"Liquity: Interest-free borrowing" - Robert Lauko

up we have got Robert loo um from liquidy I'm very excited about this talk I might be slightly biased I'm an adviser to this project disclaimer but um Robert is gonna be talking about interest rebaring on ethereum and I will let Robert take it away thank you Ashley for this introduction and I'm very glad to be here and thanks for having me um so let me share my screen and then I can just start presenting so as Ashley mentioned I will I'm going to present liquid liquid is an interest free borrowing protocol built on ethereum and now let me just start with a short like overview of what I'm going to talk about um like first of all I would like to outline like what makes a stable coin borrowing system attractive to the borrower or to the end user and then like in the next um section what I would like to do is just look at what kind of key ingredient or building blocks there must be in order to achieve those goals like to make stable coin borrowing as attractive as possible and I will cover those ingredients like efficient liquidations and and our stability mechanisms in a more details later on so let me basically start with this with the first uh like what would what should a borrowing protocol really achieve and offer to the user like it should be obvious that like 0% interest would be very attractive because nobody like likes paying interests or fees and then it's also maybe not that obvious but to some people this is a relatively important uh property that the system is immutable and really decentralized which means that the whole system acts fully in a fully autonomous way so there is nobody who can like interfere with it or temper with it and then last but not least um speaking of collect alized borrowing because liquid is all about collateralized borrowing we also wanted to make it capital efficient which basically bo boys down to having a minimum collateral ratio as slow as possible in our case we are targeting 110% just quickly if you have slides you might want to um you share them because I don't think we're seeing them oh that's strange oh sorry I'm really sorry can you see them now yes we can really sorry yeah so here are my slides um so basically I'm talking about these three uh properties we want to achieve like 0% interest rate immutability and decentralization and capital efficiency so Capital efficiency here like boils down to a low collateral ratio which in our case should be 110% this is still subject to like an ongoing agent based modeling effort but based on our first like intuition it should be around this number and now like what do those three like doida really entail or or what what can we what what do we need to do in order to achieve them in our system so here I think if starting with the interest rate it's important that we can issue our own stable coin because only if we we can basically mint our own stable coin that we give out as alone only then can we make it interest free because otherwise if we need to Source it somewhere on the Capital Market it would mean that that the system would need to pay interest rates for it so that it cannot lend it out to others without like um charging any interest rate and another aspect here is that interest rates are also important in in some systems to keep the stable coin really pegged to the US dollar or whatever underlying they have and we do this differently by making our stable coin redeemable at face Val Val so you can always get an equivalent amount uh like an amount that is equivalent to the face value of the stable coins that you redeem and the those this redeemability is coupled with a set of one of fees which even though they are fees they are like only payable like at some like at the beginning so you you are not like subject to some variable fees like as you would with an interest rate and then with immutability and decentralization what's really important for us here is that we make our system governance free and governance free not only means there is no like body that can vote on parameter changes but there is also no admin who could like just stop or interfere with the system and we would also just start or use ether as our only collateral in our first release at least so this means that ether being the really the native currency of the ethereum network this makes our system as decentralized as possible we are not relying on any kind of token that may be subject to some centralized influence and we go one step further by making it so that third parties can run front ends and they are even incentivized to do so so that nobody can just easily shut down our own servers because we we won't even run our own own like web interface we are focusing on building the protocol and then release it into the wild and then for Capital efficiency like what's important there is really a way of liquidating positions that fall below this minimum collateral ratio in a very efficient way and we should also incentivize Liquidators to that end like especially if the gas costs are high so we need to pay them some compensation for the gas cost and we also rely on an oracle or Price feed service that's fast and reliable because that's like the base input that determines whether a position or uh needs to be liquidated or not so let me first of all like give you more details on our instant efficient liquidation mechanism um if we compare it to what like existing CDP platforms do it's either they are like when a position becomes under collateralized they either do collateral auctions where they basically sell the collateral to the highest bidder or they just sell the collateral at a fixed price like fixed price probably means that there is a discount