Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Ava Labs x CBER Ep 15: Contagion in Decentralized Lending Protocols
38 Plays4 days ago
This episode examines Decentralized Lending Protocols (DLPs) and the contagion that can arise due to assets lent to one pool serving as collateral in a different pool. We discuss efficient design of DLPs and mitigations for managing risk.Guest: Julien Prat (CNRS Researcher CREST, Ecole Polytechnique, IP Paris)Paper: Shock Propagation in Decentralized Lending Networks: Evidence from the Compound Protocol
Recommended
Transcript
Introduction to the Podcast Series
00:00:06
Speaker
Hi everybody and welcome to Crafting the Crypto Economy. I am Silvia Sanchez from the Avalanche Policy Coalition and today we bring you a special podcast series in partnership with the Crypto and Blockchain Economic Research Forum, also known as the CBER Forum.
00:00:22
Speaker
This is season 3 of 5 new episodes featuring leading faculty from global universities exploring various elements in the blockchain ecosystem. From stablecoins to DAOs and so much more.
00:00:33
Speaker
These episodes are a bit longer from our usual ones since we will be getting deeper. But again, it's kind of challenging to fit in a whole research paper into just one hour. But don't worry, each episode will have its accompanying research paper posted on the website and episode description for further reading.
Decentralized Lending: An Overview
00:00:49
Speaker
Anyway, I'll hand it over to our moderators, Professors Fahad Saleh and Andreas Park. We hope you enjoy.
00:00:58
Speaker
Hi everybody and welcome to another edition of the Crafting a Crypto Economy podcast presented by OWL Explained and the Crypto and Blockchain Economics Research Forum. We're now in our third season and my co-host Fahad Saleh and I'm very happy to have Julian Pratt from รcole Polytechnique in Paris and Stefania Macassa from Sergis Paris ah University um who will be talking to us about contagion and decentralized lending protocols.
00:01:26
Speaker
um We spent a fair bit of time in our previous episodes talking about the structure of blockchain and then also on on decentralized trading. But as ah as a market, as ah if you think about crafting a crypto economy, in particular, financial ecosystem, borrowing and lending is is of first-order importance. And so we're really happy to have the opportunity to talk to you guys about your work on this space.
00:01:48
Speaker
um Even though our audience is most likely reasonably well informed about you know different DeFi applications and the like, I think it's still a good idea if we maybe start by if you you explaining to us how decentralized lending actually works and what we're really talking about here. So um I'm not sure, Stefania or ah Julian, if one of you could just lay out, say, maybe the workflow of if somebody wants to make a loan in, say, Compound,
00:02:16
Speaker
um you know which for a while was the biggest ah protocol, and as as I understand you're working on this. you just explain how that works? so Hi everyone, thanks Andreas for the introduction.
00:02:26
Speaker
So yes, I'd be happy to present the way lending protocols work. So I will focus on Compound, which is actually one of the oldest lending protocol and has served as a template for many other protocols, I think it's a good introduction to the field.
00:02:41
Speaker
So you should think of lending protocol as automated report che agreement market automated repo market. It's for short-term lending where you will have your collateral will be mostly liquid. no So on one side of the market, you have lenders, obviously.
00:02:59
Speaker
They deposit their tokens inside what we call liquidity pools, and each liquidity pool contains only one type of asset. So for instance, you will have a liquidity pool for EFER, or liquidity pool for USDC, et cetera, et cetera, right?
00:03:14
Speaker
Now, on the other side of the market, not surprisingly, you have the borrowers. And the first thing they have to do is deposit collateral. So they will have to become lender before borrowing, actually.
00:03:26
Speaker
So first they deposit some collateral in a lending pool, which might be loaned out, and we come back to that later. up In exchange, they get a token which act as a certificate of deposit for their loan.
00:03:38
Speaker
Okay? Now they can use this token as collateral to borrow in another pool, typically. So let's be concrete. You have some stablecoin USDC. You deposit them in the USDC liquidity pool. You get some what in compound is called a CUSDC. Okay, C for compound, but it's a certificate of de deposit essentially.
00:03:59
Speaker
Now you take your CUSDC, you go back to a smart contract and you say, now i would like to borrow some ETHER. and and you use And then the smart contract will lock your CUSDC as collateral.
00:04:12
Speaker
And the last thing I want, so that's very simple, I guess, very simple design. The last thing to understand is that because we are operating in a pseudonymous environment, of course, the value of your collateral has to exceed the value of what you borrow. Otherwise, you will take it and never come back. No, that's quite obvious.
00:04:29
Speaker
So that's one limitation of this protocol is that you always have to put more collateral than what you borrow. Okay, and so now, i mean, one of the key questions when you have these markets is the price. Can you explain to us how the price is being determined?
00:04:43
Speaker
So the price of what? The price of the token or the price of the loan? The price of the loan, right? Because it's borrowing lending. Okay. Obviously, a basic issue is like, let's say i want to borrow Ether against ah stable coins.
00:04:55
Speaker
I need to know the price of Ether in dollars, right? To make sure that it's over collateralized. So you have all always oracles that feed the price. all the time to make sure that the loan are indeed over collateralized. And that could be a problem, not technically, because if the oracle breaks, you might attack the protocol.
00:05:13
Speaker
And actually a lot of attacks in DeFi, many are attacks in DeFi, have occurred exactly, not inside compound, but overlanding protocol by manipulating the oracles for the price. Now your question, I think, was referring to the price of the loan itself, so at which interest rate I borrow.
00:05:30
Speaker
So when you lend, you get an interest rate, and when you borrow, you pay an interest rate. Obviously, you will pay higher interest rates than the one that is given to the lender because protocol has to make some profit. That's obvious. Now the question is, what is the level of this interest rate?
00:05:45
Speaker
It's fixed algorithmically inside the smart contract of the protocol, right? So now remember, you have this liquidity pool. Let's focus on USDC. So people have deposited some USDC, they have loaned out some USDC.
00:05:59
Speaker
Now other people come and borrow the USDC. So the pool is not full, no? So that will be the utilization ratio. So let's say 50% of the token have been borrowed.
00:06:11
Speaker
Now you will associate an interest rate to this utilization ratio. The higher utilization ratio, the higher the interest rate. So basically the mo so more empty the pool is, the higher interest rate you will have to pay.
00:06:26
Speaker
And the idea being that you want to avoid people borrowing and depleting the full pool entirely. So typically what you get is that you get close to full utilization, meaning that almost all the tokens have been borrowed, the interest rates to shut up very high so as to discourage people from borrowing and encouraging other people from loaning out so as to refill the
Interest Rates and Governance in Lending Protocols
00:06:48
Speaker
pool.
00:06:48
Speaker
So, you know, just ah formally and structurally, is this a piecewise linear function or is this an exponential function? How should I think about it? I think that piecewise linear function would like to be exponential.
