Ideas on DAI PEG, USDC-A vault, fees, etc. A basic analysis + suggestions

My apologies for the length of this I have been working on these ideas regarding PEG, USDC, liquidity and fees in general in a disjointed manner for a while now and it took me a while to assemble them all into this post below.

Situation.

  • USDC OSM locked to $1 (I don’t see any impetus to change this).

  • USDC-A vault SF=4% with liquidations off.

  • USDC-A DC 412M out of 485M DC

  • PEG holding nicely at 1.01+/-.002

  • Last CC report had accumulated USDC-A fees putting entire facility CR at 1.0042. With the DAI price at 1.01 currently we are .0052 x 412M approx 2.14M in the hole in the surplus if we had to clear this at 1.01 today.

  • Right now this is strictly a percieved accounting issue. Nothing horrible happens if we just set the SF to 0 on this vault.

  • IMO Technically there is no emergency. But there is some urgency to act/decide

  • Blacklisting risk I think is still unclear but the longer and larger we go I think the less people are concerned.

  • DAI being off PEG here consistently at 1.01 is affecting adoption and confidence in DAI hurting growth in the ecosystem.

We have various proposals:

  • MIP21 vox MIP20: Target Price Adjustment Module (Vox) - basically a DAI backing collateral devaluation model that much of the community was generally against that has passed. If we need to do this to solve the PEG issue we have something in the hopper.

    • My take/analysis on MIP20 is that it won’t have enough teeth to produce sufficient liquidity
    • ramps up/down this backing value too slowly.
    • Whether doing it is damaging to the credibility of 1:1 DAI to collateral USD value remains to be seen.
  • MIP29 PSM - MIP29 - Peg Stability Module

    • A lot of early discussions focused on details of implementation
    • At the time there were various concerns over the system taking on a huge amount of USDC without fees or minimal up front fees which is basically where we are right now.

I think everyone is agreed we would rather have sufficient collateral backing to have our users mint sufficient DAI to maintain the PEG via normal issuance. In this case our users end up taking advantage of any DAI price premium or deficit and not the protocol itself. So far large new collateral opportunities are out into the future. So Maker itself is at a kind of a hard junction. If we push down this road to effectively PEG DAI to USDC/stablecoins generally utilizing our ability to mint DAI with the protocol we affect markets securing the PEG. What I see though is general discussions of using these facilities and PSM across many stablecoins. I believe doing this across multiple stablecoins has some general unintended hazards and potentially some as of yet unknown risks (thinking of you @mrabino) but there are also some opportunies.

Over the past month or so we have observed for the most part the USDC-A with LR of 101 has us at a PEG of 1.01 with approximately 400M outstanding from that facility. I want to caution this whole situation has not been ‘battle tested’. Ala 25% 24hr ETH price drop or BT back to back 25% drops in 24hrs. We also have not even attempted to observe what happens if the DC gets maxed out.

So far it looks like USDC-A is doing a fantastic job of keeping the PEG at 1.01 as long as DC is available.

One might wonder why not just lower the LR on USDC to 100? Well short answer is the protocol basically gets no fees at all and we bear all the risk for no profit.

Right now ONLY if the PEG were to magically find its way to 1 does the profit come through intact on USDC-A with LR of 101. But this is a catch-22 kind of number cooking because so far nothing magical has happened to drive the PEG to 1 and why we are still in this situation.

So lets work out the real choices here.

  • A single USDC stablecoin vault with LR at some value approaching 100.
  • PSM with fees which itself is based on a new internal vault with a LR of 100 and upfront fee spread

Both of these vaults basically have

  • SF = 0 (PSM has an upfront fixed fee)
  • DC effectively unlimited to function properly.
  • USDC OSM locked to $1 for the foreseeable future.

MIP29 the way it is proposed captures a fee spread bracking 1:1 within the range so high low ranges like .993-.999 or 1.001 to 1.008 will not be possible. Generally there appears to be no consideration of having back up facilities in case either USDC-A or the PSM reaches maximum DC other than USDC-B limited to 30M. At this point all DC changes happen under governance and are limited in action by the GSM (now 72hrs).