on the current Oracle P now the problem with this is that such a um liquidation mechanism is generally rather slow and inefficient by Design because collateral takes time uh auctions take time and also with the fixed price sell-offs and the discounts uh it's not as efficient like price-wise as it could be so we on the other hand we try and do it differently by introducing a mechanism uh for liquidations where we can basically use the existing stakeholders in our system to liquidate our troves so we call the cdps in our system the individual borrowing borrowers positions we call them troves uh just to clarify that expression here so what we want to achieve with that is to have like this fast and efficient liquidation so that there is no need to find a buyer or bidder because they are already somewhere in our system available and there should be no risk of further price drops which could happen during like a lengthy liquidation process now how do we achieve this um this is a little bit involved but I will walk you through the main uh steps of this mechanism so we have to differentiate between two system modes on the left hand side we have the normal the the healthy mode of our system which is the case when the total system collateral ratio is above 150% again this Val is subject to um simulations and we may have to adapt it and then on the other hand there is another a special liquidation mode that would kick in or that would take place if the total system collateral ratio is below 150% but let's focus on the on the healthy on on the like normal operation mode of our system and this involves two steps one is a stability pool which you can think of as an insurance pool which is there to absorb uh every Trove that needs to be liquidated like every Trove that falls below this 110% would first be liquidated against the stability pool and if the stability pool is not sufficient because it doesn't contain enough tokens for liquidating those troves then we would fall back to another um mechanism which is a redistribution which involves all the other borrowers in our system uh so I will skip the recovery mode for now and just uh show you how this main stability pool based and redistribution based liquidation Works in a bit more detail now so um L USD this is the name of our stable coin it's obviously ped to the US dollar now every holder of this LUSD token can deposit it to what we call a stability pool so at any time any holder can do that so this pool will fill up over time and then there is also another Pool which may uh like collect e ether tokens from liquidated positions but at the beginning it will just start empty now let's say this position Falls below the minimum collateralization ratio here you see the debt in red and the collateral in blue and like we just assume here that this is below 110% as an example so what happens is that the system can simply burn like the equivalent amount of tokens like the the amount that's defined by the Deb of this position is burned from the stability pool and the de can then be cleared here so the debt is gone uh and the tokens were burned now what happens again or what happens as the next step is that The Ether that was hold by this Trove is just sent to The Ether pool now this pool contains uh some ether from the position and just the fact that those liquidations happen slightly below 110% collateralization means that whatever gain uh is sent to this pool should be higher than the amount that stability depositors lose like the amount of L USD they lose should be lower than the amount or the value of The Ether that they gain here uh and obviously or just to go back here um the the main idea is that you can basically deposit uh as a like as a holder you can deposit your LUSD to the stability pool and then you can cash out whenever you want and the gains and the losses they are calculated like in proportion uh to the deposit that you have made because there will be obviously not just one depositor there will be many so the system just takes their Stakes or their deposit as a measure to determine who gets what share of the pi now it can still happen that there are just too many positions or troves that need to be liquidated or just large ones so that uh this stability pool would get emptied or if this like basically this uh debt in red is larger than this de in uh like the tokens in held in the stability pool then we can only partially absorb this position while the stability pool there will be like a rest like you see here red is still the there is still some remainder that and remaining collateral here which we need to somehow liquidate and as I mentioned there is a secondary or fallback mechanism called redistribution which would then um Kick In And as the name says basically we are just redistributing chairs or chunks of this position to all the other borrowers in the system now again this would happen in proportion to the collateral to like those blue bars so we Chunk Up This uh the collateral of this Trove and give those shares to all the other borrowers in the system and bang they got it so you see that the bars like they are now slightly larger and we do the same with the de like in red we also just chunk it up and give it to all the borrowers in the system so this is done and now we see that the position was fully uh liquidated and uh everybody got some some collateral and some that and again the same rule or same property would hold that if we assume that we are our Oracle is fast enough and and there is somebody to trigger the liquidation um the liquidation function uh quickly enough that then this liquidations should happen at a collateral ratio which is still above 100% close to 110% ideally which means