00:07:02
Speaker
So let's you're right. yeah So it it's coded in the smart contract, right? So, and that's the beauty of it because a smart contract only has to look at the utilization ratio and then you can associate an interest rate. So that's very simple and that costs little gas. It's quite efficient.
00:07:19
Speaker
Now, you would like something which will look exponential and you would like to discret, but it's difficult to have a function that is not, ah you know, that is not, that is complex. because it costs more gas.
00:07:30
Speaker
So one way to approximate it is to do it piecewise, very flat, and then around 80 or 90%, you have a kink, and then it book it becomes almost vertical. And that's the way they implement it. But I think the conceptual idea behind it is exponential.
00:07:44
Speaker
The practical way it's done is piecewise. Okay. um Now, I know we want to talk about you know some of the fine details of the operations of the smart contract, but just before we go, and there's a lot of parameters in here, right? as So, okay, so you're saying, you know, if you have different loans or different different types of tokens, they have different prices, you need to feed the price. And, okay, that's done by an Oracle. Fine.
00:08:06
Speaker
um But then there's also, that you know, there's, I assume when you have ah a function, there's a slope. And if you say there's a kink, You have to determine where the kink is, and then you have to determine how much you can borrow against the collateral. So there's lots of different parameters. Who sets these parameters? Where they're coming from?
00:08:23
Speaker
So they are set by the governance, and that's why governance token may have a value. And actually, lending protocol is probably one of the type of protocol where I think governance is actually quite valuable because you might determine the slope of this curve, which determines the interest rate. You might determine the collateralization ratio.
00:08:41
Speaker
So these are the important parameters. And not only is that, if you look in compound, you have liquidity pools. But you cannot create a liquidity pool. If you think of an IMM, if you think of Uniswap, anybody can create a um ah pool for swapping. It's open.
00:08:58
Speaker
In compound, it's not. It has to be accepted by the governments. And there are many less pools. There are like 19 or 20, or I think in AVA, around 40. And there are good reasons. We'll come back to that later because it creates risk. now So you don't want anybody to create pools because that might be problematic in the end.
00:09:14
Speaker
So governance is basically all these things are decided by governance. So as you said, one example is the collateralization ratio. As I said, you always need more collateral than what you borrow. But how much more? 20%, 25%, 30%, 40%. Of course, it should depend on the volatility of the asset.
00:09:31
Speaker
But, you know, It has to be determined. So it's it's by vote. Okay. So essentially, it's like the people that own the bank actually run the bank too, right? So it's the opposite of what we have in delegated version where we have a CEO making decisions. If you're a token holder, you get to actively participate.
00:09:49
Speaker
You can see it like that. It's a democratic bank. it's it's very it's it's a nice It's a nice example of actually how these DAOs, decentralized autonomous organizations, can actually you know really do something. I agree. I think governance token, sometimes you wonder if if people care about governance. But here think it's a case where it's clearly important what they are voting on.
00:10:12
Speaker
Now, I have some questions. So one thing that you mentioned is, okay, so you can borrow against a collateral. and um And there's a limit how much you can borrow against each unit of collateral, let's say, right? It depends on the on the current market price or the value of the collateral.
00:10:27
Speaker
And so um so just just to be clear, is everything denominated in in the dollar? Do we denominated in Ether? what's the What's the base currency, the unit of account that we use here?
00:10:38
Speaker
The dollar. The So these but these protocols are already dollar-wise. Excellent. I think so. i'm quick not Maybe i have a doubt, but I'm quite convinced
Liquidation Processes and Flash Loans
00:10:48
Speaker
it's correct. It's dollar.
00:10:49
Speaker
I must be very much to the dismay of Christine Legard. um So, um but, okay, so ah what happens, so you know, one of the things we always worry about is when you have something where you borrow and you have a collateral is when you get a margin call, right? So what what happens if the value of the collateral drops ah beyond beyond, let's say, it doesn't have to be undercover, but beyond a certain threshold. So what what happens now?
00:11:12
Speaker
So I have a loan, I put a collateral, let's say I put them down $1,000 worth of collateral and my collateral value, I take a loan up, let's say $800 against $1,000. And now the value of my collateral drops to $850. So it's very close to to not being fully ah fully collateralized. What what happens?
00:11:31
Speaker
so And it's quite likely to happen, by the way, because crypto are very volatile. So if you collateralize with ETHER, ETHER can drop by 5% in a day easily, you know. So exactly. So what happens, let's say... um So first, as we said, you have over-collateralization ratio, which is set by the protocol. So let's say it's 20%, right?
00:11:49
Speaker
Now you borrow from ah USDC, and you have put as collateral ETHER. And now let's say you have $130 in... a collateral and then you wake up in the morning and actually if your eyes drop and now it's worth 115.
00:12:05
Speaker
So in the language of this protocol it's your it's called your health ratio and your health ratio should be higher than one. so it's which means so it's taking into account of our collateralization to be clear so now your health ratio is lower world than one which means you're open to liquidation.
00:12:22
Speaker
Your position ah become open to liquidation and everybody can liquidate you. By liquidating is meaning closing your position, right? So in this case, remember you borrowed some USDC.
00:12:37
Speaker
Now someone can come and put some USDC back in the protocol, basically repaying your loan, and they will get your collateral. And they get your collateral at a discount, obviously, because otherwise why should they do it? And they get typically 5% discount.
00:12:51
Speaker
So what you have to do, so they don't really have margin call in the sense that they tell you, like, you know, wake up, refill. You should watch your position. And once you cross the threshold, it's open and it's decentralized. So anybody can do it. So not anybody in practice, because there are big arbitrage opportunities.
00:13:10
Speaker
Now, if you're not a professional, you're not going to make money out of liquidation, right? Because there are boats running and taking advantage of this. Yeah, I've seen that actually. So I've seen online, there's actually websites where you can see Python code to and to to run liquidation, but I think that's probably too slow by now, right? You have to be a little quicker than that. Yeah, so I think a few years ago you could, but now it's professionalized.
00:13:30
Speaker
So could we kind of like walk through what could go wrong here? in the sense of because yeah you could you could think of it in a very simplistic way and say like okay so i have to over collateralize my loan um but uh but essentially my loan to value could change uh in an adverse way but at the same time before it gets to the the situation that i'm uh that that that ah that that my collateral is sort of no longer enough, I can be liquidated.
00:14:00
Speaker
um So what could go wrong there? And actually, maybe a question to slightly precede that is, what is the incentive for the ah liquidators to liquidate? So what could go wrong for the borrower or for the protocol, for the borrower, for the user? I really mean for the protocol at the end of the day.
00:14:17
Speaker
Okay. So what could go wrong? So, okay, just let me just say quickly for the user, because I guess matters too. So for the user, what could go wrong is that, as I said, you're sleeping at the wheel.
00:14:28
Speaker
And the ah there can be actually sometimes some glitch in the pricing. You know, like sometime maybe the price might drop and if your position is ah getting liquidated very quickly, you might end up losing a lot of money.
