Substantial unrecognized risk with cDai collateral proposal, please read and vote!

I am writing about an ongoing poll to include cDai as collateral. So far only 11 people have voted on it. You can see a well written proposal from SebVentures at [cDAI] MIP6 Collateral Onboarding Application

I think this collateral would carry substantial risk for Maker, most of which was not recognized in the original proposal. I hope I can convince you to vote not to prioritize this collateral. This argument will also be applicable for other interest bearing DAI account tokens (such as aDai), but I’ll focus on cDai here, since that is the one most at risk of being accepted.


At first glance, this seems like an excellent form of collateral. In the short term and in some ways in the long term, this would give us greater control over the DAI interest rate. By setting an interest rate on this collateral type that is lower than the rate found on Compound, we would immediately see the interest rate on Compound drop to the interest rate we set, since an arbitrager would borrow DAI to get cDai to lock until the interest rates matched. This would likely lower the price of DAI helping us to restore the peg and is the most compelling reason in my mind to go forward with the proposal. This reason was listed in the original proposal.

If cDai is accepted as acollateral, I’ll assume it is accepted with a lower interest rate than the rate seen on Compound, since if that is not the case, the addition should have no affect at all. Thus, I’ll be assuming that it will carry the benefits outlined above and the risks outlined below.

Some might also hope that they could profit from this arbitrage (that Maker and Compound would collectively pay for). Unfortunately, that arbitrage will probably get paid out immediately to someone with a single transaction and a flash loan and probably not to the person that writes that transaction, but to a frontrunner looking at items in the mempool. (But these details are not too important.)


There are a number of substantial risks associated with cDai collateral. As stated earlier, cDai will either be extremely popular (to arbitrage the interest rate) or not used at all. If it is extremely popular and sees use, there would be a number of downsides. Most importantly, lenders would naturally prefer to use Compound to borrow DAI than to use Oasis. This is because Compound would have the same interest rate and it provides incentives on top, including interest on collateral. That means cDai collateral would actually cannibalize other forms of collateral, including ETH, which most people in the community would agree to be a more valuable form of collateral than cDai. cDai could cannibalize literally any form of collateral that both Maker and Compound accept. This would mean that collateral would get centralized in cDai and Maker would be essentially selling its ability to choose collateral types and debt ceilings to Compound, since cDai is backed by whatever Compound is willing to borrow. This would add substantial risk to Maker if anything ever happened to Compound. These risks would obviously be mitigated by whatever debt ceiling we imposed on cDai, but that doesn’t really address the problem. Making something less bad doesn’t make it not bad. Wherever we set the debt ceiling would simply determine how much wealth we were transferring from Maker to whoever front-runs the transaction(s) to arbitrage Maker and Compound.

In addition, though a lesser concern, coupling ourselves to Compound like this would give some of the DAI policy control over to Compound. For example, shocks to DAI interest rates and supply could occur as Compound changes their incentive structures. This would give Maker less control over the peg in the longer term.

Finally, increasing our coupling with Compound would increase the chance of both protocol risk (as was acknowledged in the original proposal) and the risk of finding ourselves in the middle of a flash loan exploit (like what happened to Fulcrum).


For these reasons (mainly the first one), I think substantial risks to Maker went unnoticed and we should avoid adding cDai (and similar tokens) as collateral for DAI.

Please vote against cDai being prioritized as a collateral here.

4 Likes

I’d also be interested in getting SebVentures thoughts, since I think he’s fairly reasonable and I agree with him on various issues, such as it not being too great to have DAI be backed mainly by USDC.

Your are right in your analysis. cDAI is an extremely powerful collateral. I depicted it as leverage so in a sense, yes, you can achieve the same effect with less of other collateral. It is indeed taking a protocol risk.

The front-running of arbitrage trade (and leaving the mess to Maker) was one of my concerns for setting USDC-A at 101%. Hence my proposal for USDC-M. Community seems fine with that (arbitrageurs that front-run everyone as it’s a fair market).

Your point work well if, and only if, LR is 100%. It is less so if LR is 105%, 110% or 120%. For rates we can start at 5% and lower it down to 2%. If borrow on Compound isn’t enough (don’t increase), it will soon limit the trade. Maker can control it’s cDAI appetite.

