[Signal Request] Should we change the RPs for various collateral types?

Overview

With the base rate potentially being changed to -4% soon, we have the unique opportunity to adjust risk premiums on many collateral types without dramatically changing stability fees. There have been some concerns voiced on the forums, in community calls, and rocket chat that BAT, WBTC, USDC, ETH may have current RPs that do not accurate reflect their risk.

In particular I want to bring to attention that USDC currently has a RP of 4% and WBTC has a RP of only 2%. Both of these assets have custodial/centralization risk, whereas WBTC has both custodial risk and price volatility risk. There has been some limited discussion on thinking about collateral risk within this framework of risk categories here. Although, more discussion on this topic is certainly welcome.

Some community members have also mentioned that while BAT has a similar risk profile to a few other collateral types we recently onboarded(namely KNC and ZRX), its risk premium remains at 0%. Likewise Ethereum is a volatile asset and so not a riskless collateral type, so it is possible a 0% RP is not appropriate.

Up until this point the adjustment of collateral risk premiums have been used as strategic levers to fix the peg, and not solely to compensate mkr holders for collateral risk. The assumption made with this poll is that this strategic lever should largely be the role of the base rate, rather than individual risk premiums. If there are arguments against this assumption, input is welcome.

Let me know if there are additional collateral types of interest that we want to adjust risk premiums on, and I will add a poll for them.

Should we change the risk premiums for BAT, WBTC, USDC, and/or ETH?

  • abstain
  • Yes
  • No

0 voters

BAT risk premium

  • abstain
  • 0%(unchanged)
  • 1%
  • 2%
  • 3%
  • 4%
  • higher than 4%

0 voters

WBTC risk premium

  • abstain
  • 0%
  • 1%
  • 2%(unchanged)
  • 3%
  • 4%
  • 5%
  • 6%
  • higher than 6%

0 voters

ETH risk premium

  • abstain
  • 0%(unchanged)
  • 1%
  • 2%
  • 3%
  • 4%
  • higher than 4%

0 voters

USDC risk premium

  • abstain
  • 0%
  • 1%
  • 2%
  • 3%
  • 4%(unchanged)
  • 5%
  • 6%
  • higher than 6%

0 voters

Next steps

Poll will run for two weeks and depending on the result will move on-chain assuming the outcome of the poll deems it necessary.

8 Likes

I think that changing RP without having a solid framework is not the good solution. It’s like asking for a quote for health insurance and fire insurance without any other information.

We might get something a little bit better than currently but it will be the same empirical process.

3 Likes

This is shooting in the dark and the small changes will not help the peg. If we want income (since we cannot have the peg), we should say so and increase the rates as they will not make the peg much worse (nobody will buy DAI for $1.05).

We should also state that RP has nothing to do with making money. All risk premium should be provisioned to be used when the risk arise. Not used burning MKR.

3 Likes

I want to clarify that this has nothing to do with helping the peg right now, and more to do with getting accurate risk parameters that will help when we are raising the base rate. When we are raising the base rate later mkr holders will then be compensated for risk accurately, and also borrowers are incentivized to borrow the assets we deem as least risky over the assets we deem as more risky.

It also theoretically lowers future governance overhead because we can ask 1 question each week about the base rate, instead of asking questions about the risk parameters for every collateral type.

8 Likes

I support this view.

We must always be alert to assets with volatile prices. Considering the dramatic fluctuations in asset prices in the cryptocurrency world, we must prepare in advance.

The risk team put something together awhile back but to the best of my knowledge nothing has been open sourced so we kind of have limited visibility into their model.

There are some links in this post if you are curious.

Can we get some solid discussion on the ETH RP?

Is it possible to have any collateral with less risk than ETH on the Ethereum chain?

My thoughts are no, anything less volatile than ETH would surely carry custodial risk. If that is the case, then I believe ETH RP should be set at 0% and everything is scaled from there.

If we can imagine a collateral less risky than ETH, then ETH RP should be increased to allow this hypothetical collateral to be set at 0%.

Edit: I had typed SF previously. I have corrected to RP.

2 Likes

The MAKER community voted on the proposal that the base interest rate is -4%. Here, we are only discussing the risk premium of collateral, not the basic interest rate. I also believe that any collateral must have a risk premium and cannot be zero risk. We cannot confuse the risk premium with the base interest rate. As for adjusting the basic interest rate to a negative number, this is a temporary method for the community.

1 Like

I voted for ETH 1%, but i would rather go very small, like 0.25%.

2 Likes

I don’t follow. It’s not like current RP were determined by solid framework, right? What changed?

1 Like

I don’t see how changing the RP are useful.

There is a peg issue and a risk management issue. Changing the RP might get us think the risk issue will be addressed but it will not.

Raising RP without changing the surplus buffer (which will be full soon at such rates) doesn’t make sense and doesn’t remove any risk.

We should focus on what matter.