00:14:41
Speaker
And actually, ah we found some example of that where someone was very high. Then there was some like very quick drop in the price reported by the Oracle, maybe because there was liquidity issue in the ah in the centralized market and then you get liquidated.
00:14:56
Speaker
Right? So that this kind of thing can go wrong. And we have an example. So, well, someone lost millions like this. So that's what can go wrong. Now, in practice, normally you should watch your position and you should always be able to avoid liquidation. You can just do a flash loan, no?
00:15:12
Speaker
Because it's at 5%. So normally you should go at a more interesting rate. than the rate of the discount. So you should always have about been liquidated. now that's theoretical. In practice, the market might not be liquid enough for you to find the collateral when you need it, to refill your position. There might be issues like that.
00:15:29
Speaker
ah Hold on, I have to interrupt you. So we we have to come back to the topic of a flash loan, which you just mentioned on passing, right? So just to make sure that everybody in the audience understands that. But maybe we can punt that for a moment, just that everybody knows we have to come back and explain what that is.
00:15:42
Speaker
So you want me to explain? and what yeah Why not? why I would add it. Just quickly explain. What is a flash loan? So flash loan is a DeFi voodoo, actually. I mean, it's one thing you can do in DeFi and you cannot do anywhere else. No, don't you agree on the rest?
00:15:54
Speaker
So flash loan is when you borrow. Basically, you don't really need collateral because you repay the same transaction. Okay. So let's take the example where you want to close a position.
00:16:05
Speaker
You know that if you close the position, you're going to get a discount of 5%, right? So what do you do? You go borrow, so maybe you can do it in Compound, actually, in Compound itself. So you go, to go back to our example, you had to borrow some USDC to close the position, right?
00:16:22
Speaker
So you borrow some USDC, you close the position, you get the effort, Then you go on and on an AMM, decentralized exchange, you exchange this against USDC and you repay your loan, right?
00:16:37
Speaker
And you will make money because you get your yeah you get a 5% discount when you liquidated the position, right? So in principle, you should be able to do a flash loan to close it.
00:16:47
Speaker
And the advantage of a flash loan because it's instantaneous, right? The interest rate is close to zero, it's negligible, and you don't need collateral. So many people should do that. Now, if you have a huge i'm coming father if you have a huge position, but you might need millions, and you might not have the the liquidity to do that.
00:17:06
Speaker
But my point is that this shows that you should avoid liquidation if you're not sleeping at the wheel. But sometimes the markets are not liquid enough, or maybe everybody wants to liquidate together. um Or maybe you, and what we have found in our analysis is that some users are not watching their position enough.
00:17:23
Speaker
So they seem to be sleeping at the wheel and and and lose a lot of money sometimes. So so just just to be clear before we go into this, so this is, I mean, you called it voodoo, I think as a finance professional, I would say this is like essentially the the grease that you need for markets to work. This is the perfect thing where you want to make sure that if you need, this is something that you really want in order to make markets as efficient as possible.
00:17:45
Speaker
And in in this case, so just so that everybody understands, what you're saying is you yourself, if you're close to liquidation, can take up the flash loan and just reduce the position just to make sure that you you get that problem solved very quickly. right i think this is kind of the idea that you have.
00:17:59
Speaker
Now, once you're liquidatable, anybody can actually apply the flash loan idea. There's like a borrow of money. I put no capital at risk, I just go bam, bam, bam, do the transaction and make my money off of it. right so that's i mean you know That's essentially the mechanism that you have in mind. right And so why you have the incentive to monitor these markets too. well Because they could you could make a very quick buck for putting essentially no capital at risk other than your computer that actually monitors these markets instantaneously.
00:18:28
Speaker
Anyway, so sorry for I had to interrupt you, but you you you want to go. no no No worries. so um Before we get into sort of the your work um on this, I wanted to sort of double click on the frictions at play here, because i think the way you're describing, even even in the Flash Loan context, although I'm thinking about, you know, context beyond that, you know,
00:18:51
Speaker
yeah there are certain things that can kind of get in the way, right? And so I'm alluding to everything from ah gas fees to block times, et cetera. So could you say a little bit about, or even for instance, like price pads, like are, are do prices follow diffusions or do they jump significantly? And does that make a difference here?
00:19:13
Speaker
i So could you sort of get into how, um, how this might end up badly ah for, ah well, for the moment, we can even say for for the lender.
00:19:27
Speaker
So, yes, exactly. All you said is exactly are the type of issue you might face. So, Andreas, you said that's exactly what we need to make market efficient. And I think you're right. But it's also true that in blockchain, we have congestion.
00:19:39
Speaker
And that we don't necessarily have in centralized markets. That's the kind of thing that could go wrong. So let's say you have a huge price drop in the price of ETHER. Everybody wants to refill or everybody wants to liquidate the position in in compound.
00:19:54
Speaker
Then maybe the block space is not enough. And it might be exactly at this time that people want to do all kind of stuff on other DeFi protocol. So you may want to refill your position, but it might be that the gas cost is massive. You don't manage to enter the block and things like this. So that's the kind of thing that can happen, definitely, when you are exposed to a limitation of the throughput of the blockchain.
Risk and Resilience in DeFi
00:20:17
Speaker
And also, as you said, um the price of crypto are not diffusions. there's no There are not diffusions. There are diffusions with jump.
00:20:28
Speaker
for sure, and they are quite insignificant, and that's exactly where liquidation is problematic. So actually, to link to our work, that's bit what we're doing now. We are estimating diffusion process with jump so as to calibrate the risk of liquidation, because that's definitely when you get the jump that things get wrong go wrong.
00:20:49
Speaker
So maybe as a bit of a tangential question, do you have a sense of how... ah ah Do you have a sense of the solvency of these platforms, meaning like the extent to which the frictions do get in the way and and then there are actually losses at the protocols?
00:21:04
Speaker
So you get losses. So we have this paper. We have one paper where we look at them. You have... ah thousands of people will get liquidated. You have even debt that is so under collateralized that it cannot be recovered. So big because if the price drop is so big that you get even under, fully under collateralized, then you lose.
00:21:23
Speaker
And that but that's bad debt that the protocol will never recover. And you get it by the millions. That being said, um compared to the volume and the size of this protocol, it's not yet systemic. This seems to be robust.
00:21:35
Speaker
And they went through a difficult time. ah They were operating during FTX, Luna, you know, when crypto was, ah and they and they survived. And they actually worked. And I think that's because they have a huge over-collateralization ratio.
00:21:50
Speaker
Now, at the same time, that's also the problem, right? Because it's not very attractive to borrow in this protocol. You need time you need to put so much collateral, right? So I think the question is, can they lower this over-collateralization ratio to be more competitive with traditional repo market, while at the same time remaining robust as they have been until now?