Like always it’s a trade-off between custody risk (and almost no fees) for stablecoins and a protocol risk (with fees we will be able to enforce) for cDAI/aDAI.

I want to highlight that I don’t see ETH as a great collateral neither due to the black thursday event. We need to diversify (but not USDT please).

Let me know if there is a mistake in my reasoning.

1 Like

You are raising good points, but I think the situation is not as dramatic as you described. For instance, borrowing DAI at Compound has been much cheaper than at Maker for months and we haven’t seen that much refinancing. For instance, borrowing 100k DAI at Compound by using 500 ETH as collateral costs you -3%. So you actually get rewarded for borrowing DAI there already. And there is about 300m unutilized DAI available for borrowing now.

I do agree though there is a breaking point somewhere where Maker users wouldn’t tolerate it anymore and would instead just move to secondary lenders. But Maker controls everything as Sebastian pointed out, from SF, LR to DC and we get to decide how much of a liquidity and at what price we want to provide to secondary lenders. It basically means Maker participates in secondary lending activity and takes a cut of it. Rune described it pretty well in his tweet.

I do think that we should equally focus on cUSDC as collateral currently, because it helps with the peg more and also related to some of the potential issues you mentioned. If we can make cUSDC as collateral that also collects COMP rewards, people would switch from DAI recycling to USDC recycling on Compound by using Maker to leverage and have a massive effect on the peg.

I also wouldn’t like to see DAI borrowers on ETH leaving to secondary lenders in large amounts at this stage because we made the difference on borrowing it at secondary lenders versus Maker more than 5% for instance. Even though we can catch some of that yield through cDAI SF we would be incurring a lot of additional protocol risk that is harder to evaluate. But I don’t see it would be risky to implement it, we might want to be just a bit cautious at setting the right parameters.

2 Likes

Oh, interesting proposal re USDC-M.

I think the point about flash loans is less applicable with greater LR. However, flash loans aren’t a big concern here. I was considering not mentioning them at all. I mainly mentioned them to note that if you wanted cDai collateral not because you thought it was good for Maker, but because you thought you could personally benefit from arbitrage, that this was unlikely to be the case.

As mentioned, I agree with your point about rates. The arbitrage will either be profitable or it won’t and we’ll either hit our debt ceiling or have no interest. In the latter case, the proposal isn’t too interesting. In the former case, our appetite would mainly be controlled by the debt ceiling. However, as I mentioned even with a low debt ceiling, we’re just replacing good collateral with worse collateral, we why have any appetite for this at all?

A fair point, it’s always important to note what you’re trading off against. If you’re arguing for this against USDC, it does look better, but ideally we’d have neither.

Ok, I think we disagree on ETH. I think diversification is important and I wouldn’t be interested in 100% ETH, but compared to some other forms of collateral USDC seems great.

I think we’re talking past each other a bit regarding the LR. I’m still concerned that this plan would involve Maker subsidizing arbitragers, taking on worse collateral and introducing more risk. I think most of our difference stems from a previously unstated difference in how highly we prioritize the peg vs these risks and how willing we’d be to accept the peg being less stable in the short run for better collateral. Does that sound right to you?

1 Like

That’s a good point. I think arbitraging this cDai vault would be very fast, but closing out Maker vaults and migrating to Compound will be much slower. I definitely agree with you that it shouldn’t lead to a collapse in our ETH vault all at once. I also agree that perhaps a lot of the damage has already been done, with most DAI borrowing taking place outside of Oasis.

Interesting. That was well written.

I’m not sure what you’re saying here.


Thank you both for your responses!

Overall, I had 2 classes of concerns here:

  • This collateral type involves Maker giving away a lot of control to Compound and will centralize our risk. I still think this is the case (though I’ve read some good points and am somewhat less concerned than before).
  • The community has a poor understanding of the risks involved and has a significant chance of setting parameters that would quickly have very bad effects on Maker. After reading the responses here, I no longer thinks this is the case. So while I still think this collateral type is a bad move at the moment, I’m much less concerned overall.
3 Likes