Put simply if I had to choose an approach to PEG DAI to USDC I am going to pick the PSM one because it is cleaner in fees even if it lumps any USDC-USD pricing hazard under the OSM lock to $1. Inherently it is cleaner than a vault both in structure and fees as well as pricing control but the PEG management becomes dominated by the USDC the PSM DC can trade for run out of DC or USDC and the PEG will go off again.

Key point: There are no mechanics in either of these approaches that create a mechanism for the PEG to move to 1 other than narrowing the fee levels on the high side of the PEG (tout) on the PSM or the LR on the vault.

The simplest way to put this is Maker with either of these has clamped the trading prices with unlimited DC on the upsdie and 1 with the USDC vault (or lower with tin and PSM) on the downside as the protocol takes on more and more stablecoins.

I personally would like to see a PSM that is a little more dynamic in terms of price bracketing so the pricing bands move as the top or the bottom of the band is utilized for liquidity and thereby act a bit more like a normal market. In this case I imagine multiple PSMs operating with different DC levels in different bands to anticipate unexpected market events.

Off the top of my head multiple PSM example anticipating a PSM DC limit being hit: In ourcase say a first PSM with 500M DC between 1 and 1.005, another with up to 1B between .999 and 1.01 and yet another with 2B between .99 and 1.02. This way there is always DC available at some price. Use an automated MIP17 MIP17: Weekly Actual Debt Ceiling and Actual Risk Premium Adjustments to leak out liquidity on these at some rate so no-one can just take all the liqudity at once. (think flash loan here) may be another way to handle not hitting DC on a single PSM btw but this means the DC can run out occasionally and for limited times until the high DC limit is hit.

If we are going to extend this to other stablecoins I think this opens up a lot of unforseen interactions that may be complicated to deal with and this is particularly true if we are OSMing these stablecoin prices all to $1 and the markets are saying otherwise.

Frankly these are not optimal solutions but the methodology seems to work even if being successful in clamping the PEG price into a bracketed range reduces trading opportunities. The biggest hazard is not taking into account the real stablecoin USD OSM price in these models/calculations even if we have liquidations off on these.

Fee Opportunities in the above situations.

Now as I know from the last CC others have been looking at ways for Maker to utilize these stablecoin vaults from the PSM or otherwise to earn revenue. I wanted to add here my own views on key issues related to earning a return with stablecoins and present some ideas based on a basic approach.

Thoughts I have had on this whole stablecoin issue isn’t just that we

  • take on all this risk for a flat non-recurring fee
  • but that we can’t use the capital we get to do other things. Maker has always been a protocol with guaranteed liquidity. We dont loan out our assets. But lets think through the following scenario.

Either in PSM or USDC-A people are going to deposit USDC and mint DAI. Now lets say we take some chunk of that USDC and pair it with DAI again and put it into the Uniswap DAI-USDC contract so people trying to do uniswap exchanges have less slippage and Maker to earn fees.

Example: 1M USDC gets paired with either 1M DAI minted from thin air or even 1M DAI from surplus (take your pick here where the DAI comes from because in the end it doesn’t matter).
In uniswap liquidity is guaranteed - you can burn all your LP tokens and get back the corresponding USDC and DAI instantly.

Now there are three cases(and if I have done my math correctly):

  1. PEG is > 1 - so in this case there will be more USDC in the LP so the deficit DAI is made up almost entirely by USDC. Lets say 1.01 to illustrate. so 1M USDC: 1M DAI becomes approx 1004974USDC and 995026 DAI so here at 1.01 4974USDC buys 4924 worth of DAI. We lose 50 DAI which is .005%
  2. PEG is < 1 - so there is more DAI to USDC and the protocol wins DAI but if it has to make up USDC by buying DAI it if the PEG is .99 it costs the protocol 50USDC or 50.5DAI
  3. PEG = 1: DAI is perfectly balanced at the OSM price there is no loss or gain.