00:22:10
Speaker
And that's more the open question to me. do you have a sense of... how things like faster blockchains like layer twos have affected this like do you have any sense of for instance the differences no and i think it's a very good question so to be honest we have focused all our research on layer one but now there a profusion of this protocol on layer of two and connected chains and uh we have not paid ah attention to it so i know they have like also of course landing protocol on l2 and uh Right now, we have yes, I prefer not to say anything about that, but I think it's a very relevant topic.
00:22:46
Speaker
So can ask a process question here? So, I mean, you said basically when I want to become a borrower, i first have to become a lender. And so I put my book collateral down that goes into a lending pool.
00:22:58
Speaker
Now, the value of my collateral drops. um And it is in the lending pool, right? So somebody wants to liquidate me, but as you say, right, so there's a certain utilization of these pools, right? So people take out the, so presumably the money that I put as collateral in the pool can be borrowed by somebody else.
00:23:17
Speaker
So now what happens if I'm really big, I put a lot of money in there and you know i'm I'm at the point where I should get liquidated. And is there guaranteed that but the money that I put into this into this lending pool is always there? I mean, is is there is there always money in the lending pool? For all of what happens oh short answer, so no.
00:23:36
Speaker
i think it's fair I don't have a historical case where it happened, but it's possible that a huge lender that has to be liquidated might deplete more than what's available in the pool.
00:23:46
Speaker
So it's more of a question of if if if there's this sudden... so you know, I mean, people can withdraw their funds, right, at any point. So if i if i if I put money in the pool and I want to withdraw it, and I need to go back to the pool and get money out, right? But what happens if everything is lent out, right? Right.
00:24:02
Speaker
um Same with if I want to get liquidated, then the the stuff that it wants to get that has to get liquidated it needs to would also leave the pool. so oh So your question is what happened concretely? You want to know? But what happened concretely is that the trade rebounds.
00:24:17
Speaker
Oh, okay. Fair. You have to wait, you have to come back. it's like i go to my So that's essentially like I go to my bank and tell me i want i want to have my money and they say that we don't have it. Yeah, yeah.
00:24:30
Speaker
But wait, there is a bit of ah a market mechanism at play here too, right? And what I mean by that, for instance, is that if if there is, you know, you've lent some capital and maybe most of it is is borrowed out, but if then you come in and you take some of your capital back from whatever is is is left there, you're actually pushing up the utilization rate and you're pushing up the interest rate for borrowers.
00:24:54
Speaker
And in principle, if they actually do have to face it, this might actually lead them to actually bringing back ah the borrowing. and And in some sense, this gets back to the liquidation mechanism, which is to say, like,
00:25:06
Speaker
what is it that actually compels the borrowers to have to pay those and interest rates? Because then know ultimately, i think Julian alluded that to this, but when the utilization rates get very close to 100%, the interest rates get to be ridiculously high.
00:25:20
Speaker
So you know if you have to pay 60% year to keep the loan out and you really do need to pay that, um then that's that's potentially going to be an economic... That's a significant economic incentive to pay it back.
00:25:32
Speaker
But of course, that's actually... um I don't know, if like maybe we should talk about this briefly, which is that um like the I do think it matters the way that borrowers have to pay back, um because that that then ties us back to sort of the the the liquidation mechanism. And what I mean by that is that, so let's say I'm borrowing Ether and and I'm told, hey, I have an 8% interest rate.
00:25:55
Speaker
How is that actually enforced? Maybe, Julian, you could kind of clarify that. Like, am I paying in Ether? Am I paying in USDC? Am I paying every block? um Is somebody going to call me if I don't pay?
00:26:08
Speaker
You know, where how does that tie back to the liquidation mechanism? So what happened is that as a borrower, you get continuously compounded interest that you have to pay. And these are added to your position, meaning that you need your collateralization ratio decreases over time, right?
00:26:27
Speaker
Because the protocol realized that it adds up this interest to what you have borrowed. So let's say when you started, you had 130%, and now over time, it's going decrease slowly as you increase your interest. So if you don't repay, you will enter the liquidation region. That's the way it works, right?
00:26:43
Speaker
And it's added in the currency. yeah So it's
00:26:49
Speaker
in dollar, I think, and converted. Yeah. But that's a good point now, I know you have adopt i think it's in dollar converted, but maybe it's in the I would have to check that.
00:27:04
Speaker
right So it affects the life ratio. That's yeah exactly what know where to put it. So I think it's it's it's in that way, it's efficient. I mean, OK, you cannot repay, but then if you don't repay, you're going to enter eventually this liquidation region and going to be liquidated.
00:27:18
Speaker
Well, fair enough then. um So let's let's go back to your concrete to your project. um You are particularly interested in the topic of contagion. So can you explain to us what what exactly is the contagion that you're that you're worried about or that you're trying to identify? um Maybe, Stefania, you want to go chime in on this one?
00:27:39
Speaker
ah Well, I can tell you about the technical part mostly. I think I'll do an introduction and then, Stefania, maybe you can you can pick it up if you feel better. So the point is that the type of contagion we are thinking is more, in a way, maybe theoretical.
00:27:57
Speaker
Okay? maybe theoretical, in the sense that we are looking a bit at what happened with fractional reserve. Let's say, like, if there is a run on one pool, and as we say, it's depleted, then what you have is that this is the collateral for the borough in another pool.
00:28:14
Speaker
And but in that sense, it makes the position, okay, in the other pool, not collateralized. Okay, so now it affects the health factor of the pool. affects the Now, the effect that it reduces the asset of the pool because the collateral is actually not available because the pool from which you get the collateral is depleted.
00:28:35
Speaker
And that could induce a chain reaction where all the pool get contaminated, right? It's the same type of things you get. So there is a very wide literature on that in financial institutions. So this is something people paid a lot of attention during the 2009, 2009 crisis.
00:28:51
Speaker
where a relatively small shock to subprime markets eventually toppled down, not only is the financial system in the US, but also in Europe. and so People in Europe at the beginning were like, oh, it's just subprime in the US, s and eventually all of the the eurozone was affected, right?
00:29:07
Speaker
It's exactly this type of thing. You may have a pool that is... becoming illiquid, okay, because let's say it's depleted, because maybe there's a run on it, and then you get a contagion towards the overpull because it's used as collateral to borrow in the overpull. It's actually the collateral. So that's what we look at.
00:29:23
Speaker
And to do that, we use techniques that have been established in traditional finance to study financial contagion across actors, across bankings, or sorry, across banks, or countries, or, you know. So can I just quickly chime in? So... um We want to differentiate contagion on the individual level from the pool level, right? So you're looking at it at the at the pool level. Pool meaning the particular asset that is being available for borrowing and lending. Is that right?
00:29:52
Speaker
So actually, we are we want to look at the user level. That's something we have open right now. But now what we've done <unk> just say... Let's say the price of one pool of one token goes down, then this will affect the collateral value of all the overpull and how this might spread to all the pool in the system.