that those borrowers who receive the shares from the liquidated position they would not lose anything uh so they would would rather gain uh like a net value from the position so this was all about liquidations now I would like to um cover another topic which is also important and this is how our system achieves price stability for its stable coin now again here we do things a bit differently from existing systems uh those CDP platforms out there they generally use variable interest rates which are set by human government mechanism this is a bit like a central bank where you can basically modulate uh the attractiveness of the loans by setting an interest rate so that you can um you can uh basically contract or expand money supply by making loans more less attractive and there is also something which I call a soft Peg mechanism which basically involves that whenever stable coin has like a long-term perspective of being stabiliz Peg to1 Us doll then you can say that if there is like a shortterm deviation from the peg let's say slightly above $1 uh that means that people would have an incentive now to take out loans because they would hope that in the future that would go back to $1 dollar so that when they repay their loans they would need slightly less money to repay basically their debt and the other ways uh the opposite is true if it's below $1 it makes sense for you to repay maybe even a bit earlier than you would have repaid if it was exactly $1 now we on the other hand um we have a zero interest rate system as I mentioned initially and also we have fees in our system but they are algorithmically governed so the system just looks at it its own um variables and like the state of everything and determines the fees and what it also has is this redeemability which I will just cover in a sec which is not present in existing um platforms and now with those uh like with this with the fact that we have a redeemable token and we also have a relatively low um collateralization ratio at 110% we can create a price floor and a price sailing so we can basically squeeze the price between those two boundaries which would act as relatively hard boundaries uh compared to the soft Peg mechanism that I I mentioned here um but we'll see how that works uh so generally we have the borrower operations like the borrowers who just deposit ether as their collateral and take out loans in LUSD that's like similar to what what is already there in the market but what we add here is a Redemption uh right or a Redemption mechanism which basically entitles any holder of LUSD to uh basically exchange LUSD for ether at face value which means that you get like you get like a value of ether in return which just corresponds to the dollar value of your LUSD like basically you get like let's say you redeem 100 uh LUSD you would get back $100 worth of ether and this now the question is like where does the system take uh like The Ether from and obviously it needs to take it from the borrowers because it does not come with its own like pre-field reserve so what we do here is in that we basically sort the tropes by collateral ratio you see here this is the riskiest one this is a bit less riskier and so on and this has like this is really well over collateralized so this is still our our borrower Bob and now when Ashley comes along and wants to redeem um LUSD for ether like what happens is basically that the system starts at this position and it uses the The Ether that are redeemed to repay Bob's position like repay Bob's debt so this happens then and what also happens is that the system takes uh the amount from Bob's uh collateral which corresponds to the amount of debt repaid and gives it to Alice and if there is a I mean there should be a surplus here a remain Remer because this position was collateralized at above 110% this remains with Bob so Bob does not like uh lose anything it's just uh the fact that he loses basically the the the opportunity of staying invested and if um the the rede redeemed amount is more than than the debt of Bob it would just proceed with the next riskiest position and do the same there so repay the debt of this one and give the collateral an equivalent amount of collateral to the person and so on so this obviously means that there is a risk of being affected or hit by redemptions as a borrower but you can lower your risk by increasing your collateral ratio but making sure that you are not on the at the bottom end of this sorted list but let's say in the middle or in a higher range now what we also do is we take fees on or we charge fees on both the loan issuance which is like a oneoff fee like when you take out your loan you may have to pay an insurance fee by default this fee is set to 0% but it can go up and similarly there is also oneof Redemption fee which is charged on the redeemed amount like both the issuance fee and the Redemption fee they are onetime fees and they are like determined as percentages of the loaned amount or the um redeemed amount and the fees they are charged in LUSD here and in uh ether for the Redemption and they they are collected in a staking contract which holds our secondary token called lqt um so the stakers of these lqt tokens they would get like uh like a share of the fee revenue generated by the system and you see that initi fee has a very strong immediate effect on the attractivity of of new loans basically because I mean if you have to pay let's say 5% on the loan amount as a fee you would rather consider whether you want to take out the loan now or or wait maybe a few days or weeks until this the fee would go down to zero because we see that the fees they they are algorithmically governed and they are governed in a way that