dai liquidity (due to all the additional Dai being deposited on Compound), but it is unlikely to directly affect the Dai price in the short run, since none of the new Dai being generated is actually being sold. Improved liquidity on Compound (in theory) will slowly migrate to other protocols (and improve the price), but it is not direct.
  • Risks:
    • cDAI is essentially a claim on all collateral assets at Compound.
    • Compound doesn’t have a capital solvency feature like Maker. Losses are socialized by depositors.
    • In order to properly assess a risk premium on cDAI, one would need to examine Compound’s general credit risk on the cDAI market. This includes an analysis of current exposure and risks associated with liquidations and auctions.
    • For this strategy to work we need to have a low LR, which means we are more exposed to credit loss events at Compound.
    • We may inadvertently increase systemic risks as many protocols (YFI for example) could start utilizing cDAI Maker recycling as a strategy. Those could automatically switch to cDAI recycling via Maker and our indirect portfolio exposure to Compound could increase significantly. Proper debt ceilings should be applied, as this Vault is likely to fill up quickly.
  • Positive: Potential to earn high fees, easiest way to flood Compound with DAI to lower COMP rewards, improve Dai liquidity on secondary markets (liquidity could flow to Curve, for example)

cUSDC:

  • cUSDC is currently not an optimal strategy by yield farmers on Compound, since the overhang of USDC supply is so high (i.e., utilization rate is low). USDC is less attractive than Dai because it is difficult to artificially increase the utilization rate (difference between borrow and supply is $100 million, and whales can easily come in and supply more USDC (i.e., there is no shortage of USDC).
  • Enticing USDC recyclers is easier since the APY is lower. A 120% CR should achieve this (allowing 6x leverage).
  • Key Point: Whereas cDai merely only improved Dai liquidity but did not directly improve the Dai price, the situation is completely different for USDC. A farmer who deposits cUSDC into a Maker vault, and borrows Dai, needs to sell Dai for USDC (on Curve or other liquid venue), to then deposit USDC back into Compound, to then re-borrow from Maker. This will provide enormous downward pressure on the Dai price.
  • Again, however, the increased supply numbers on Compound would skew the farming APY for COMP rewards, so it is a bit difficult to predict what type of leverage Maker would need to accommodate to make this option attractive for farmers…
  • Risks:
    • Pretty much the same as described above for cDAI.
  • Positive: Potential to earn high fees and simultaneously fix the peg.

Process of building Compound-Maker COMP farm position using cUSDC ~105% LR Vault

1.1 Deposit 100 USDC in Compound and receive 4728.5 cUSDC

1.2 Deposit 4728.5 cUSDC in Maker and mint 95 DAI

1.3 Sell 95 DAI for 95 USDC

2.1 Deposit 95 USDC for 4492.156779 cUSDC

2.2 Deposit 4492.156779 cUSDC in Maker and mint 90.25 DAI

2.3 Sell 90.25 DAI for 90.25 USDC ->

  • These steps would be repeated many times over, or by utilizing flash loans or some external capital, it is possible to build ~2000 USDC supplied in Compound based on the initial 100 USDC owned. Below is the same process illustrated by corresponding state changes in Maker and Compound.

Capital changes in Compound and Maker during cUSDC Vault, COMP USDC supply farm buildup

cUSDC Vault Debt Ceiling effect on the state of capital in Compound markets and corresponding interest rates and COMP APYs

  • Analysis and simulations below assume no changes other than direct impact of cUSDC Maker vault on supply of USDC in Compound.
  • Due to high leverage achievable by LR in Maker, the final Compound yield for this strategy would be more profitable than regular USDC-USDC recycling. This could potentially trigger some capital reallocation within Compound and also wider DeFi.
  • As farmers will utilize this new strategy, the supplied balance of USDC in Compound will increase (the increase in USDC supply in Compound should be roughly proportional to the DC on cUSDC Vault). Increase in USDC supply leads to decreased COMP rewards for suppliers in the USDC market and since it effectively lowers the pool utilization, it also decreases both the supply and borrow interest rates.
  • The first table below shows how the IR and COMP rewards per borrow and supply market change, based on the effective increase in the USDC supply market on Compound. The supply increase is correspondent to cUSDC DC.
  • The second table below shows how these changes in rates affect the yield for multiple strategies, DAI-DAI recycle on Compound, USDC-USDC recycle on Compound and finally cUSDC Vault strategy based on several different collateral ratios.