00:30:13
Speaker
But i'm not talking it's not really like how it's going to happen in reality. It's like in the future time in terms of a accounting, how it will we reduce the balance sheet of the overpull because the collateral value has decreased. And actually, so maybe as I apologize, Stefania, for having interrupted you before, but maybe you can... So after we've done some econometrics, maybe you can explain the type of findings we we got in acronometric econometric findings we got on this. Yeah, I guess the first we have to explain which measures we use in our econometric model.
00:30:43
Speaker
ah So, no, right? ah So what Julien has been saying is that, so we try to determine which features, if any, ah is the main determinant in the ah what we call the debt rank, this measure that is using finance to...
00:31:05
Speaker
oh to basically provide well a measure of systemic distress. And we have several features available on the pool side and on the user side.
00:31:22
Speaker
um And we do something that is ah innovative, you can say, on this type of latin literature, ah because we...
00:31:35
Speaker
focus mainly on the user side and we we use we borrow some tools from the special econometrics and we weight the features of the users by their exposure.
00:31:52
Speaker
and And we don't do that for the for the pool's futures. ah So our model is constructed in this way. So we have a dependent variable that is the debt rank that we have estimated for different...
00:32:09
Speaker
um ah size of shocks. And on the right hand side, we have several independent variables where and there are the features of the pools and the weighted features of the users, borrowers and lenders.
00:32:25
Speaker
And what we find in a nutshell is that there's a feature that ah that they kind of sort out every in any type of ah specification that we have. That is the ah lender out a degree ah feature, which basically measures the the level of connectivity of lenders.
00:32:58
Speaker
And that is robust to many specifications that we use. And it actually, it is also very powerful in terms of ah explaining the ah that a big fraction of the variance of the model. So what we want to claim is that, ah so the...
00:33:22
Speaker
top Topological features are important as has been shown in other ah financial models similar to ours. But what we add is that we have many more users features and we add this ah aspect of the borrow from special econometrics.
00:33:43
Speaker
And also um these particular features of the users ah is new in the findings compared to the standard financial literature.
00:34:01
Speaker
So it's really not my particular area in traditional finance, but let me try to see if I can understand this a little deeper. So normally when we think of the measures being applied in traditional finance, we do this between financial institutions, right? We look at JP Morgan versus Bank of America versus AIG or the like, right? And so what are their What is their connection with one another?
00:34:24
Speaker
And what's there for the risk that you have? But you actually go much deeper here, right? You basically look at the user level. So it would be as if we knew what all the different counterparties that JP Morgan deals with are doing and what the counterparties of a city and so on doing. And so then we can actually bring this down if you want, we can bring this down to the culprit. So if you think of, let's let's say, the LCTM crisis in 1999, so it's like, okay, so there was relatively visible because there was this one fund that was a problem.
00:34:52
Speaker
but um But it's essentially that. right it's like you know It's not like how whoever was whoever was most exposed to them was Salomon. I'm not sure if I'm getting this right, but but you really, so if if we could have a world and abstracting away from, you know, this particular lending protocol in which we had the information that you have now in the blockchain world, we could understand financial crisis or risks of financial crisis actually much better if we actually could drill down on the user level. Is that right?
00:35:22
Speaker
Is that the implication really of your model? That's pretty cool. Yeah, well, and I think that we can say that we identify some particular feature of the user that can give us an idea of what can lead, ah what is mostly linked or highly correlated to the death rank.
00:35:46
Speaker
ah we're not talking much about causality yet. Oh, sure, sure, i understand. Yeah, sure. So, yeah, we are just making a link between the user futures. And we I think the the next project ah is about trying to identify the the specific users that are at the root of the...
00:36:13
Speaker
of the distress. I agree with what you said because one motivation we had when we started is data in DeFi is so much better than anything you can get in Trad5.
00:36:26
Speaker
So as you said, what we have is not only, you can think of pools as bank and you can think of users as clients. So we have the position of all the clients at every point in time without noise.
00:36:39
Speaker
So that's just... know the kind of thing you can never think of getting in TradFi. And so the idea was this, like, we can build okay tools to monitor risk in DeFi that are much better than what we have in TradFi. Not only because we have better data, more precise, more no noise, but also we have data in real time. And that's a big deal because if you think about regulators, they ah know that the crisis is coming, but it's like a quarter too late, right?
00:37:08
Speaker
While in DeFi, in principle, you could say, okay, guys, now... We have all these indicators, the topology of the network looks bad. we i mean, things are going wrong and we could have early warning signs before it gets, before it's too too late.
00:37:21
Speaker
And that's what' one of the motivations we have. So I know Fahad is raising his hand, but I'm just going to butt in anyway. um So, but I just want to and so, because I think this is the next question, if I would be a financial professional, just um Because I'm very excited. I'm really excited now about what I'm hearing here. and um I already think that there's some mechanisms also in in in DeFi or and and in in yeah and just say in defi and the blockchain world, like proof of solvency and so on that exchanges are doing, which is superior to what we have in the banking world.
00:37:49
Speaker
I just want to make sure that I close this one loop, which is this. It's like, can we can Does this really translate from DeFi to traditional finance or are there is there anything specific, I mean, very unique about the blockchain world that makes this different?
00:38:05
Speaker
um I hope the answer is this. It's it's sufficiently close. But I'm just wondering if there's if you can think of anything here that we would say, well, this is really just a toy world and it's not the real thing. And, you know, people in TradFi so much cleverer that they don't, you know, that this they can still manage this without having this little detailed data. They could get the same outcome or something of that nature. You think they could build... No, but I mean, you you said they could build similarly precise ah supervision tools?
00:38:32
Speaker
I don't think so. Well, and well so so I want to put it differently. right so okay so We have a lending market in DeFi. um In many ways, it has it shares features that we see in traditional finance too. We have collateralized lending or you know think of it as short-term repo market.
00:38:48
Speaker
ah There's trades, there's people borrowing lending. you know Prices can move against you get a margin call, you get liquidated. um you know Sometimes the pool the the liquidity liquidity in particular markets is drying up, which then leads to contagion across other markets.
00:39:03
Speaker
So we have all of these. We've observed all of this in traditional finance too. just wondering if there is anything that is really so fundamentally different that you can say that the learnings that you create here with your particular new measure, where you where you look at the client and the connectivity of a client, that that does not apply.
00:39:21
Speaker
ah You mean that will not apply? No, I think ah the insight should apply also to finance in some extent. Of course, it's not copy-paste because, ah i mean, another thing which is cool in DeFi is that because a lot of the parameters are encoded in the smart contract, you know how things are going to react. Right.
00:39:37
Speaker
Yeah. Right? Not only do you have better data, but often you know exactly because it's encoded. No surprises. You don't have, you know, so some stuff is actually... happening exactly how it's supposed to do. So all of this together makes it more predictable, but the general insight should carry, I think. some i'm just kind of So i want to I want to push in here a little bit on the TradFi, DeFi analogy, because it seems to me there are kind of...