whenever there is a Redemption they would go up because the system uh can detect with the Redemption or just infer that whenever somebody redeems Tokens The Tokens would be below $1 so if the price of one LD is below $1 it makes sense for you to redeem because what you get get back from the system in ether is worth more than this $1 so I mean and in in such a situation it makes sense for the system to raise the the issuance fee at least because by making the loans more like expensive it can throttle the growth of the system because whenever the price is below $1 there is this kind of risk that the system is just growing too fast and it needs to be throttled and then we also have like a Time base Decay where the the current uh base rate which serves as a basis for the Redemption fee and the is fee would just Decay and go back to 0% so in a in in a state where the system has no redemptions nobody redeems the fee should eventually go back to 0% and this if you put that all together leads to uh like a nice price uh Dynamics because we can basically have a price floor down here because if the price is below $1 minus the fee that you would pay on redemptions in that uh area it makes sense for you to redeem and there is also price sailing which uh is a result of the fact that if the price trades above 110 like $110 then you can basically take out the maximum amount of uh LUSD against your ether at this 11 10% collateral ratio and then sell your LD on the market and make an Arbitrage gain so we basically have Arbitrage opportunities above the sailing and also below the floor and in between we we still have a soft Peg mechanism similarly to what's already like the case with with similar systems and I will skip skip this this is just like another way of of showing how the price Dynamics work out in practice and I just want to quickly cover like the last topic of this uh talk which is an incentive mechanism for front ends and early adopters as I mentioned in the beginning we really want to make it so that uh the system will be decentralized so and we don't want to just rely on our own front end so we need to have some enshrined incentives to make it attractive to front ends to run a web interface and the way this works is uh we will create like a front and launch kit which would be relatively easy to install for anybody and also release like an SDK for our system which allows uh existing platforms integrator wallets to um build uh upon or integrate our system so those front ends they um can be part of our system and they can set a kickback rate between 0% and 100% And now when the system is in action and people like borrowers and stability depositors use a front end basically to interact with the back end like with the protocol then the system would measure the turnover generated by those early adopters and based on the turnover it would calculate like an allocation of lqy tokens which is our secondary token the one that captures the fees so the more turnover a frontend generates the more the higher will be its share of lqt tokens that it receives over time so this is a continuous allocation early adopters will get more so the Curve will will will start high and then and go down over time and the kickback rate also means that there is a part which goes to the or stays with the front end but the rest which is defined as the kickback is basically sent to the users uh we haven't like fully defined yet what would be the basis for this incentive mechanisms whether it would be the stability deposits made by them or the borrowed amount or some combination of the two but anyway the idea is that uh you the users would get uh those lqt tokens uh based on the kickback rate and based on the amounts that they transact through the front end so there is a very high incentive to become an early adopted user or a front end for our system so that's it I'm here to answer questions and I just wanted to say that we are um planning launch for q1 next year and we already have a Dei live on our all important test Nets which can be accessed via devi. liqu it.org thank you so much for your attention awesome thanks Robert um we did have a question so toilet head asks uh is this only for short-term loans like a Um this can be for short-term or much longer term loans so there is no um limit on the duration of loans the only limit you have is that you have to keep your collateralization above the minimum threshold which is one 10% there is no minimum or maximum period of of of your debt okay great um there's another question in the chat about um assessing loan risk for each loan so does the system uh assess loan risk individually or is there um an an aggregate way of assessing risk uh I'm not sure if I understand the question uh right but the the idea is that the loans they are collateralized they do not depend on the credit The credibility or the the credit worthiness of the borrower as a person but of course the the lower the collateralization ratio the riskier the loan because with the price movements of The Ether used as collateral there is a higher chance that they may fall below or may drop uh in this range where they can get liquidated so and we use this kind of riskiness to evaluate which loan should be uh hit by a redemption or subject to redemptions okay great awesome and when is uh when is liquid going to be live uh so liquid should be live in q1 next year um this is on our schedule and uh yeah we we do our best to make it happen awesome very good well thank you so much Robert um really appreciate you joining if no one else has any questions I don't know Robert you might want to monitor the chat um in case there are any additional questions but that was all that I saw for now um