Now while this appears to be setting up the protocol for a loss if we are off PEG realize we just need to earn .005% in LP fees to balance out a loss of 1% on the PEG in the uniswap contract. Right now the Uniswap DAI-USDC pool has approximately 5M of liquidity trading 220K/day earning 650 in fees. The protocol makes a 1% PEG up/down price change Impermanent loss in 6hours via LP fees with 2M/7M or 28% of the DAI-USDC uniswap liquidity. We expect the PEG management to be better than .99 to 1.01. Is there a huge protocol risk here? Only if you believe we will see a PEG value of .9 or 1.1 and even then the losses are in the 5% range (50K/1M of liquidity)

In principle this approach can be used across all stablecoins to boost protocol PSM/stablecoin holding returns. It also has the added benefit of providing liquidity to markets to reduce slippage on trades to stablecoins.

Are there better ways for the protocol itself to earn yield by pairing DAI to stablecoins. I think others will come up with ideas. My concern is they could be liquidity constrained where as Uniswap LP is not. Will or can we use a UNI based LP vault (or even other trading venues) to mint more DAI to pair with more stablecoins to lever up - possibly but then at least on Uniswap we multiply our IL losses and our LP gains ONLY if we don’t have a high % of the LP liquidity.

I want to point out that all the DAI (even if this was unbacked DAI) under most normal circumstances just sits in the LP doing nothing except increasing the % of fees being earned for the protocol and reducing slippage going through this pair. This DAI can’t go into the markets without corresponding amount of stablecoins added to the contract. Think of UNiswap LP contract as the dynamic PSM and from this perspective it behooves Maker to actually take funds and deposit into these contracts not just to provide liquidity inbetween the PSM or Vault PEG spread but for liquidity when we move out of this spread.

I want to point out here the above is premised on people still using these DAI-stablecoin Uniswap pairs to complete trades for the LP (i.e. Maker earning fees in these LPs). IF Maker narrows the PSM spread or lowers the vault LR too close to 100/1 people are going to just pick whichever stablecoin gives them the best return either by getting best price and skipping Maker DAI-stablecoin swap or going through Maker as the last step to their preferred stablecoin. One case means Maker only gets their % of the Uniswap LP .3%, the other means Maker gets nothing off the uniswap LP but earns the fees directly in the PSM. Either way the protocol benefits

To conclude this section. I see little to no downside to Maker taking some of the DAI profits (or even minting unbacked DAI) and pairing it with stablecoins from the PSMs in the Uniswap LPs simply because this liquidity actually acts exactly like a dynamic PSM providing the protocol with Uniswap LP fees and guaranteed liqudity and stablecoin backing of DAI.

6 Likes

As it is stablecoin, we might want to compare with Curve more than Uniswap.

I’m quite sure @hexonaut will come up with this idea when the PSM will have USDC, TUSD and GUSD in stock.

Our goals is still to limit the exposure to those stablecoins anyway but in the last 6 months we decided to prioritize the peg over this risk. I’m bullish on RWA to solve the bulk of the issue and cDAI/aDAI for handling the remaining short term volatility.

3 Likes

I have not dug into Curve or Aave but a prime requirement on this stuff for Maker is guaranteed withdrawal liquidity. Compound doesn’t have guaranteed liquidity I am not sure the other protocols have guaranteed lliquidity as well.

I picked Uniswap because I understand it the best and it has guaranteed liquidity. If liquidity is guaranteed and the returns are better then I am all for it. If liquidity can be constrained then I have some serious issues because I consider taking assets and in effect wiring them through multiple protocols possibly to create liquidity hazards that should be avoided. It doesn’t matter if one can get 20%yr/yr return if you can’t pull the liqidity when you need it.

I am open ears though and when I have time I will learn more about curve, aave, etc.