00:40:02
Speaker
ah differences here, yeah even at a qualitative level before getting to quantitative dimension. So for instance, like if we're just thinking about kind counterparty risk in the traditional financial system, what is the analog of the like the arbitrageurs who conduct liquidations in the context of the lending protocols?
00:40:19
Speaker
Right. So like I, I, it's not clear to me that this is actually such a tight analogy. And let me push that a little bit further. So, you know, one obvious point is, of course, that um the ah one one obvious point is that um some of the more recent innovations, including Compound's third version, ah sort of stopped the ability for people to borrow out collateral assets. Now, you could say, but then that's idle capital and it's inefficient, etc. But you could imagine a world where the collateral asset is put into a different application such that when people want to liquidate, it's just pulled back, which is very different than, of course, giving it to a pseudonymous borrower
00:41:02
Speaker
who you know has his own incentives as to whether he wants to bring it back or not right i'm saying i don't really see what the analogy of those sorts of functions are in the trad fi world um so i guess i said a a bunch of different things there but so maybe maybe maybe as a as a smart just and I like lot your second idea, which should be the protocol like that. But think you're right. I mean, principally, you don't have to loan out the collateral. You could make it in an, like you could stake it, for instance. I don't know if you didn't stake it. It will be there now.
00:41:31
Speaker
So, ah but the analogies, so the analogies, so think of CCP, Central Quota Party, you know. So what you get, the take care of the margin call.
00:41:42
Speaker
Right. And then usually they have a phone that is ah but that's a club. Now there is a phone that is financed by everyone and that is used to mop up the bad debts. So there is a central centralization. You have a centralized system and then you have someone handling the CCP and they take care of this so liquidation. Right.
00:42:06
Speaker
Typically, and they sometimes, and they usually are financed by the funds ah provided by all the participants to the CCP. That's how it works. So, of course, it's not like DeFi where you can, yeah, it's the far west, where everybody can come in and liquidate.
00:42:21
Speaker
No? Yeah, so CPPs actually have a... That looks more efficient, right? Because if everybody can come in and liquidate, in principle, you should go all the way ah you know like to make a liquidation of it because they should fight. I mean, the gas is the MEV. It will go to the miner, basically.
00:42:36
Speaker
When the CCP, it's a rent. um I mean, in a way, because they are the only ones liquidating and they charge probably a premium. on I mean, i I would expect them to charge a premium on you know for their services, but...
00:42:49
Speaker
So I'm i'm just throwing very quickly because our um our podcast is called Crafting the Crypto Economy. And I think in a way, so the idea is really that, you know, what what would we do if we would actually move, as is now proposed in the US, we move actually markets on chain, right? So then we need to understand these borrowing and lending protocols well.
00:43:08
Speaker
And I think this is something that we're, so this is really an important insight that in a way you can get a much better sense of, ah the risk of contagion and the risk of of ah of a systemic spread, right, of ah of a problem.
00:43:21
Speaker
Now, can I ask the question here? So, Stefania, so you have now created a measure that allows you to identify the ah the the the contagion risk, if you want, from individuals, right, by their connectivity.
00:43:36
Speaker
Can we learn something here um in terms of is Is there a regulatory intervention that we can think of? or Maybe not regulatory, but ah like a protocol intervention where we in some form or other have to look at these, and we have to price what these people are doing differently um because they create greater risk? Or how do we how do we think about this? is there Is there an unpriced risk that these guys are creating?
00:44:04
Speaker
um And therefore, is there market that that we that would that would be opening?
00:44:10
Speaker
I'm not sure. Maybe Julien can help me. I'm not sure either, actually. So why why do you think there will be... um So there is probably one, but it there's always inefficiencies somewhere where when you think hard enough.
00:44:21
Speaker
But why do you say that? Is there obvious one you can think of? Well, I mean, if you think about it right? So you're saying basically we can identify... ah that some people are riskier than others because of their connectivity. I think this is what you're saying, right?
00:44:33
Speaker
Ah, yes. So if I'm that, right, then I'm creating a ah greater risk to this this ecosystem or to the market than others. And apparently i'm not I don't have to pay for it. Do do I? um Yeah, okay. So that's our next project. So, okay. should okay that's So now we want to evaluate the risks that each user makes on the system based on their connectivity and their story and everything.
00:44:56
Speaker
Because right if you think about it, the rate at which you borrow is the same for everyone in the pool. Because it's only based on the utilization ratio of the pool, right?
00:45:07
Speaker
So everybody that borrow USDC pay the same price. But of course, a user that is highly connected, maybe with a lower leverage, and maybe a history of defaulting left and right, you know, should borrow at a higher rate than, you know, someone who has maybe a borrowed only is not, okay. So you should have a user-based interest rate, but ah this would make the dimensionality of the problem much higher and the ending of the smart contract much more complex because now they only set one interest rate per asset. Now they would have to, you know,
00:45:43
Speaker
But ah yeah, for sure. But isn't pseudonymity kind of a challenge for that? Sure, yes. But at least you observe the connectivity and you can observe the history. But let's forget about history.
00:45:55
Speaker
What our model shows us is that the way you're connected makes you more systemic in the sense that if I'm a user who's connected to many pools, if I blow up, I'm going to contaminate all the connected pool, right?
00:46:08
Speaker
If I'm connected to only one pool, it's more like contained. And so um maybe I should charge more the user, which is very connected, right? But but would so why why not just go with what sort of compound de eventually did, which is break the connections across the pools?
00:46:23
Speaker
Ah, okay. so if you break the um So that's a different question. So if you break the connection between the pool, it means you have lower capital efficiency because the only way you can break it is is that you don't loan out the collateral.
00:46:35
Speaker
No, no, but that that goes back to what I was saying earlier. You're not... low so part of the issue is the way in which the collateral is earning yield because you have a pseudonymous borrower running away with the capital and you go like, well, you know, we can incentivize him to come back, but...
00:46:50
Speaker
but that's that's But in some sense, there's not really a ah lever to pull to bring it back, right? Which would be very different than, for instance, depositing it into a different DAP where you could actually pull it back when somebody wants to liquidate.
00:47:05
Speaker
i agree. You could do that. It's a different model. Makes sense. That's a good idea. and has it were Has anybody done it? There are things sort of... not a kind of idea on podcasts, right?
00:47:16
Speaker
um yeah theres There are things in the context of of sort of like yeah Uniswap v four that are sort of getting into that direction. no You're right. I mean, you agree. I mean, you could improve capital efficiency by loaning out, let's say, liquid staking, no?
00:47:31
Speaker
Staking. For sure. Yes, no that could be one way, but I think so there are different models. I'm not saying one is to superior to the original model compound.
