Different Approach to Rate Setting

Where we are now

The DAI supply has increased by more than 250% (about 300m) in the 2-3 months since Compound kickstarted the yield farming craze. Before that period, the DAI peg was still recovering from the Black Thursday event when the DAI price reached $1.12 and needed about 3 months to finally recover in mid June, right before Compound yield farming was announced. This recovery was made possible due to bullish sentiment surrounding ETH in the second quarter this year.

Since then, ETH is still performing well and organic demand for leverage has only increased. However, DAI selling from leverage seekers could not offset the yield farming craze and on average DAI has traded at around $1.02. Potentially, market making activities also decreased due to lucrative opportunities from yield farming. Importantly, this all happened despite heavily increased debt ceilings, the onboarding of new collateral assets and 0% rates across the board. The issue is that all the conventional monetary tools at hand simply cannot compete with massive DAI demand from yield farmers and new price mechanics introduced by several AMMs like Curve.

Unconventional Solutions

There are multiple ideas floating around in Maker Governance about how we could address the current situation:

  • PSM
  • Negative rates
  • Strategic reserves through MKR dilution
  • Direct protocol DAI minting and participation in yield farming opportunities or arbitrage directly

Unfortunately, all of them have some technical or legal implications in the short term. Furthermore, certain solutions also seem to divide the community on topics such as:

  • Whether negative rates are positive for DAI adoption and potential technical implementation issues for partners
  • Should MKR build strategic reserves, dilute itself and spend its capital that’s actually needed for coverage of loss events
  • Does DAI become a USDC wrapper

It feels like these kinds of issues accompanied with the technical constraints of implementing unconventional tools to manage the peg mean that these solutions need more time for discussion, or a more serious peg deviation to implement. At time of writing, the DAI price stands at $1.01.

Current State of the Peg

In our recent analysis we showed how the DAI price moves by a few cents intraday only due to new yield farming opportunities. We also showed that such movements can be in our favour and we recently predicted one led by Curve farming. About two thirds of DAI issued by Vaults is currently being deposited in yield farms directly by them. This shows that if stability fees are changed in this dynamic landscape, this potentially doesn’t affect the peg because other more powerful forces are driving the change in price.

In other words, a few percentage difference in fee charged to Maker Vault owners doesn’t dissuade them from either speculating on 50-1000% APY yield farming opportunities or speculating on the current bull run of crypto assets used as collateral.

Different Approach

I have written in the past how Maker actually pursues three goals when setting rates: Risk Policy, Monetary Policy and Business Policy. Business policy is something that was overlooked thus far since our primary goal was and still is to manage DAI price as close as possible to $1. If we are able to reach DAI dollar parity, Risk and Business Policy can both be pursued with more flexibility. Let me first explain why Business Policy is important and then address how we can sort out peg issues.

MakerDAO Vault users prefer stable and competitive rates. If we are able to solve peg issues unrelated to how we set fees, we can finally start becoming more rate competitive and reliable, which is of high importance when we start adding real world assets (RWA) as collateral. This also means we need to change the way rates are set through Base rate on a weekly basis and set rates less frequently by having more focus on the competitive landscape and portfolio risk. This however doesn’t mean our conventional monetary levers are completely ignored; if we are able to increase stability fee, DSR can be used again as a more regularly used monetary lever. Also in extreme situations we can still use more aggressive stability fee or debt ceiling adjustments to help with peg deviations.

Finally, the community is aware of how low fees are unsustainable long term in cases when funds are needed to cover potential shortfall, particularly now that debt exposure is increasing at a fast pace. Assuming debt exposure doesn’t change as much, Maker protocol would benefit from 4m - 12m revenue per year assuming increased average fees in range of 1% to 3%. Assuming additional revenue is collected from penalty fees, we could be aiming towards 15m revenue per year, which would be a nice boost to MakerDAO’s reserves to protect against losses. Importantly, maintaining a system surplus will allow us to onboard and compensate new EPCs in order for MakerDAO to become self-sustainable. For this to happen it is critical that the Protocol has a more regular inflow of funds.

