Ethereum Macroeconomics via Dynamics

Transcript

be a technical talk. Okay. Um, hello. I'm Daniellea from 20 squares. We're an economics research firm, R&D firm. Uh, and we have worked in the Ethereum ecosystem for a long time. And today I want to tell you a bit about this um, uh, very interesting piece of research we have worked on uh, that was supported by the Ethereum Foundation. Uh I mean in general last year we did a lot of research on uh Ethereum insurance. I mean it was the talk of the town right? Yeah. Inflation.

Um so this talk will be about Huh. Okay. Yeah. That part wasn't showing up. Okay. So this talk will be uh about uh like it will be an a bridged version of uh uh complete report we wrote. Uh so my awesome colleague Eric who's here um has wrote a very long- winded blog post uh about uh the results of this research and uh what's that about okay so we used a stock and flow model and if you don't know what stock and flow models are uh don't worry I will tell you in a bit a simplified version obviously uh

uh of the Ethereum staking dynamics the staking flaw and um we have looked at three uh big problems that are looming over Ethereum's future. Okay, one of these is runaway staking. So the tendency towards a scenario in which near 100% of the ether are staked and the chain is very liquid. Uh the second one which is the thing that's most interesting for uh like most of the people that have been concerned with changes to the isance curve is excessive inflation or runaway inflation. Um and

then also I mean this is uh uh like I think that doesn't get talked about that much. I mean in in the context of Ethereum issuance but governance centralization is a big problem that's related to that and uh yeah so by governance centralization I don't mean physical centralization so um a small number of validators actual validators or like um uh uh all the validators concentrated in a in a given geographical place I in a very small amount of economic entities that are

holding all the st. Okay. Uh but today we're going to focus on the first one, runaway staking. Uh I mean the the second one is kind of built in, but let's focus on runaway staking. Okay. So here's a graph of the Ethereum staking model. This is a stock and flow model. I will also give you the other parts but uh this is pretty much accurate actually. So what is a stock and flow model? Okay, it's uh uh so you can imagine these places the this uh circles as tanks that are holding a liquid and in this case the liquid is

all the ether that was ever minted. And uh these uh arrows are just tubes between these tanks with oneway valves. So they only move in one direction. Okay. Uh so the tanks or the stocks here are uh well U is unstaked ether. S is staked ether and O is the inaccessible portion of either. So either burned or slashed or lost the key. Um the flows here are so here we have this F uh which is uh reinvestment no sorry it's it's um fees okay so fees uh like uh and also me

you can include me in it if you want but it's not important uh so it's basically whatever validators get out of transaction fees uh Q plus is new stake Q minus is well unstaked ETH and R is the reinvestment of like validator rewards. Uh actually this would be a bit more accurate if we add another uh smaller tank that's inside U let's call it V which is uh the validator rewards. Okay. And uh this B and J were just so B that goes from unstake ETH to inaccessible ETH is just uh uh like

EIP559 burn. Okay. And J is lashing. Okay. So all these things are subject to some constraints. Here's a nice table with names and the domain domain and co-omains. As you can see the B I told you about the the subset of U is uh here and uh I mean these are very reasonable constraints. I'm not going to explain all of them. Uh but uh so well one another important part is this is not a closed system. Okay here this dashed arrow I comes from nothing. Okay. And this is the

inflationary part of of Ethereum. This is issuance. Okay, it comes from the outside. It it adds more liquid to the tanks. So these constraints are reasonable to have like you can read them in the blog post by the way. So let let's just go through one. Okay. So base fees plus priority fees so are equal to the total fee that is uh total fees that are paid for a transaction uh have to be less than all the unstake ether otherwise I mean this wouldn't run uh it's a very reasonable thing to assume right um and uh be careful uh all these flows are considered over moving averages over

spot values at a given block. So in the in our research we picked quarterly um averages but I mean whatever is reasonable to you. Um okay so well okay some parts are not showing up in the slides. There should be dots over uh e a v whatever because these are differential equations right? the uh for example let's read the first one the first law that actually regulates uh the Ethereum staking model is that total ether okay ever minted the variation of that

there should be a dot like a time derivative over the E okay the change in total ether over time is just given by isance so I was as I was telling you this is not a closed model and this gives you like what gets added or I mean another reasonable one is the change in accessible ETH so state plus and state is equals to um well um issuance minus burn minus slashing. These are very reasonable things to assume. Uh one thing that could be added is actually stuff from um EIP7514

which uh actually poses an upper limit on churn rate. uh but uh the point of this research was also to look at uh uh like a case in which we don't have that because this is uh uh an EIP that was introduced to mitigate some of the effects of very short-term runaway staking. Uh we also make some more assumptions on this model. The first one is that um like the yield out of is sublinear and this is important in order to avoid discouragement attacks and if you don't know what discouragement attacks are uh

these are a very important and crucial point when talking about issuance and I suggest you go and read Vitalik's paper on discouragement attacks. Um I mean this is a an actual elephant in the room and also we assume that uh basically our moving averages do not do not differ uh a lot from spot values which is another reasonable thing to assume. Um so how do how did we actually analyze the thing? Well we built open source library. Okay you can go on GitHub uh and and play with it. It's a wrapper around differential equation solvers because these things are