00:47:42
Speaker
They wanted to have high capital efficiency. So they loan out some of the collateral. This creates connection between the pool because basically it's fractional results. Now you have more different model like Morpho, which is growing very quickly.
00:47:55
Speaker
And they don't have this by default. You can connect the pool, you cannot use collateral, you cannot have the collateral, but it's not by default. By default, you don't do it. So then the pools, the liquidity pools are zeroed. They call them volts, but they are pool.
00:48:08
Speaker
So in that sense, they have less risk of contagion, but they have compensate by having a higher utilization ratio. So they are closer to having their pool depleted. Now you have overmoded. You're right. Maybe instead of loaning out the collateral, why not use it somewhere else to earn some yield? Remember, you you can have a high ah utilization rates um as as long as they're not on the collateral assets, right? So there's not actually there's not there's not a direct tension there.
00:48:35
Speaker
um ah It's just that, yeah, the collateral asset, what is it actually yielding? And that's where, as I said, you know you can look into other applications where there's more of a mechanical process of pulling back. In some sense, there's some value in knowing where the collateral asset is rather than just having it sort of borrowed by some amorphous, pseudonymous ah identity.
00:48:54
Speaker
Yes, so exactly. No, i agree. And I think in a way that's where DeFi gets fascinating because you have all these possibilities, right?
Innovation and Risk Management in DeFi
00:49:00
Speaker
And now it's just a modular approach that makes it, ah it's quite easy to rethink and say what you're going to do with the collateral and issue token in different types of protocol.
00:49:11
Speaker
Yeah. no like And but yeah only let me just say one last thing. Naturally, when they started Compound, they were almost alone, no? So it was obvious that they they were really reloading out.
00:49:22
Speaker
Now that you have an ecosystem, you can be much more creative, as you suggested. I mean, I think one thing in the background here is like when we think about things like risk pricing and sort of charging people for the externalities they're imposing on other people, I think a lot of people in the crypto community and maybe in the tech community more generally would say that the record of TradFi in accurately pricing risks, particularly in the last maybe generation, is is only so-so. And so it's one thing to say, like, I have a model that, you know, presumes a certain, you know, underlying probabilistic structure. And within this, I'm going to price it. and It's another, it's quite another thing to actually, um to actually be sure that that model is a proper reflection of of reality.
00:50:05
Speaker
ah um and And then to like reliably, then then to tend to then to sort of reliably ah lean on on the ah on the risk pricing from that model. So can I, um you know, being thinking about risks, just is there also something like an absolute risk of some sort?
00:50:22
Speaker
um So as Fahad before pointed out, so there is usually an incentive that you have that is if you have um everything is basically, ah you know, if the if you have a very, very steep part of the borrowing curve, if you have to pay an interest rate off.
00:50:35
Speaker
It can be up to several hundred percent very quickly ah for a particular loan that you have. um So the question is here is more something like this. It's like if the price of a crypto asset drops very rapidly, right? And so essentially, um you know, it drops faster. It it drops over time so quickly that even the liquidation is too slow in order to capture the the arbitrage that you want. yeah You just can't get it in. The network is congested of some form.
00:51:00
Speaker
And then all of the assets are depleted from the pool. And then nobody but decides to ever pay anything back, right? There's nothing in the pool. There's nothing to liquidate. I mean, you can you can rack up all kinds of ah loan um obligations that you in principle had. If there's nothing to to liquidate anymore, then there's really nothing to lose for you. So is there something like a nuclear option where this whole thing can go away and lenders basically are left with back in hand and borrowers are running a party?
00:51:29
Speaker
Is that possible? Yeah, I think it's very unlikely as long as this collateralization ratio is high enough, but it's theoretically possible, yeah. What you said. And I imagine, and and then right now there is no run now because there has never been a huge run because, yeah, but I mean, we know it's not because it did not happen, but yeah, yeah you're right. i mean Well, I mean, run is run, right? So, I mean, you know Diamond Dip Week doesn't go away that easily, right? so So, I think it's so that's what the Fad said before. The hope is that there's a stabilizing force because the interest rate goes up enough so that people will bring back, you know, liquidity in the pool because interest rates will be huge to loan out.
00:52:12
Speaker
And that has worked up to now. Okay. okay so So can I also ask, us just generally speaking, so, I mean, you have an agenda on this, right, on this topic. So, I mean, this is, I personally find what you find really quite exciting. I think this this individual level and, you know, and and measurement of systemic risk, is I think is really something,
00:52:32
Speaker
ah novel here, right? um So what else are we looking forward to here? What what else are you working on in this space? Because ah here you have an entire agenda. Maybe you want to map that out to us. So the agenda, it took us a while get the data and to put them in the form of a network.
00:52:47
Speaker
Because the first thing was to go there, get the data, and then we had to build the balance sheet So by the way, that's funny because when you look at the protocol of lending, they only give the asset side of the of their asset.
00:52:59
Speaker
So all these protocols, they are built on providing the asset side, but they don't build the liability side, which is the way you do in finance. So we have to do that. We build the liability that we have built the balance sheet.
00:53:10
Speaker
So that was a lot of work. And by the way, I have to give a shout to my co-author Matt Kamontovaniฤ because he did the heavy lifting, right? He's now at the University of Vienna and he did all the heavy lifting on this data work. Now that we have it, we know how to do it.
00:53:24
Speaker
And actually we have worked on compound, but now we already have done it on Havรฉ. So it's quite fast to do it once once you have done it on one protocol. Okay, you have to look a little bit at the new graph, but you know, you get gist of it. So so now we have this data and the idea was okay now we have this kind of graph we want to build modern next generation supervision tools that's our agenda okay and eventually provide to all these protocols maybe dashboards monitoring tools where they could not only um look at what happened but also look at what's gonna happen and have like early warning signs and things like this
00:54:07
Speaker
And maybe even potentially down the line, as you said, you could invent things where you price at the user level. I mean, all these things could be used then to do more advanced type of protocols. That's the agenda.
00:54:18
Speaker
So now we have done that. Now we are working on modeling contagion, but we have a lot of work to do on how to model more realistic contagion. I think Farad pointed to that. So when does this thing really break? We still have to think more of it, how to simulate it.
00:54:33
Speaker
That's what we're doing. And we want to build more... find more econometric results. That's what Stefania is working on. She works really on extracting because once you simulated this algorithm, it's called complexity science that you don't understand much. It's like machine learning.
00:54:50
Speaker
Then you want to interpret it. and That's also what we try to do. We want to go not just produce simulation, but try to understand what's going on. And I think, ah but we have not yet worked on it, but I think AI might enter the picture also in it because you have like a graph neural network that are arising where they are able to extract structure in neural network and they are able to use graph as input and understand what's going on. So, I mean, because it's a huge data set. So here maybe AI could be useful down the line, but the objective is just, I think so. Let me say it differently. In DeFi, the culture of innovation strong.