Conclusions

To sum up above concerns:

  • DAI price doesn’t necessarily react heavily to increased SF due to other more powerful forces that determine the price currently (DAI demand from yield farming)
  • Stable and higher but still competitive rates should be preferred from business policy standpoint
  • MakerDAO needs more income from rates to cover potential shortfall and make governance self sustainable and grow domain teams

Cause and Effect

In practice, the major concern about increasing stability fees from a DAI peg viewpoint is potentially causing liquidity drain, vault owners refinancing to other stablecoin debt denominated borrowings and causing the DAI price to increase. Therefore it’s important to always pick a fee that doesn’t cause massive refinancing and is in line with our Business Policy and position in the competitive landscape.

“Organic” Vaults are one of MakerDAO’s long term core stakeholders and retaining them is of large importance. Luckily ETH is in a bull run and as long as we don’t increase fees too much over competitive stablecoin borrowings, we shouldn’t lose them. Even at 0% fees, using MakerDAO to borrow DAI is more expensive than borrowing DAI at Compound, where COMP rewards already offset the current borrow rate for a user (net APY of 7.5%). This tells me users still prefer to use MakerDAO when it comes to DAI borrowing, but surely there is a breakeven rate.

Defending the Peg

In case we trigger a huge DAI price increase due to increased SF, this could be mitigated by a very low liquidation ratio (LR) for stablecoin vaults, which may have a similar effect as to what PSM proposal was trying to achieve. If governance sets the USDC LR to a level of 102%, we effectively put a ceiling on DAI price of $1.02, assuming the debt ceiling isn’t utilized fully.

Of course such a strategy also has side effects and these were discussed in the PSM proposal. It is unavoidable that with lower LR on USDC we keep increasing USDC exposure, particularly in times of high DAI demand (new DAI yield farm becoming hot again). We also must be aware that we will likely need to have low fees on USDC vaults should LR be kept really low. Further, some of the USDC vaults might also need to be manually liquidated as yield farmers could abandon them in case DAI price decrease doesn’t offset their liability from fees and remaining collateral. Here I think governance needs to decide how much USDC exposure is willing to tolerate versus how much peg deviation is acceptable. Importantly, governance also needs to consider other stable coin collateral onboarding to diversify risks associated with stable coins.

The other important fact is that we also need to increase the number of onboarded collateral assets with high minting potential and charge fees. We will need more community contributors to assist with risk analysis and if protocol starts earning money from higher fees, the protocol should have the funds to compensate these contributors.

Yield Farming

A short note on yield farming. As noted, yield farming changed the landscape in DeFi and affected MakerDAO. There are ways to mitigate it and meanwhile collect fees. Again as we showed, about two thirds of current DAI supply minted by certain Vaults is used in yield farming. Those Vaults shouldn’t be hurt by a few percentage points lower yield and they can’t really go anywhere else to borrow large amounts of DAI that is needed to farm. We must find the best strategy to a) retain them, as many of them are providing DAI liquidity in venues such as Curve and b) charge fees that are still tolerable for them. In such a way MakerDAO gets indirect participation in their APY. There are also more exotic alternatives available such as onboarding farmed assets as collateral (yCRV, cDAI, cUSDC), but risks need to be properly addressed first.

Next Steps

  • Discussion around this proposal, specifically:
    • Is the general strategy of introducing more business policy logic into governance sound enough and doesn’t affect our approach to risk and peg?
    • Is governance willing to change the way how rates are set?
    • How does governance view low LR on USDC and associated risks with potentially increased exposure? Do we want to use low LR on USDC only when peg escalates or do we put a ceiling on DAI price immediately by implementing it?
    • What is the peg deviation tolerance on DAI to assess proper LR for USDC should we decrease it immediately to defend against price appreciation?
  • Proceed with a Signal request.
25 Likes

Thanks for this. I would love to increase the base rate so that we can get 1-2% fees as I don’t think that would affect DAI supply too much. The biggest challenge is swaying popular sentiment and more importantly whale voters. It’s nice to hear experts advocating for a rate increase and hopefully that allows us to tip the scales to start pushing in that direction. It would be nice to be able to start paying more people for things so we can grow in ways other than supply which would actually have a more holistic effect on the supply, peg, new collateral, etc. Not to mention starting to insure the system for the next inevitable collapse.

