 Hello everyone, so welcome to today's seminar series of the CBNDC, so today we are happy to have Wilco of D&B to hold this session, Wilco is all yours. Thank you, thank you Russell, and thank you for inviting me as a moderator to this Central Banking and Digital Currency online seminar series. Before we start it's useful to keep track of the housekeeping rules, the housekeeping notes, so our presenter today has been allotted 25 minutes, can go ahead for 25 minutes, then the discussion takes over and he has 10 minutes to discuss and then after that we will have some time for a Q&A and that will take about 20 minutes and we try to sort of conclude in an hour time from now and during this Q&A session panelists can unmute themselves to ask a question or make this themselves clear and the other attendees please use the Q&A area, the Q&A box to ask questions and I will do my best as a moderator to select the interesting, while all questions are interesting of course, to select some of these questions and post them to the presenter and the discussant and also note that this recording or this conference or this seminar is being recorded and the video will be posted later on the website, so just to inform you of that well let's quickly move on to our seminar speaker today, today's speaker is Antoine Martain, well he has been around a while and he's been working at the Federal Reserve Bank of New York for a longer time and he's a senior vice president and also the head of the money and payments department, his main interests and research interests include money and banking, financial intermediation and economics of payments. Antoine the online floor is yours, you have 25 minutes, I'll give you a and hands up three minutes before time is up if that's okay with you and perhaps also good to note that if the panelists have a clarifying question please go ahead but let's limit that to only clarifying questions and the bigger discussions for the Q&A, okay Antoine it's all yours please go ahead. Wonderful thank you so much Wilco, so this is joint work with Michael Lee who is also at the New York Fed and Rob Townsend from MIT and before I jump in I have to remind you that these are my own views and not necessarily the views of the Federal Reserve Bank of New York or the Federal Reserve System. Okay so in this presentation today we're interested in an age-old problem of limited commitment in financial markets today, traders meet and agree on a trade some time before the action necessary to settle those trades can happen and the delay between the moment where these traders make promises to each other and the time where they have to fulfill those promises creates the opportunity for reneging on them because of limited commitment and markets have found a variety of ways to try to limit this issue such as third-party mediation with intermediaries or platform margin requirements or other use of collateral long-term relationship reputations things of that sort. Now despite that, settlement fails still occur so the peak of the global financial crisis in the U.S. Treasury market there was about 400 billion of settlement fails per day and this problem was sufficiently serious that a failed charge was introduced which basically put a minimum cost of failing in the Treasury market and despite that what while these measures did reduce the incidence of settlement fails despite that someone fails still occur from time to time. Now this is of interest so the persistence of settlement fail is both an institutional and a technological feature of the current arrangement as I mentioned a minute ago traders today have to submit instructions in order to fulfill the promises that they've made at the trading stage when it's time to settle and there's a delay between those two activities occur and if incentives break down if at the time where they have to take those actions to fulfill their promises they no longer have the incentive to take them settlement can break down and this is of interest because with new technologies particularly the advent of distributed ledger technologies and the possibility of having smart contracts self executing programs people are hoping that it might be possible to resolve settlement fails technologically and in this talk I'll be talking about the token system as a new type of settlement arrangement that could eradicate settlement fails through the programmability of assets so programmability would enable traders to commit to settlement that is at the time where they are making promises they could guarantee that these promises will be fulfilled and as I'll elaborate on in a minute this in order for that to happen there has to be a collapse between the time of trade and the time of settlement that is settlement has to be committed at the time where the traders are negotiating the trade and this is a feature that that's typically described in atomic settlement this collapse between trading and settlement and so this would entirely eliminate the potential for fails so in today's talk I want to talk about a market where traders have to enter interdependent trades to achieve the optimal allocation of a long-lived asset and I'll describe that in more detail but I'm going to hardwire in the necessity for for intermediation there's going to be limited commitments traders will be tempted to break the contracts after the trade or they will they will make promises to each other and they might have an incentive to renege on these promises later and what we're going to do is compare equilibrium trade and settlement under the legacy system that is a system where this inability to commit creates frictions and a token system where it's possible to eliminate settlement fails technologically but that might create other issues as I'll show so the question that we want to ask in this paper is does a token system strictly improve on the legacy on the legacy system and perhaps surprisingly that's not true in a setting where trading is endogenous and intermediation is essential right so these will be two key features of the model so you can eliminate settlement risk but in order to do that you exacerbate an information problem basically what I'll argue is that the token system implicitly requires trader to reveal more information about the positions that they hold and so what it does provide is it provides commitment but with strings attached so to speak like clearly in the environment I described if you could just commit unconditionally then you could achieve a better allocation than then under the legacy system but the token system does more than allowing you to commit it commits it allows you to commit under the condition that you reveal more information and that creates information problems I'll describe so just to give you some intuition suppose that we have an arrangement where a and b agree to trade in a token