[Signal Request] Initial Flash Mint Module Fee

hexonaut · February 25, 2021, 7:04pm

Although there is no firm timeline of when the flash mint module is going in, I would like to poll for the initial mint fee in advance so we are not blocked on this. Please note that debt ceiling does not need to be set to any number in particular, so we are going to roll out with something big enough to cover most use-cases (100M or so). There are a couple considerations in whether we want a fee or not, and if so what that fee should be. I will enumerate them below:

Competition

There are two main flash loan lenders currently - dYdX and Aave. Aave currently has the most liquidity, but they charge a fee of 0.09%. dYdX is free, but they lack liquidity. Here are the pool sizes as of writing this:

Aave (0.09% fee)

DAI - 9M
USDC - 7M
ETH - 300M

dYdX (Free)

DAI - 138k
USDC - 2.2M
ETH - 185M

As you can see there is a serious lack of stablecoin liquidity in particular, so we have an opportunity to be the go-to source of liquidity for stablecoin flash loans moving forward. According to this site, Aave has done 2B in flash loan volume for a total profit of 1.9M in fees over the lifetime of the project.

I would suggest we should probably use the 0.09% as an upper bound for the fee to pick, but I’ll include a higher option as we do have access to unlimited Dai liquidity.

Market Efficiency

The closer the fee is to 0%, the more arbitrage benefit we get from the module. This should help to improve DAI’s standing as a trading pair and improve peg stability. This effect may actually be worth more to the protocol then the ~1M or so in expected fees at ~0.09%.

Public Relations

Coming out with a 0% fee flash mint module for the good of the ecosystem is a good PR move. We will be the go-to place for most flash loans as we have unlimited liquidity at 0% fees, and it is a good look for people who are interested in democratization of financial arbitrage. A 0% fee removes capital barriers of entry to users, and allows that kid learning to code on Ethereum the ability to construct profitable trades without having to compete against those with more capital.

Planning Ahead

Although Aave is currently making fees on flash loans, I believe the days of being able to charge a fee will be limited. WETH10 is already planned with a built in (free) flash mint feature, and it seems likely to me that other newly created tokens may include this as there is no extra cost. Market forces should drive the fee to 0% on all flash loan/mint features in the long run as there is no cost to operate them.

What should the initial dss-flash fee be?

0.00%
0.05%
0.09% (Aave Fee)
0.10%
Higher than 0.10%
Abstain

0 voters

This poll will run until Thursday, February 18th to be included in an on-chain vote the following Monday, February 22nd if any options gain >= 50% of the vote (ignoring Abstains). A ranked choice vote will be used if multiple options are >= 50%.

alexis · February 10, 2021, 8:55pm

0.01% is that possible

ultraschuppi · February 10, 2021, 9:12pm

Voted for 0%

No risk, great PR and we are the only Lego that could do that.

SebVentures · February 10, 2021, 9:39pm

I can see that dYdx is not really a competitor and at 0.05% we still get >98% of the business.

Moreover, free flash loans might make a lot of arbitrages between DEX profitable. Not sure increasing the load on Ethereum is good PR. Gas is high enough.

Uncle · February 10, 2021, 10:11pm

0.05-0.09% for unknown-unknown risks

MakerMan · February 10, 2021, 10:25pm

If one is going to do this and compete I usually don’t go for 1/2 but in this case I’d say .05%

hinchliff · February 10, 2021, 10:29pm

I agree that charging fees for flash loans will soon be a thing of the past. We are better off having zero fees and good PR than a lukewarm launch only to drop the fees to zero soon after.

alexis · February 10, 2021, 10:30pm

Unfortunately nothing is free in this world.

LongForWisdom · February 10, 2021, 11:00pm

This feels like it might be an important point. Flash loans are great because they improve arbitrage efficiency across different on-chain platforms. But… In the current environment increasing arbitrage efficiency may mean more transactions as people use flashloans to take advantage of ever-smaller margins.

That said, this is a function of available flash-loan liquidity as well as price. Infinite liquidity at zero fees would mean the only cost of arbitrage is gas, which would increase the number of viable arbitrage transactions as gas prices decreased.

So, maybe there is some compromise where we keep fees artificially high, or keep available liquidity artificially low in order to minimize this.

Or, on the other hand, maybe it’s inevitable that someone offers low-fee high-liquidity flash-loans/mints at some point soon anyway, in which case it may as well be us.

Jiecut · February 11, 2021, 12:48am

I don’t really understand the PR benefit of 0% fees.

swakya · February 11, 2021, 1:28am

Am I wrong to think there is still no barrier to entry with a > 0% fee, you just exclude arbitrages that are less than >fee profitable?

hexonaut · February 11, 2021, 1:33am

I was more thinking that the pool of developers who can compete on any given opportunity will increase with less fees. In the extreme opposite case where we set the fees to 100% or something crazy then arbitrage returns entirely to those with capital as the flash loans are completely uncompetitive.

Jiecut · February 11, 2021, 5:59am

With fee wars, a 0% fee eventually results with miners just making more money.

ultraschuppi · February 11, 2021, 6:23am

It is not only about arbing, it is also about offering really cool usecases that other apps already do: leverage or payback with flashloaned DAI so you don`t need to spread your activity over multiple transactions. it would implement some of them on our own, it seem greedy if we do them with 0%

Aaron_Bartsch · February 11, 2021, 1:51pm

I see it as being easier to lower fees later once we see how it is received. It’s a lot harder from a PR/market position to raise. Would be nice to collect some income from our (@hexonaut) work.

mrabino1 · February 11, 2021, 2:58pm

This is great progress on this module.

Going back, I do want to stress the incredible power this will bring to DAI (and those that will utilize flash loans). AND with this great power comes responsibility. Maker already went through iterations to defend against flash loan governance attack angles. I strongly suspect other DAOs have not implemented such measures.

To that end, I recommend to start off with a “sane” max flash debt ceiling that is comparable to dYdX. I do believe it wise to scale that debt ceiling over time. Starting off at infinity leaves the door open to possible damage to other DAOs.

SebVentures · February 11, 2021, 4:07pm

I also wonder if we shouldn’t make 2 modules. One for everyone with a fee and a somewhat low DC (to be defined). And another one for Maker ecosystem projects like DefiSaver, Instadapp stuff like that. As they provide nice tools to our customers we should not take a fee on that.

Aaron_Bartsch · February 11, 2021, 4:13pm

Whitelisted flash mint sounds nice.

hexonaut · February 25, 2021, 7:09pm

So the poll ended with 0.05% as the winner. We will look to get the flash mint module up in the next few weeks, but I can’t offer a firm timeline with higher priority stuff such as MIP21, MIP22 and MIP45 taking precedence.