[Discussion] Adjusting the BOX Parameter

[OneBiteLawnOrder]

Given the exponential growth of DAI outstanding and the value of ETH I wanted to get a conversation going about the Pros/Cons of adjusting the Box Parameter. However, I don’t think we are quite ready for a signal request before we hear more from Risk on this subject.

I don’t have a good sense of the model that was used to select the current BOX of 15,000,000 but I assume that when the debt outstanding and liquidity of DAI have both substantially increased that this would also justify increasing the BOX. Given that we have over 600 Million DAI from ETH-A alone if we have a vicious pullback/deleveraging event and have substantial liquidations occurring a la Black Thursday there is no way that 15,000,000 will be enough to clear auctions with a 6HR TTL without losses.

Happy to hear what others think and hopefully we can come up with a range of possible BOX values for a future signal request.

[Jiecut]

I agree that the increased Debt exposure and ETH-B debt necessitates a higher BOX size. Currently the BOX is set to 1.9% of outstanding debt (excluding stablecoins).

One counterargument is Keeper liquidity. It is hard to quantify. And if there’s not enough keeper liquidity auctions can fail.

And my counter-counter argument to that is that you can also rely on manual auction bids during times of extreme volatility. There’s just needs to be enough publicity and an easy way to do it. There’s 20B TVL in DeFI, and Billions of stablecoins. There’s a lot more on-chain liquidity/capital than in March.

So to pull a starting number. 50M DAI for the size of the box. That’s 1,000 auctions with a 50k lot size. 6.2% of the current outstanding debt. 3.7% of outstanding DAI. And it’s the size of the ETH-B vault.

Anyways, that’s my 2 DAI. I’ll try to shut up. There’s definitely real risks on both sides, and bad things could happen wherever we set the BOX size. (On one side, imagine MCD with a 0 DAI box parameter. On the other side 0 DAI bids).

Also, I wish there was an hourly throttle instead of a 6h throughput throttle. (This isn’t a comment regarding TTL)

[g_dip]

I think everyone shares your frustrations around the auction time and the box parameter. That being said, I think it’s an incredibly bad idea to change this variable until liquidations 2.0 is live. The Keeper ecosystem is likely worse than it was even a few months ago due to the lack of significant liquidations (traders do not run complex idle infrastructure). I think the limited ecosystem of Keepers that we have would oppose any reduction in the timeframe as they need every hour can they can get to source the Dai.

Just keep in mind that the potential outcome to a higher box or shorter auction timeframe is zero-bids…

Edit: RE: ETH-B, if that’s an issue I’d personally prefer to just cut it back and wait for liq2.0 to scale it up again

[OneBiteLawnOrder]

It’s a good point about keeper liquidity, and I am not suggesting changing the TTL. I see the reasons for the longer TTL and it has so far worked fairly well for us. I’m more thinking that maybe we can increase the BOX given the increase of ETH value, greater DAI liquidity and currently (for now) fixed peg.

It is simply easier to source DAI now than it was when we first implemented the current liquidation system. My understanding was that part of the rationale behind the 15,000,000 BOX was that even assuming keepers had the capital necessary, it may be difficult for them to source this much DAI within the TTL (also why we implemented USDC-B). In that respect the situation looks different now and may warrant a change.

[g_dip]

@OneBiteLawnOrder

It is simply easier to source DAI now than it was when we first implemented

The issue, at least for the Keepers that I know of, is not the liquidity of Dai - it’s their own balance sheets. They don’t have more than 15M to deploy. If the community could demonstrate a well-capitalized keeper that makes a commitment to running the infrastructure and bidding on auctions, I’d definitely change my mind.

[OneBiteLawnOrder]

Makes sense. I understand the constraint there. I guess this begs the further question of whether Keeper balance sheets may have similarly scaled due to the recent ETH appreciation. My guess is that they have but I could be wrong.

Either way I feel like we need to take a dive and increase the BOX at least somewhat. I realize that this is just a temporary problem until Liquidations 2.0 but I just think we can’t keep expanding the DAI supply without scaling BOX as well. If we can’t increase BOX maybe we have to stop increasing debt ceilings until liquidations 2.0 is available… I don’t think anyone wants that as a solution.

[LongForWisdom]

This is a little outside my area, but I will say that the likelihood and severity of loss with a too-high box seems much greater to me than with a too-low box.

Multiple things have to go fairly badly wrong for a too-low box to cause us serious damage.

• We’d need to fill and exceed the box.
• And have collateral prices continue to decline to less than 50% (ish, for ETH-A, assuming most debt is 200% backed, ETH-B probably loses us money regardless if it gets liquidations in a crash scenario)
• And get to a point where the lost funds from debt at <100% collateralization at time of auction outweighed the income from the liquidation penalty on the debt at >100% collateralization at time of auction.

For a too-high box to go badly for Maker, a lot less has to go wrong.

• We’d need to exceed keeper capacity.
• That’s it.

I feel like the payoff matrix is something like this (just defining scores as relative to each other.)

	Light Liquidations	Medium Liquidations	High Liquidations
Too-low box	+1	+1	-3
Just-Right box	+1	+1	-1
Too-high box	+1	-1	-2

[g_dip]

I even think this payoff matrix understates the risk of too-high box. One zero-bid is potentially equivalent to 100 liquidations at 99% of the Dai debt. I actually think it would be exponentially worse, because under too-high box it would be even more zero-bids.

[OneBiteLawnOrder]

Got it. That makes sense to me that the biggest risk to the system is zero-bids. Having auctions kicked at a lower price 6 hours later is not the end of the world in comparison. So keeping the Box a little too low is way better than keeping the box a little too high.
This is exactly what I was hoping to learn from this thread so thank you!

[Primoz]

I agree with the discussion held here, but just to be clear, the main problem is that we still don’t have a good benchmark of how much capital keepers bidding in Maker auctions have. The 15m box that was voted was originally was proposed when @g_dip made some concerns few months ago when talking to some keepers. And this was the only input we got at the time.

If we were to take this more professionally, we should take a look at all the addresses that are bidding in FLIP auctions and try to evaluate how much capital they hold. I somehow feel keepers have more capital these days than 4 months ago. But we would also need to check if these keepers also bid in other auctions at Compound and Aave which are happening 1 hour prior to Maker’s. It’s not an easy task but worth doing it. Maybe our on-chain expert @lix could help us here.

Alternatively we could invite keepers to reveal their reserves anonymously and propose this number but in that case we would need to trust them and always be a bit conservative. Hostile keepers would want a higher box number so they have a better chance of 0 bids.

[g_dip]

I think this is a great idea. My only point has been that I would not want to see any change without this kind of independent due diligence.

[Jiecut]

I was thinking the Flopper could also benefit from a BOX parameter.