system right so what what happens and I'll go into more details about this later in the presentation at the time when they're negotiating the trade a and b are jointly writing a program programming the asset so that settlement cannot occur so that both both legs of that trade will happen with certainty and a basic requirement for this to happen is that each trader must have ownership rights to the asset at the time the settlement must take place right it's it's easy to understand intuitively that if you just program assets are not yours then you could just you know make take somebody else's asset and make that asset be delivered in your trade that would create obvious problems but so knowing that you must own the asset in order to program it means that your trading partner has information about you that they can exploit in order to extract a better price from you so this aggravates a holdup problem and it can break down trade altogether all right in the interest of time I'm going to skip the literature review and I'm going to jump right into the model so we have a model with three risk neutral traders a b and c and it's going to be a long-lived asset only one long-lived asset and it's initially owned by trader a and traders are going to have period dependent payoffs which I'll describe in a minute that can be low medium or high and in terms of timeline there's going to be two stages so first traders are going to meet and there's going to be two bilateral meetings between those traders during those meetings are going to make promises to each other they're going to negotiate a potential trade and then after that takes place there's going to be settlements the settlement stage payoffs to the agents only occur at the settlement stage depending on whether they hold the asset I'll describe that on the next slide and because there's this delay between trading and settlement it is possible that agents will have incentives not to to execute the trades they've promised to make at the settlement at the trading stage so what are the the traders payoffs so that the payoffs differ by period and by agents they can be high medium or low or or zero altogether so if you look at the first line this is the payoff of agent A the agent A cares a little bit about holding the asset in period one and in period two but cares a lot about holding the asset in period three and agents B and C also have payoffs that that are random in the sense that at the moment they trade they don't know what will be the value they attach to holding the asset so in particular during period two agent B might like to hold the asset moderately or not at all and as will be more important later for agent C that agent will might care to hold the asset a lot at period three or not at all that will be known but after the trading stage so at the moment where agency trades this is something that the realization of H tilde is unknown so the maximum payoff for holding asset in this economy is comes with B holding the asset during the first period C holding the asset during the second period and the asset returning to A during the third period so from a planner's perspective this is the allocation you would like to be able to achieve and that means that the efficient trades are going to be for A to lend the asset to B during periods one and two so that B holds the asset during period one then B lends the asset to C during period two and C returns the asset to B who immediately returns it to A so that A holds the asset during periods three so can you achieve that in equilibrium that's going to be what we're going to try to find out so what happens at this trading stage as I mentioned there's two meetings that occur sequentially one of these meetings is that A meets B and the other meeting is between B and C now the order of these meetings is random now since B is part of both meetings B knows who she meets first but A and B and C never meet and A and C don't know if they get to meet B first or if they meet C second so in this economy B has to act as an intermediary between A and C that's an important aspect for a result now you could endogenize this and and you could create an environment where B has some superior information or other properties that makes her a particularly desirable intermediary we're going to abstract for that but we're but our results should extend to an environment where we can we endogenize the role of B as an intermediary okay so in a meeting traders are going to negotiate a securities lending contract for settlement at dates one two and three so the kind of arrangement they agree to is something like A lends the asset to B in period one at price B and up to two trades can occur during this trading stage one per per trading stage meeting and then at the settlement stage so so the trades specify our promises to make transfers between between these three traders and so those promises look like A lending to B in period one so that requires two transfers or transfers from A to B at the beginning of period one and a transfer from B to A at the end of period one that was correspond to that and and so those actions have to take place during the settlement stage and might be subject to to incentive constraints as as I'll describe okay so what we want to do is we want to study two settlement technologies one that I call legacy system where incentive problems are where the the inability to commit creates a problem because because traders might have an incentive to renege on on their promises in one the token system where technologically you can solve the commitment problem but you introduce information leakage so let me get into that a little bit so how does someone happen in the legacy system in the legacy systems during this trading stage the only way an asset gets transferred from one trader to the other is if the owner of that asset initiates the transfer and we're going to assume that said that failing to act on a promise that has been made at the trading stage will cause a trader to suffer cost delta we can think of that as a reputational cost or as a penalty like the fails charge that was introduced in the treasury market and we're going to assume that delta is an intermediate value it's not so high that it prevents all fails that would make the model less interesting but it's not so low that fails happen all the time but again you there is a sudden there's a cost for failing to settle but there's no way for agents in this economy to commit to settlement and their incentives might lead them to renege on their promises now in the token system bargaining the act of negotiating a trade with uh with with another trader and programming the assets happens simultaneously and programming the assets allows you to commit to delivery of the asset in the future so the asset is programmed during the meeting that occur at the trading