dynamical systems and it's specifically built for uh this kind of microeconomic analysis of Ethereum. Okay. Uh and I mean myself and my colleagues at 20 squares firmly believe this is the way to approach issuance research and in general macroeconomics research in Ethereum. uh it shouldn't be Ethereum economics shouldn't be by based okay we actually need to look into the actual effects of stuff with a like in a precise and scientific manner before pushing I don't know quasan issuance okay or deflationary e

uh so please go and play with this learn about this stuff it's important So from our research we have u like figured out some very important uh way to go about uh issuance. Okay. Um so if you keep medium reinvestment rate, low inflation, okay, it has to be positive but low and even lower fees then it's more likely okay that we will converge to a a feature that's uh in which we don't have super non super long

horizon runaway staking. Okay, we will see that like we can push the problem down the line but it's like the problem is kind of baked in. Um so why low inflation though? Okay so in the case of strong deflation so this is for all of you who care about your bags and want to reduce Ethereum insurance because this will appreciate your bags. I mean in this case uh uh like in the case of strong deflation in which you have uh like more deflation than the reinvestment rate uh this is very unstable uh that that gives you a very

unstable dynamical system and in the case of zero or very weak deflation so still not positive. Okay. uh then uh uh the tendency towards runaway staking can be moderated only by either eye churn rate or slashing which is undesirable. I mean we don't want validators to be slashed. So in short, okay, runaway staking that's not super long-term can be avoided only while inflation is at the same time low enough that concerns over inflation obviously uh do not dominate their investment rate

and at the same time high enough not to be dominated by their investment of priority fees. So, it has to be like low but not too low. Um, so here's the important question though. I I I'm sure you're all here for for this. So, should we reduce inflation? The short answer is no. No, this is very delicate. Okay. uh I mean uh before thorough investigation which by the way this research was part of we should like do more research on this kind of topics

because this is what keeps Ethereum running right the economics uh we shouldn't touch this stuff okay we shouldn't push for strong deflationary ETH or quasan issuance or whatever without first uh having a thorough look into it. Okay, but Danielle, uh I really care about my bags. Maybe we can decrease uh inflation issuance and uh we will have the magic EIP7514 that poses limits on the churn rate uh save us. Well, not really actually.

Okay. So, uh and yeah, so uh the the the problem is that this is designed to tackle the problem of runaway staking only on over very short time horizons. So, spikes of either mass exit or mass entering in the validator set. uh and in general as I was telling you earlier near 100% uh staking is uh kind of uh uh like baked in okay uh for our current understanding of issuance in which we have a static ission curb um so but if you're interested I mean the repo is open there's also the uh

notebooks we used for uh doing our analysis and I suggest you go and play with it or play with whatever else uh like way you have to model uh Ethereum microeconomics and play with the constants in EIP7514. Maybe something will come out of it. Uh personally I have some ideas around how to tackle this problem with dynamically adjusted curves. So not a static curve. We will see if you're interested come to talk with me about this. Um so the talk is over. Thank you for coming and yeah feel free to ask any question.

Thank you very much. Do we have any question? Oh, I come. Yeah. When when you say that a 100% ETH will be staked long term, maybe it's best to explain um what you mean by that. Like what what does that mean in in uh uh who would be staking it? um why does or or who who are the big players maybe why don't you tell us

about that I think I I mean uh this is very difficult to decouple from very strong economic centralization right so there will be a very small number of large economic players uh that uh have uh all this uh ETH in stake I mean 100% ETH staked is very catastrophic IC scenario but like near 100% is a future we are actually starting to get accustomed to right so for example think about the L2 centric future that is being pushed on uh like the uh public opinion over Ethereum uh so the idea is okay a lot of ETH is

staked the chain is very liquid and what's like the the the very little liquidity we have left over is used to settle uh like uh for L2s. Okay. Um for example, blob prices or whatever like roll up. Thank you for the questions. Yes, welcome. Hi. So um the longer I'm working or looking at like the whole crypto uh I see more and more parallels to the traditional finance. So how much of your analysis can be traced back to the traditional

finance or did you um yeah how how how do you compare those two if you're looking in like inflation and all those? Yeah, I I mean obviously dynamical systems modeling is uh widespread in finance like traditional finance. Obviously the staking model is totally different because uh I mean the whole I mean I mean from one point of view we're looking at a like the the intersection of the interesting problems that uh we we're looking at is uh very thin. Okay, we care about inflation but also we care about the all these other problems which

would kill Ethereum, right? Uh so there's there's some overlap but it's very little because you have to change stuff around to accommodate uh like research about Ethereum like the the the the whole model. I mean blockchain is a whole different thing from Trafi even economically and technically. So that's thank you. Thank you very much. Uh I guess first off, awesome talk.

Thank you. Thank you. Um I I guess in theory my my question would be around if more and more Ethereum goes into staking. My assumption would be that yields would need to come down which would make it kind of less people would be less motivated to stake at that level. Is this something that you're seeing in your data or uh c can you can sorry can you raise your voice sure sure um sorry about that no don't worry as staking uh nears like 100% my assumption would be that yields would need to come down yes which would push me out of it but even looking at the yield curve this is not uh like discouraged by over long long time horizons okay we're talking

about uh yeah so from what you're seeing like you don't think the yield curve would come down enough to demotivate people from moving out of staking. I mean the current yield curve no uh I mean this uh uh see like reducing issuance doesn't solve it either. So this is the important part I mean this is this comes from the actual simulations. Yes. All right. Thank you. I'm very grateful for your work. Thank you. Thank you. And now we can finish the presentation and uh we will back at 1 p 1 p.m. Thank you.

[Music]