8 Likes

I echo your sentiment to have some kind of positive stability fee for three reasons:

  • There is no evidence that lowering the base rate to -4% has helped correct the DAI peg in this yield farming environment. Likewise, there is no evidence that a non-zero stability fee will hurt the peg. Simply, the demand for DAI significantly outweighs any nominal fee
  • The DAI supply may be able to grow faster using the surplus to compensate additional domain team work. We have a significant backlog of collateral on-boarding, which at the current rate, will take years to get through
  • Absence of a positive DSR is handicapping the use case of Dai for the key end user (“normal” people who aren’t yield farming).
9 Likes

Let me start with the fact that strategic reserves are not needing a MKR token holder dilution. It is one way out of 3. Moreover, with the USDC-M proposal (using cDAI), it would be enough to use 100k$ out of the surplus buffer to achieve unlimited leverage (and keep all the profit) and therefore keeping the promise 1 DAI = 1$ of collateral. This would separate the business policy from the peg policy in my mind and bring a new revenue stream. At the extreme (assuming enough diversification of workable collaterals), we no longer need vault holders (but obviously we love them when SF > 0%). We could put SF @ 1-2% without fear (the exact SF would be set to maximize fees not to solve the peg). There is currently low traction for that (2 likes), but I will push that again in the future.

I’m not a big fan of low LR on USDC-A because it increases the risk, gives no profit to MKR, and we don’t control the DAI usage (some will likely be used to farm Compound if the past is predictive). The facility will not be used 100% (or even 50%) for solving the peg issue => more risk and no fees for the same effect. Moreover, we can’t dispose of USDC in an emergency because setting USDC-A risk premium @ 1000% to liquidate vaults quickly would be PR suicide (as it should be). In any case, @Aaron_Bartsch posted a good signal request in this direction.

Indeed, we fully agree on the role of farming on the DAI peg and the need of revenues to have a workforce and sustainability. I feel MakerDAO is moving toward the solution. Fully agree also on setting the rates using the competition as benchmark if that’s compatible with our risk appetite.

2 Likes

I agree it seems likely that a small positive stability fee will be well tolerated by current DAI borrowers. It would be interesting to see the results of experimenting with this.

2 Likes

I insist that SF is 0 is not desirable, it will not help DAI peg the dollar, but it will affect our system surplus. The facts tell us that a SF of 0 does not have the effect we have seen, so please abandon this conservative plan. If we want to improve DAI’s peg to the US dollar exchange rate, please choose another plan. An appropriate increase of SF to 1% to 3% will increase the system surplus and will not affect the exchange rate pegged to the US dollar.

4 Likes

This is a great post Primoz. I accept your premise that Yield Farming has returns that are an order of magnitude larger than the Stability Fees, so small changes in the Fees will not make a significant difference. Taking competitive inflection points into account is also helpful to maximize operational revenue while not losing all our business. To be clear, are you wanting to raise the Base Rate to 1-3% and reduce the frequency of the weekly adjustment vote or are you thinking of some entirely new mechanism?

If we are to go down this route, my one concern is that we have a concrete plan for how this revenue will be spent to fix the Dai peg. Our bottleneck for collateral onboarding currently seems to be the domain teams, so do we have some way to parallelize that with more money available? I understand subsidizing risk compensation to MKR holders is clearly a temporary solution, but as I’ve said in the past, I think the biggest risk we face is a broken peg. The compensation will come when we reach market maturity. For now we want to grow.

There is also still the foundation. Are they out of money? If we have a plan to increase collateral onboarding throughput and we need money to do so, why lean on vault owners when the foundation could cover it?

So in summary, I could be on board with this plan provided there is a more clear action plan for how the new revenue will be spent to fix the peg. I would vote no for this plan if the only reason for doing so was to give us a surplus buffer for risk compensation.

5 Likes

Great post. The business side of this is essential and must not be forgotten.

Further, DAI needs more diverse collateral. At this risk of being a broken-record on this, the real-world asset initiative is paramount to introducing stability over time. Primary as it acts as a counterbalance for any rate increases on the digital asset side.

While farming may not be rate sensitive today, collectively the future is uncertain as to what the rates will be. Moreover, Governance must balance bringing on yield assets while not injuring just the average guy / gal that is borrowing against his / her digital asset and has been for months.

Going back for a moment, the community (via risk teams) needs to asses what is the risk associated with every collateral type and what is the “target risk premium” for that collateral. Today implementing those risk premiums out of the gate is not appropriate. It would cause an increase in the peg which collectively is not acceptable. Right now, the community (correctly) had needed somewhat put risk in the backseat in exchange for the peg. That said, we should all strive to know what the target risk really is.

