Dai is back under $1. Time to get prepared for tightening that might be needed soon. What's next?

Let’s be very proactive as a community and make sure we don’t allow the price to go down very much. The last significant tightening cycle we were far behind the curve and the price of Dai sunk as low as $0.95, causing a lot of pain in the ecosystem. We don’t want to repeat that mistake!

Once it becomes clear that we need to start tightening (which is certainly not now, but could come very quickly), where do we start? My recommendations, one at a time and seeing how the Dai price responds:

  1. Boost SF for USDC. Current SF is 0.75%, move up to: 2.75%

  2. Boost SF for WBTC. Current SF is 1%, move up to: 2%

  3. Boost SF for ETH/BAT. Current SF is 0%, move up to: 1%

  4. Later on, start looking at increasing DSR

This is a non-linear system and when the regime changes it can whipsaw very quickly. We really need to monitor Dai price and not let things get out of hand!

7 Likes

We should immediately start to increase ETH / BAT SF to 0.5%. Others remain unchanged.

Boost SF for USDC. Current SF is 0.75%, move up to: 0.75%

Boost SF for WBTC. Current SF is 1%, move up to: 1%

Boost SF for ETH / BAT. Current SF is 0%, move up to: 0.5%

Later on, start looking at increasing DSR.

So, based on the recent stability fee structure changes, the only weekly fee-related vote will be the base-rate changes as detailed in the recent signal request: [Signal Request] Change the Stability Fee Structure

Currently, the cadence for the current set of parameters is as follows:

Base Rate: Every week
DSR Spread: Every 2 weeks
Risk Premium: As-and-when someone signals / or when proposed by a risk team.

The last is a little vague. The wording of the signal stated ‘we should vote on vault-specific fees on an as needed basis’ however it did not comment on how to determine when such a change was needed. In the absence of a specific method, I believe it would make sense to fall back on the signaling process.

So, with regards to this thread, I would suggest that you frame suggestions in terms of base-rates changes, DSR-Spread changes, and Risk Premium changes rather than as absolute values for individual collateral types.

8 Likes

We were over the peg for a couple months. We are fine being under the peg for a short duration to ensure that we are not “jumping the gun”. It is extremely easy for MKR voters to correct the peg upwards (as there are quality levers to use such as raising stability fees and raising the Dai Savings Rate.) This is not true for being over the peg once rates approach zero as we have seen this last two months. To sum it up, it is easier for us to raise rates than lower them so we should be sure we are under peg before we raise them. That said, once we are sure we are below peg I would support what was suggested here as a start in correcting the peg.

2 Likes

Apples and Oranges comparing the last time we were below peg to this time (should we get there). Last time we had no DSR and the system was totally different as well as single collateral. There is very little to stop us in this current system from correcting the peg upwards when we are ready to do so.

Agreed. Also, we’re not even below peg yet per dai.stablecoin.science.

No need to rush anything.

2 Likes

The DAI price is exactly where it was a few months ago - above the peg. Sometimes more (+1%), sometines less (like now, +0.6%).

I see no improvement whatsoever. Everyone here is denying that the problem with the current design exists despite DAI being above the peg for 3 months.

1 Like

First we will need to onboard the two handfuls of new collateral types that are currently pending. Then we observe the effects. After that we can discuss the next steps.

I think the effect of SF is 0% or SF is 0.25% is almost the same. Our key problem is not only the SF problem, but how to print DAI in large quantities.

1 Like

I suggest trying hundreds first and then observe the effects.

Thank you for bringing this up @IslandHunting. From some of your previous posts, I remember you saying that you’re a very active participant in the DAI markets. You probably have your ear finely tuned to the depth of the DAI books, correct?

late Jan - early May 2019 was when you were calling for big increases then, too correct? Let me see if I can find some of the reddit links.
I think this post sums up the previous “hawks vs doves” situation best: https://old.reddit.com/r/mkrgov/comments/b9mh4t/bold_and_drastic_action_is_a_bad_idea/
another early one on why bigger rate increases were needed:
https://old.reddit.com/r/MakerDAO/comments/b8r04h/broken_peg_market_makers_strike_very_little/
in hindsight, you were correct - more volatile rates were needed to stay above 0.98. To add to this, I think you were seeing the symptom (DAI price slippage) to rate arbitrate pressure on DAI. (borrow from maker at X% short on a perpetual swap contract Y%). Remember how quickly the debt ceilings would get filled soon after they were increased? It was because a decent amount (sorry, no real source of this number, only first hand accounts of etherscan pillaging) of this volume was finding its way directly to the centralized world. There were many other reasons too, like compound exploding in volume.

100%. So, let’s look into the rate arb opportunities and see which way the lending rates have been blowing. I tend to be a bit more hawkish as well, but I’m not afraid to acknowledge that 50% of the time DAI price does the exact opposite of what I think it will. Calling for higher/lower rates when the price is so close to 1 is going to take a good bit of evidence to convince the doves.

Fig 1

Fig 2

Fig 3 - Previous 24 hour funding rate

The funding rates did have meaningful change in the past 24 hours. Will that positive funding rate (longs pay shorts) be sustained as it usually is in bull markets? I have no idea. The rate arbitrage needs to be in place for long enough that it becomes tempting to capture. Maybe it will happen faster now that I posted it in this forum? It’s hard to say how sweet the deal needs to be. Regardless, whenever the unrelenting rate arb pressure comes (like it did in early 2019), it will be two-fold what it was last time because of WBTC.

To reiterate - I’m not sounding the alarm to raise the SFs. We’ll know when the funding rates are positive for at least a week and we see the DC getting used up at a faster than normal pace. @IslandHunting is on to something here though. Eventually, volatile rate and DSR increases will need to “close the arbitrage gap” between maker and centralized rates.

Edit: expanding on this with some code. 0.053% every 8 hours may not sound like a lot, but lets see what that is compared to our yearly rate.

fee_per_8_hours = 1.00053
print("deribit fee per 8 hours: {0} = {1}%".format(fee_per_8_hours, (fee_per_8_hours - 1) * 100 ))

initial_principal = 100
fees_per_year = initial_principal
for k in range(365*3):
	fees_per_year = fees_per_year * fee_per_8_hours
fees_per_year = round(fees_per_year - initial_principal, 2)
print("total fees deribit after 1 year: {0} = {1}% APyr".format(fees_per_year, (fees_per_year/initial_principal) * 100))

initial_principal = 100
fees_per_year = initial_principal
# this hex value comes from this spell 0x606395f1b167aa73134b8a2b34c25bb7d0564920 and should be 20.5%
fee_per_sec_hex = '33b2e3cf1e0820ac6d7b91b'

fee_per_sec = int(fee_per_sec_hex, 16) / 1000000000000000000000000000
print("maker fee per second: {0}".format(fee_per_sec))
for k in range(365*24*60*60):
	fees_per_year = fees_per_year * fee_per_sec
fees_per_year = round(fees_per_year - initial_principal, 2)
print("total fees maker after 1 year: {0} = {1}% APyr".format(fees_per_year, (fees_per_year/initial_principal) * 100))

output:
deribit fee per 8 hours: 1.00053 = 0.05299999999999194%
total fees deribit after 1 year: 78.64 = 78.64% APyr
maker fee per second: 1.0000000059132284
total fees maker after 1 year: 20.5 = 20.5% APyr
If rates on centralized platforms like deribit are sustained at 0.02% or higher (~24% yearly equivalent) we should expect lots of arbitrage pressure if we’re in the single digits.

4 Likes