00:55:28
Speaker
Okay, and that's a cool part of that. I think why we all do DeFi because it's exciting. There's so much going on. But then the culture of risk is lagging. Okay, because that's not where you make money. And that's a bit of way that someone has to do it. And we are academics and we want to help. And and I think there's no reason for that because the quality of the data DeFi could actually even easily, you know,
00:55:50
Speaker
but Yeah, you compare and actually put to shame, let's say, like what is done in Tract 5 to measure risk because we have just better data. Yeah, of course. I mean, it it seems like something that, I mean, this is something that meant many of us think, right? So we we can see so much more in DeFi and so we can get better inferences, we get better information and we can but build better models.
00:56:10
Speaker
And therefore, if you think of prudential regulation or any risk regulation or any risk, not forget about regulation, this is maybe an overstated term, but risk management, right? You want to be able to like to to have you know the more effectively, right? And so this is, I think this is, I think if you think of the, getting back to the name of the podcast, the crafting of the crypto economy, this is essentially the ah the frontier of of making a ah financial work better. So i like good to you, good for you.
00:56:39
Speaker
Exactly. so Yes, sorry. So yes, I think one way to go to the regulator is to go for to the regulator or to financial institution is to show them that actually we have tools that are not even as good as what they are, but just like the next generation.
00:56:57
Speaker
of what they could mean you know Ultimately, i mean so if you if you i mean we can go to regulators, but ultimately, regulators are there to also solve information inefficiency problems. right So that's kind of what they can overcome. and some i mean In some sense, you know yeah we we have to ask the question of whether or not DeFi, bit with its higher level of transparency, can overcome.
00:57:16
Speaker
but what What of the information problems can it overcome here? and And what therefore, you know we can then model and and assess probably with market mechanisms. I mean, that kind of has to be our goal, right? can we Instead of having a regulator needs to look at, you have to use measure X so that you know the world doesn't fall apart, and here's the requirements that you have.
00:57:36
Speaker
well, here's all the data that you on on your ledger and on on the things that you do in your system, right? So as you say, you have to here's your liabilities, right, as a platform. You know, here's the risk associated with these. Now the ah remainder of the world can assess it, right?
00:57:50
Speaker
Yes, exactly. So you could make stuff compliant by design. You could make them very transparent. and you can And that's what we try to do as Nexetra, even build more advanced measures because we have just more data. So in a sense, there's no reason that why DeFi could not compete on that front.
00:58:06
Speaker
Because that's, so that's all philosophy, you know, and and we and so I think, and by the way, now, okay, so um not all of our data, but we are making them more and more public. So because we think it's a public, ah it's a common effort. We don't want to just like, you know, keep it to us.
00:58:22
Speaker
So we make them more and more public to people who want to work on that because we think yeah it's a general idea and it goes beyond us. No, i think it should be a, common effort of a researcher and people working in the the field. Well, this is almost a a great way to to get to the end of our podcast but because there's this call for, hey, you know, I but i have cool stuff. Let's go.
00:58:43
Speaker
You guys go come go crazy on it. It's actually a very good way to do it. But and maybe I should ask Fahad if he wants to chime in here because I'm monologuing. No, I mean, I think i think one thing to to think about, i mean, because so I think the last bit here was was very optimistic towards DeFi and like the potential of it. And and and there is, I think, i you know, clearly...
00:59:07
Speaker
things that when you have everything on the ledger, then you can sort of design economic mechanisms in a way that they wouldn't have happened in the TradFi world. Like, you know, I'm i'm thinking now Julian, you were talking about bank runs and so on, right? Like when and asset prices fall, that the when when when when the when the assets that the bank has invested in start to collapse in value, you know, there's not really frictional liquidation in that context, right?
00:59:34
Speaker
um And so... There is a lot of value, I think, in kind of having this common ledger where we can essentially write code to to implement policies and implement economic mechanisms. And I was alluding to the L2s earlier on in this conversation in part because there are still frictions in terms of like the block times and the congestion and so on. But in some sense, those are all being alleviated through, you know, now multiple blocks per second on the fastest L2s and so on.
01:00:02
Speaker
um But I kind of think there is this, this big elephant in the room, which is that, um Ultimately, a lot of the relevant assets in our world are physical assets, right?
01:00:16
Speaker
and And so enforcement, you know you you can you can have an NFT that tracks the ownership of a particular physical asset, um but but enforcement becomes a difficulty. And it seems to me that's actually the pain point for ah for DeFi. um And so to make it concrete and maybe bring it to decentralized lending protocols, right?
01:00:37
Speaker
um i you would like to think that ah that businesses could take advantage of something like a decentralized lending protocol and get a much more competitive rate on their small business loans.
01:00:52
Speaker
But that's ah that's an obvious example where the collateral that a small business is typically using is probably some physical property. um and And it's not quite clear how that translates to the to the to the DeFi setting.
01:01:06
Speaker
um i but you know certainly you can you can you can do things beyond what's on the ledger to take care of enforcement, but it's but it's it's not I guess it's not clear to me when you sort of do those other things whether you lose kind of what what makes DeFi very effective within the space that it currently occupies. so you know I wonder, if Julian um yeah and Stefania, if you have any thoughts on on sort of the idea of, I mean, I suppose I'm talking about real-world assets, but um
01:01:37
Speaker
if we're going to talk about lending, I think about small business lending. And and then then you're talking about physical assets. And then in the blockchain context, I guess we're thinking about real world assets. And I kind of wonder what the what the evolution of these and decentralized lending protocols will be with respect to real world assets going forward.
01:01:52
Speaker
So I think, but this is, the okay, what I heard is that several of these lending protocols are thinking of more and more of themselves as banks. And one horizon would be to extend loans to people in the real world.
01:02:08
Speaker
So ah my understanding is that some of these protocols are moving in that direction. Okay? Potentially, yeah and you also get this thing with stable coins issuer and things like this, no? Because you could start to, if you start to issue stable coins, you're not very i far from being a bank, no?
01:02:26
Speaker
The slope is there. So it's one direction. And I think it's promising, but it will have to be fought through. And here I think they will need more economists. but Sorry to tell that because like DeFi has been very good because also it was very virtual, but as soon as you get to the real world, you get all this asymmetric information problem that is much more much more prevalent that so Andreas was talking about. And here there's long history. So there would I think it's interesting, but it will be challenging.
01:02:56
Speaker
That's my feeling. Yeah, i think I think that's actually a very important point that you that you're concluding with there. but i think I think that's probably the way we want to conclude here, Fahad, unless you have something to add. um i want to I want to thank you for and incredibly insight kind insightful discussion. I learned an awful lot, and I hope the yeah the audience did too.
01:03:15
Speaker
Thank you.
01:03:20
Speaker
We hope you enjoyed this podcast. Thank you for listening. As a reminder, you can find additional materials on owlexplains.com and can stay updated by following us on social media. That's all for today.