By doing so and having a reference point, one day, when real-world assets are on-boarded in bulk, MKR governance can start to move the RPs for all digital assets toward their target RPs and then use the DSR as the control mechanism with slight modifications to the RPs across the board. This will drive both stability for DAI and revenue to the MKR protocol.

Until those days, defending the peg is essential and a keystone for the entire project. Stablecoins luckily can help get us through this rough patch in the project until it can organically have enough liquidity and market forces on its own.

3 Likes

Thanks everyone for valuable feedback. Let me try to address some concerns.

Apologies if my post implied that. I specifically wanted to address only that MKR capital is better to be used against potential losses rather than mitigating peg issues. I do agree though that reserves in some form are interesting versus MKR dilution, but only for the sake of protection, not peg mitigation. Also money raised through MKR capital (in any form) may not be enough to solve peg issues.

Sorry if I wasn’t clear. I think we need a new mechanism that doesn’t use Base Rate. I know base rate simplifies things for governance and I was in favour of implementing it. But if we lower the frequency of rate setting, make rates more stable and competitive, the overhead of changing each collateral’s rate shouldn’t be too big. Rate setting becomes market driven and making sure we don’t underscore our portfolio risk by too much. Risk premium is still used and compared with SF that we set, so that we know how protected we are. This is still better than current situation where both portfolio risk and underpriced rates are basically a hostage of DAI price. If we can solve DAI peg with lower USDC liquidation ratio, or simply hope that higher SF won’t cause peg issues, we can start acting more market driven and assure our exposure is better protected. Otherwise we are always having this trilemma, where it is hard to satisfy all three conditions with one SF: peg in balance, stable market driven rate and compensation for portfolio risk.

And to be clear 1-3% range for SF is not something I am proposing. Instead I would want people to focus more on competitive landscape and see what SF should be properly applied for each collateral type. There are constraints of course as we don’t want to cause massive unwinding of positions and peg deviation. Because even if we have a lower USDC LR in place to protect against peg issues, governance needs to consider what exposure to USDC is still tolerant. 1-3% range was just some range I came up with.

As noted above, the idea above pursues three key things: risk compensation, more market driven fairly priced rates and peg mitigation through a potential lower USDC LR solution. Having more funds from higher rates though has positive side effects: we have long term self-sustainability to pay people who perform collateral onboarding assessments that scales DAI supply. By Cyrus leaving I can’t stress enough we need more people in risk.

7 Likes

If I summarize:
SF( C) = max(risk_premium( C), min_lending_rate_competitors(DAI_for_C_pair) )

for C each collateral we provide. Therefore we have always the lowest interest rate while avoiding taking risk we don’t want.

Correct?

This assume we take care of the DAI peg by other means (100% agree we must separate those topics).

PS: Regarding strategic reserves, I’m just correcting my own mistake, it is 100% my fault if it blended with MKR as I implied it at the beginning.

1 Like

Great post, Primoz. While negative rates are still my preferred solution to the peg problem, I am willing to try another route to fix Dai since MIP20 was so controversial. Even though it is risky, I think we’re left with no choice other than becoming a centralized wrapper to the decentralized world.

On the PSM:
The PSM is an overreach of power on our part. We need to let price discovery happen outside of Maker’s contracts. Cronje’s recent post sums it up nicely:

It’s about supporting each other and elevating those around us, not for our own personal well being, but for the space as a whole.

While 102% CR on USDC is similar to the PSM, I think an important distinction is that we are leaving the trading fees to other platforms. This gives the space more room to grow independent of our success, which is the way it needs to be. A low CR also allows for more price discovery below the threshold price.

If we do lower the CR to 102% on USDC, then I would expect the debt ceiling to balloon upwards towards infinity. We need to be ok with this huge ceiling. USDC will most likely become the primary collateral in our portfolio, and this cannot be discouraged with a high risk premium or debt ceiling limitations.

If we want people to go up to 50x short Dai, then like you said above, fees need to be low. A 4% risk premium multiplied by 50x leverage is unreasonable. The risk premium needs to reflect this “pivot” in Makers direction. Not only are we accepting centralized collateral, by lowering the CR to 102%, we’re saying we prefer centralized collateral. I’m alright with this, only because it seems we’re out of alternative solutions.
I suppose one could say that I’ve been “red pilled.” If negative rates are off the table; either the project becomes a wrapper for centralized assets, or the peg is at the whim of farmers and we earn 0 Dai in fees. Not an easy pill to swallow.