As in my post I think we all agree we’d rather have RWA here to grow liquidity to push the PEG down. We also do have to keep in mind that when there is excess liquidity everyone is of the mind that rates can sop it up fast enough, but that may not be the case. Right now a 400M USDC buffer feels pretty good putting a bottom on the PEG but once that is gone (or if it gets eaten rapidly) the only way to sop up liquidity is by borrowers having liquidity to repay loans. It is the short term fast liquidity expansion and contraction we need mechanisms for. The PSM fails to control the PEG on the downside once the USDC exhausts.

I find myself repeatedly coming back to the same key topic at Maker - DAI liquidity control/mechanics. What is clear is this is a dominant issue that isn’t going to just dissappear and the community tends to not want to do the basic things to manage within the protocol how to expand and contract liquidity quickly within the other constraints imposed by governance. Everyone tends to have this idea that all the decentralized players are going to step up - what people are not asking is who steps up, how as a DAI liquidity supplier or remover of last resort when markets and people can’t or won’t.

1 Like

@MakerMan thanks for this very nice post.

I agree 100% on the need of putting our money on guaranteed liquidity systems.

This said, as @SebVentures suggested, I also believe Curve has this property and a better curve (different than c = x * y) designed fro stablecoins.

I just want to add one basic consideration:

Observation: MakerDAO was not designed to work like this, i.e., on relying on a stablecoin (USDC) to mantain the peg. It turns out, however, that this sadly looks like the only realistic option (while we wait for RWA) now.

This said, I would like the discussion to focus on: Even assuming that we manage to fix the peg with PSM, by embarking, say, 700m more USDC.

Main Question: How do we plan to get out of this situation? I.e., how do we get rid of our stablecoin exposure at some point?

This is a point that is still a bit obscure to me.
Sub-questions include:

  1. Do we generally expect to be able to really fix the peg (1:1, not 1:1.001 nor 1:0001) with the PSM?
  2. Do we expect that our USDC stablecoin exposure will decrease naturally once the peg is fixed (or semi-fixed, i.e., 1:001) ?
  3. If so, how? Or do we plan to wait for RWA? If so, How much RWA? Are we talking billions?

It is a bit unfortunate that DAI needs centralised stablecoins to be stable. We need to actively think on how to fix this situation (beside fixing the peg, of course).

3 Likes

I like to think of it in this way as it makes me feel better.

Literally we are backing DAI 1:1 with USDC. There isn’t anything to get out from under. It doesn’t matter how much because if DAI were to go down we now have 412M happy USDC that will buy DAI at 1 or anything lower. Think of USDC as insurance that the PEG doesn’t drop below 1 and WE don’t have to pay for the money. They paid us 1% to keep their USDC - so LETS DO THAT. :slight_smile: Forgive the loans and take custody of the USDC and put that USDC to work while we wait for a buyer for our sweet beautiful to look at and take out on a date USDC we are complaining about. :wink:

Because our hot USDC chick is going to get married to RWA or something else at some point and we might miss her later.

So yeah bring on USDC as our PEG control. Should we have a back up. I honestly think we gave USDT a bad rap. Putting the SF to 4% like everything else - it would be interesting to see that one get 5M DC and if it gets used to mint DAI like PAXG, TUSD are doing for us. If it got big enough it could be also used for some additional PEG duties. But I want to leave on a good note and not push the slutty USDT on anyone.

What IS interesting is that the first shot at DAI PEG to USDC via USDC-A so far is a success. Lets claim the fees to hold USDC as insurance the DAI price won’t drop below 1. All we have here is a kind of hand that if Circle or govt plays the blacklist event - everyone loses and I think numerous lawsuits fly over the event. Our issue is Circle and USDC at 3B is pocket change to 1T so as Maker move to 10B we need to get our act together.

liquidity, liquidity, liquidity how to mint and burn a B DAI like a pro. First you start with ETH, then you add in some USDC, some wBTC (I’d like to see some encouragement of more wBTC to come here btw - maybe slightly lower the SF offer some MKR to these BTC people honestly) focus on RWA so we can start doubling B’s here 2,4,8,16,32, etc. :wink:

BTW I did the math on current DAI return for Maker just to perk up the crew. I am coming up 15.13M/yr DAI interest (382.6*.02+187*.04) excluding USDC from the top 5 Maker DAI collateral backing caps. Return rate here on our total USDC, ETH, wBTC, TUSD, PAXG. at a 531M MKR valuation. 2.85% for 35P/E on Maker ‘right now’.

