PreMIP discussion: Vaults with LR << CR

Maker Improvement Proposals Proposal Ideas
#1

We are struggling with keeping up with the DAI demand and trading off the risks for some time already. I guess we can all agree that driving stability fees to 20% and increasing the stability buffer to infinity is not a sustainable way, so I was trying to find a possible help on that topic for some time and I would like to bounce around this idea with all of you.

In case of a sharp dip, our biggest risk is that too much DAI needs to get flipped because vault-owners don’t pay enough attention and cannot deleverage quick enough. What if we could offer a safety net for those who want to stay heavily overcollateralized:

We could offer a vault-type with a quite high CR (let’s say 300%) but keep the LR a lot lower (for example 150%) so vault-owners would not be able to mint DAI so close to getting liquidated on the next (small) dip.

I guess a lot of the risks by liquidations can be mitigated (if LR is 2x the CR, we would need a 50% dip for the vaults going underwater), those vaults can benefit from a much lower stability as well.

I don’t know what that would imply smart contract wise, maybe this can be handled in the flipper, maybe a new JoinAdapter?

WDYT?

7 Likes
Weekly MIPs Update #26
#2

Yes, AAVE has a different LTV and LR.

I’m guessing that it’s not easily implementable for MakerDAO.

1 Like
Weekly MIPs Update #27
#3

This could be added as an additional ilk parameter in the cat (soon to be dog). CR could be set as normal in the vat, and the CR to LR additional buffer can be configured in the cat.

3 Likes
#4

unrelated… but @hexonaut is there a “cheat sheet” on all of these cat, dog, ilk references? it is like another language… any link is appreciated…

1 Like
#5

So this is just configuration right now? No code needed?

#6

This is the best English descriptions: https://docs.makerdao.com/. That or the code itself.

#7

New code is needed, but it’s not a large addition.

1 Like
#8

Looks like the same idea resurfaces again. :grinning:

2 Likes
#9

Some here - https://makerdao.com/purple/#sec-7

Could be out of date but might give you some idea.

1 Like
#10

This is why I suggested a long time ago to have multiple vault tiers with different CRs, SFs, DCs etc.

It isn’t just a way to manage the debt profiles but also to gauge market sentiment.

There is another way to encourage maintenance of high CRs - a rebate on SF. Honestly could just drip back a Rebate_DAI to vaults based on their CRs relative to LR that can be periodically exchanged for DAI via a market mechanism (reverse auctions of DAI for Rebate_DAI).

Really what one wants is the way to ‘declump’ risk and the only way to address this was to have multiple vaults with staggered LRs, DCs and SFs.

Not simple though and something governance at the time seemed unwilling to do. But we now have ETH-A and ETH-B and are considering ETH-C so I stand by my statement that having multiple vaults on each collateral with different parameters may be a better way to manage risk. It also requires no change to any of the contracts simply an addition of more vaults which is easy.

5 Likes
[Signal Request] New vault-type for ETH with a higher LR
#11

Added a PR for this to go in LIQ-2.0 although I can’t speak to whether this will make the cut for the initial release as they are in final code review.

4 Likes
#12

going to post this in here again bc it seems like the more relevant place to have that discussion.

1 Like
#13

I think the concept is not to have people maintain their vaults at that level, but to limit their ability to generate further Dai below it. This should have some very nice effects for the overall risk of the system, including further mitigating OSM attacks.

1 Like
#14

This statement depends on if the desired CR is maintained on average. If people just initially deposit some collateral and let the SF drag their vault CR back to 250% before depositing a bit more you are in the exact same boat as you are in today just you’ll be charging less

How viable would it be to introduce some sort of penalty for going under the desired CR ratio? Maybe you don’t get liquidated but all the sudden your rate goes from 3.5% -> 5%.

How much additional effort is that?

Maybe it is just a one-time penalty of like 5% tacked onto your debt balance. Is that easier?

1 Like
#15

Hm I’m not quite sure what you mean? It’s not perfect, but it will definitely be a risk improvement versus being able to generate additional Dai below that level.

Regarding scaling rates to CR, this would be an awesome tool. It’s not too different from the process that Instadapp integrated with Gelato, where if you go under 150% in ETH-A, they’ll automatically put you into ETH-B. (Correct me if I’m wrong @sowmay @Samyak )

2 Likes
#16

that would be a lot better than just hoping for users to repay once they drop below CR. great idea!

2 Likes
#17

There’s already a discussion you started with the different vault type ETH-C which I totally support - [Signal Request] New vault-type for ETH with a higher LR

Inclusion of new vault type with lower stability fees and higher L.R (200-300%) makes total sense. Although, Higher C.R & Lower L.R with a difference you suggested 150% & 300% feels a little problematic as a user in the range of 150-300% won’t be able to pay back some of his debt with collateral if the end ratio is going to be 150-300%.

What I suggest would be a better solution is to include ETH-C vaults with higher L.R and provide tools to shift them between ETH-A & ETH-B to save themselves at time of ETH crash.

We’re keeping a close eye on ETH-C vaults and also planning to support refinance automation for ETH-C vaults the moment it goes live.

2 Likes
#18
2 Likes