3 Likes

So I think there are a few separate issues here. I’ll try to break them out:

First, let’s leave the lower USDC LR as a separate issue. I agree it should be lower to provide an upper limit on the Dai price. We can do this regardless of fee policy.

With regards to your other two points:

Risk Compensation: Long term we should totally compensate MKR holders for collateral / portfolio risk. In the short-term, however, I view the peg as a much greater risk to MKR holders. We are not in a mature market, and if you are looking at MKR from a purely risk/reward perspective, the returns from price appreciation (coming from a solid, working product) should outweigh the 1-5% from Stability Fees in a market that is 1/10 or 1/100 of the mature market size. It is because of this that I am not concerned at all with risk compensation at this stage.

Market driven fairly price rates: I’d be more okay with this once the peg is fixed. At best moving to only slightly undercut the competition will have a neutral effect on the peg, and possibly some additional upward pressure (albeit a small amount compared to the farming pressure).

This is maybe a bit more speculative, but I do not view this farming craze as sustainable. I think we just need to try and grow the supply as best we can to meet the demand, and wait it out. Things such as the yearn.finance yETH vault may help a lot to reduce the ridiculous APYs. Right now we are at the mercy of farming, but we may return to leverage being the dominant factor again in a few months. I think it’s important to maintain the Base Rate lever in this case.

With regards to getting more people into the domain teams, I think this is important as well, but is money our biggest obstacle? How will increasing Maker protocol revenue assist with this in the short term?

RWA are probably something that require more fixed, stable rates, but I still like using the Base Rate for general market dynamics on crypto-assets. We could exempt RWAs from the Base Rate model without needing to axe it altogether.

3 Likes

I’ll chime in here. From the technical perspective, the Base Rate is becoming somewhat unwieldy to manage on the smart contract side. It requires off-chain management of what the current base rate is supposed to be, which is currently maintained in a forum post, and because some executives don’t pass, we don’t have a good source of truth at a glance what this value should be at any given time. So, some amount of research is required weekly to figure out each collateral’s risk premium, the current base rate, and the expected base rate, and calculate these values accordingly.

The system is now rapidly adding collateral types, and while voting around the base rate at negative values has little effect on collateral types that remain at 0%, every adjustment to the base rate once we’re positive is going to result in an edit to every system collateral stability fee. This will increase the cost and overhead of each spell for each new collateral added to the system, to the point where a single base rate change could modify dozens to hundreds of collateral values. We are being diligent with our audits and working on tooling to streamline the process, but it should go without saying that every modification to a system value introduces a potential for error.

I am a supporter of tying stability fees to individual collateral risk and making adjustments on a per-collateral basis and more deliberate cadence.

8 Likes

As someone with a sized vault got hurt twice by two digits SF and black thursday I am not in favor of this approach first dai peg needs to go to 1

Not everyone is playing with farming and these high yields will be gone shortly. Pushing vault owners to play with farming to compensate SF is not fair

3 Likes

@BrianMcMakerDAO totally agree with you on this. Base Rate was a good idea in theory, but events did not play out in a way where the Base Rate would have been useful. With the onboarding of Real World Assets, where interest rate changes are a much more sensitive issue, the Base Rate/Risk Premium concept is getting old fast.

4 Likes

So we’ve identified a problem that Base Rates are becoming unwieldy. What is the solution here?

Should we scrap base rate and just have RP’s per collateral? Can we create a module for BR adjustments to offload that work?

I’d also like to add, what if we just scrapped risk assessments for collateral onboarding for now and retroactively did them at a later date as we have the manpower? That way we can get the Vaults online and creating DAI and worry about the risk later. Not like we care that much with -6% BR anyways.

Not entirely disagreeing but the problem is how do we vote increases & decreases in basis of peg alignment, per collateral on an absolute basis? Agreeing on increases or decreases will likely be more time consuming

This on the other side could open the door to fixed one year or monthly rates for long term ETH deposits and apply peg variable rates on the remnant.

1 Like

Automating this process I think would be best. I like the BR/RP dynamic as it is analogous to central bank lending rates.

1 Like

With regards to the base rate. One solution is to set it to 0%, update all the stability fees to sensible values, and then stop voting base rate on a weekly basis. In this way we would still have it in the toolbox, but it wouldn’t change regularly.

5 Likes

Yea, it might make sense to scrap the cadence and move it to signal request/emergency only. The larger we scale the less we have to play with the BR I think.

1 Like