Have been discussed before the creation of some market simulator, that we could provide statistics and get a simulation of how could some RP affect the DAI price for many parameters (this applies to many other parameters of course)?

Such simulator could help us to start establishing a framework to determine the best RP combination, which could have in its core results from a simulation based on statistic collected. Also we could study on it the risks (like running those “worst case scenarios” we think about). We’re right now using social-intelligence to get the best RP combination, but I should agree that this is not an efficient methodology, and we’re spending many voting polls, weeks (and GAS) to test what this “social intelligence” considers the best.

Please note, that I’m not saying such simulation should tell us which the values will be, but if correctly tested with the past (and I’m aware the past is not indicator of the future), it could give us clear signals of how bad or good some configuration of RP should be, and we could have even a base to prepare our polls (I’m not sure that polls with values: 0%, 1%, 4%… makes always sense, for some assets, maybe something like 1%, 1.25%, 1.5% is the best thing to vote on, but it is very hard for us to see these differences right now).

I don’t think we fool ourselves. It’s about more transparency of amount of subsidizing (i.e. negative base a rate). As mentioned in OP, BAT currently has 0% RP.

I am not sure if i follow here… what’s wrong with full surplus buffer and burning mkr?

It is possible. The uniswap tokens have less price volatility risk than ETH, and if both tokens are non-custodial, such as dai-eth, the uniswap tokens are also non custodial. Not to mention we have yet to concretely quantify what custodial risk should mean for risk parameters. It is likely that some tokens which represent derivatives of the cryptocurrency market will have lower overall risk that standard volatile cryptocurrencies.

2 Likes

The risk premium is a way to compensate the risk of a collateral hitting the solvency structure or MakerDAO. Setting 4% RP for USDC is like saying there is a 4% chance USDC goes bankrupt per year (losing keys, being frozen by the US gov whatever). Maybe a little less if you put risk remuneration here. The loss when a risk arise doesn’t have to be 100%, but it’s not impossible here. Let’s pretend it’s a 1% risk (100% loss).

Currently, the equity part of the MakerDAO balance sheet is the surplus buffer (0.3M$, max 0.5M$). We in have 50M DAI backed by USDC.

MakerDAO will become insolvent in such case and will have to raise 49.5M DAI. Sure it can work, there is a lot of immaterial value in MakerDAO but can also fail. In any case, it wouldn’t be unfair to wipe all current MKR holders.

4% RP for USDC-A mean risk about 2M DAI per year. With a surplus buffer of 500k, you are expressing that you expect to lose 4x the surplus buffer each year. As nothing happened with USDC for some time, you expect to lose way more than the buffer with one issue leading to MKR issuance. Either there are other reserves in the system that’ I’m not aware, that’s not risk management, that’s YOLO.

From my perspective you can’t discuss risk premium without have a framework to manage risk.

With BR @ -4%, the peg in its state, and the lack of diversification we don’t have to move fast as it’s not something that can be solved in the next few months. That’s why we have time to discuss the framework before moving in action.

Since this post is getting traction and sentiment seems to be changing, I just wanted to plug my signal request for increasing the ETH-A RP. “No” is still winning but based on this polls results I’m wondering if some people haven’t changed/added their vote yet. Please be sure to signal if you can. Thanks.

3 Likes

If we simplify, why would buffer size matter? Important thing is how much (dai) fees are collected annually. If there is MKR issuance, auctioned price would be influenced by previous MRK burn rate.

(And i am not saying, we should not increase buffer size).

You can. You just won’t have very good estimates. I don’t need to have a formal “risk framework” to determine BAT RP of 0, is too low.

1 Like

You’re right that earning (fees collected) should somehow have a value. That’s exactly why Synthetix is backed by nothing more than expected earning. That’s unheard of in TradFi but why not.

The issue is that such expected earnings also have a risk premium on them, or a discount rate. If MakerDAO need $50M, a bidder can provide it to resolve the insolvency issue or copy/paste MakerDAO code in a MakerDAO2 with a better solvency.

You don’t need to have correct RP or even generate fees for that to work. Expectation is all that matter. If you know the supply rate in Compound/Aave is around 5%, it seems okay to charge 2% on the whole DAI supply. You don’t have to do that to be counted in your MKR valuation.

You are hopping that there will be buyers of last resort, that MakerDAO will be too big to fail and that they will be nice/fair to existing MKR holders. I wouldn’t bet on that.

PS : Not saying that BAT at RP 0% makes sense, just saying it’s the least of MakerDAO problems right now…

1 Like

Completely agree with now being the time to set the risk premiums for collaterals that have been grandfathered in at 0%. That being said, we should be deferring to the risk team as to what those rates should be. It’s their job to do the analysis and come up with a number based on science. I think we should stop having these types of popularity polls for Risk Premiums where the community guesses at what a good number is. We have a risk team. Let’s use them.

8 Likes