Maker is earning an honest 1.26M DAI/month towards surplus. We have money now we need to be smart about what we are doing with it. Both in surplus management, operations, treasury (i.e. curve/uniswap LP etc. mining and return seeking), community growth, etc. I’d love to see us start using some % of the MKR purchased for burn to go out as a an additional sourceCred perk beyond their grant for everyone working on the project. Make that sourcecred + whatever else MKR vest somewhere - so people can vote it (lock it in DSSChief maybe out of the gate for 1-3 years). People can vote it but can’t withdrawal until their vesting period completes. MKR As possibl rewards for vault holders to give feedback via the UI their on-chain borrowing transactions. It doesn’t have to be much but just connected to some on-chain metric with these folks. This might encourage vault holders to also hold MKR at a minimum it will give the MKR markets more volume… This way all interests align on improving the burn rate of MKR. When the DAO burns Maker those working for the project earn a share of MKR (through governance to say sourceCred or directly to Domain Teams, whatever but straight into DSSChief to vest/vote). The whole point isn’t to make MKR so rare by buying it up and burning it all but buying it up to burn less and redistribute more of the MKR to all of those who are working on it. In the end we still burn but generate a hell of a cash flow doing it. Later we basically split MKR 10:1 lol.

Man so late and I have to work a 12hr in 5hrs.

I think I have to swap some ETH for MKR because if ETH does well - Maker already is - MKR could be worth like 20ETH or more in as little as 3-5 years. It is going to be 3 ETH in literally a drop of a hat.

Ladies, gentlemen I can see the path to MKR=10-20K here.

all we have to do is NOT BREAK THE MACHINE.
Happy day after 1B DAI day.

1 Like

Curve definitely makes more sense than depositing 50/50 liquidity into DAI/USDC Uniswap.

1 Like

Yeah I definitely have to educate myself on curve. Like I said in the post I was just using uniswap as the example because I understand it. The only requirement was liquidity. If curve is a better place for return on the liquidity and a better place to cross trade stablecoins I am all for it. :wink:

The actual numbers were not the point - but simply illustrative of what could be done. I am thankful others are already thinking about how to optimize it using defi that I am still not fully up to speed on. :wink: The second part of that post was the suggestion part. Basically a general idea laid out with a specific example.

1 Like

PSM is interesting by setting up big sell walls. Dynamic AMMs can help to marketmake in the middle.

We should be careful before choosing to set up our big sell walls at 1.001 (current winning option for the PSM). It really doesn’t give us that much room to maneuver for the risk being taken. Also people are talking about the upfront fees, but why are we settling for 0.1%?

1 Like

Yep I pointed out this problem as well with a tight spread. Some real thought needs to be put into this regarding where we want to have the brackets pricing wise. It was why I kind of wanted the PSM bracket pricing to change according to a well defined liquidity price model so we don’t just set up hard price walls. The point was to try to maximize the fees but also allow for some market dynamics within the spreads to still exist. Make the walls some what soft, and movable but as the prices move away from nominal they get harder and move less.

1 Like

What risk? You mean warehousing some huge amount of USDC, TUSD, and PAX? Well, it’s too late to worry about that. We’re already warehousing way too much stablecoin. PSM doesn’t really make it worse.

Hey at least we are getting 4% on that TUSD and PAXG so I don’t see the issue there or are all the LRs 101 now. My oh me if I missed that update…

I agree. DAI is backed by too many centralized coins for this to matter much anymore.