COMP and IR APYs changes due to inflated USDC supply in Compound as a result of a sUSDC Vault

COMP yield farming simulations based on different LRs and DCs. Red numbers yield less than regular USDC-USDC recycled farming

  • As we can see, the 125% LR on cUSDC Vault already renders this farm less profitable than regular USDC-USDC recycling even at the current state of capital markets in Compound.
  • Dynamics determining relative profitability of this farm, which also determines how large of a DC we can afford before the farm is rendered unprofitable is as follows; as we increase the DC, we lower rates on USDC supply in Compound, thus, we must also lower the LR in order for vaults to achieve higher leverage and effective yield.
  • With LR of 118%, we could generate ~20m dai, 115%; ~50m dai, 110%; ~100m dai and 105%; 250m dai.
  • Above calculations can be seen 1) Simulations, Steps, and 2) Compound Rates
  • The analysis does not assume other yielding strategies, for example USDC-A Vault and DAI-DAI recycle on Compound, which yields ~52%, or just simply depositing USDC in one of the Curve pools, where above 100% APY can be achieved. Effectively cUSDC LR would need to be 110% or lower for this to be viable. At 110% LR we could generate 50m DAI, where the APY is only slightly higher at ~54% compared to USDC-A DAI-DAI farm.
  • There are other potential tokens which can have a greater effect on the DAI price management; mainly LP tokens of several AMMs with dai. For example Curve pools, which currently play a big role in determining the price of dai.

Lead Researchers: Primož Kordež & Marko Štemberger

6 Likes

Nice work!

Just linking related topics here:

I’ve submitted cDAI as aDai might change soon in Aave 2.0. I think the knowledge we will get from cDAI will inform us on the interest of having cUSDC. Also USDC is not the most liked collateral due to the custody risk.

My biggest problem with the cTokens was lack of direct market liquidity cDAI->DAI or cDAI ->Otherstuff

The second problem is that unless something changed in compound after I stopped using them that any facility can have withdrawal liquidity issues. Now compound may be building a nice reserve to buffer withdrawals but the fact a facility can end up very close to 100% utilization means that buying cDAI say at auction a keeper may find there is no liquidity to redeem cDAI for DAI on compound to recycle DAI. If the cToken markets are thin there may not be enough market liqudity to do DAI cycling without serious slippage.

Willing to be corrected as I no longer follow compound as closely as I used to and am for the most part unfamiliar with all the different products popping up. BTW: Generally speaking the more of these assets that pop up the more they are trying to claim similar pools of real liquidity. I am aware they also ‘provide’ liquidity but when we start coupling these representations of liquidity deposits into collateral for borrows we really have to pay keen mind to the additional systemic constraints to actual underlying liquidity in these markets when they get stressed.

The key issue here isn’t whether we can, but whether it is healthy for these markets and whether we should add these assets. Right now based on my possibly outdated info for cTokens I am generally against adding these mostly due to the liquidity issues. Uniswap has no such issues, neither does Maker as the liquidity in these facilities is not re-loaned.

2 Likes

@MakerMan Well, the liquidity issues is one of the reasons to actually add cDAI. Maker can act as the powerful role as a primary issuer, to provide DAI liquidity during liquidity crunches.

Also even if utilization is at 100%, you can still swap some ctokens on curve.

And cDAI shouldn’t be liquidated because of a price drop. Immediate liquidity is therefore less important.

1 Like