stage those transfer instructions are self-executing they will happen even if no agent take any action further it's impossible for a trader to prevent the transfer of program asset from occurring right so by by programming you have self-executing movements of the asset that cannot be that cannot be prevented and so this this creates the ability to commit to two promises but it requires the holder of the asset it requires the trader to hold the asset at the time where that commitment is made right again you can only transfer your own assets you can't transfer other people's assets so you have to hold the asset and that means that both parties know at the time of trade the the position of the lender if i'm entering in a negotiation with you to lend or sell you an asset i must own the asset and this is information that you're going to be able to use in order to extract a better price for me and i'll i'll clarify that okay so let's let's see what happens in equilibrium first in the legacy system and then i'll talk about the token system so there are two frictions that are going to make it difficult in the legacy system to achieve the optimal allocation one is a limited commitment problem i'm going to focus mainly on that but i'll also talk about the holder problem because both are interesting so the limited commitment problem should be pretty clear what you're trying to do again as you remember is to transfer the asset from a to b for the first period from b to c for the second period and a really wants to hold the asset in the third period but c under with some probability really likes to hold the asset in the third period and the value for c of holding the asset is greater than the cost of reneging on that promise of failing to settle and so if it was probably lambda c agency really likes the asset then then trader c will will fail will keep the asset one more period and will fail on a promise to return the asset to b and that will prevent b from returning the asset to a so in this environment b is the only trader who was matched with both a and c and so for c to require the ownership of the asset b must successfully negotiate both sides of the trades and that means that it's possible for b to not value the asset very much but under the right circumstances to want to do this intermediation because it's beneficial but if um but because the the trades occur in asynchronously b must make market by completing one side of the trade in advance of the other anticipating future trade and that creates the potential for a hold up problem right it's possible for c to make a low ball offer what's happening here so so b needs to pay a price to a that may be in excess of her own valuation of holding the asset and c could make a low ball offer to b because in case b has already acquired the asset because because the value of that asset is not so high is not so high for b and and c can offer something that's just as good or slightly better but if b was able to anticipate that she would not want to engage in intermediation that's the nature of the hold up problem so what is the trader's c's temptation the trader sees temptation is to offer something that's that's less than than b would like to to receive in order to make intermediation profitable so both a and b know that in the legacy system there is some probably that c will not return the asset if if i was probably lambda c the value of holding the assets is very high now for b the reservation price at which she will always agree to intermediate on behalf of c corresponds to a's value for the asset in period two which is l plus the settlement risk premium the the risk to a that he will not get the asset back which is which constitute the cost lambda c h plus what i call the daisy chain premium b knows that which probably lambda c c will not return the asset that will force b to fail on the promise to a and that will cost b lambda c delta so if if b can be confident that c will be willing to pay at least this much to borrow the assets then b is always completely satisfied to do the intermediation because intermediation will be at least a zero profit or or or better that value is increasing with lambda c if b is worried that c might offer something that's lower than that amount that might discourage b from from being an intermediary from from conducting that trade so what happens if b is already traded with a right so first b means was a agrees to pay a price to borrow the assets for two period now that trade is sunk that means that at the time where b is meeting with c b knows that she will be holding the asset at period two and and if c offers a low value essentially the expected value of b to hold the asset and the expected cost of the daisy chain failure in that case b will agree to lend to c even though the total intermediation will will will generate a negative profit but it's too late to to change that now what limits this risk is that c does not know if b has met a prior and c does not know if b has obtained the assets for me and in this opacity about the order of trade order of meeting and and whether b has obtained the asset is going to discourage c from taking advantage of the situation and making a low ball offer so what we can show is that if the limited commitment promise not too severe then the optimal trades will it will be achieved with certainty if the limited commitment problem is severe then not all of the good trades will happen with certainty and basically in order to guarantee that c offers a high enough price b will be forced when she meets a first to with some probably just refuse to trade and and because c knows that that this might happen that reduces the incentive to make a low ball offer and that means that some optimal trades cannot occur in in this economy so what are the strong feature of the system there's this complete decoupling between trading and settlement traders must enter contract well sorry are allowed to enter a contract without explicit proof that they can fulfill those contracts and that allows b to hide from c whether she has met a before or whether she has purchased the assets and so you have this interesting situation that there's an interplay a valuable interplay between the trading system and the settlement system the fact that trades occur asynchronously and that there's no transparency over the history of trades might seem an isolation to be problematic but they are essential this decoupling is essential to facilitate the intermediation of this long-lived assets between multiple traders it is very useful to have opacity in order to allow trades to occur and resolve the potential holdup problems so what happens in a token system in a token system as you might remember the whole point is that you can eliminate the risk of of settlement fail um see we know we want to hold the asset at at date three but b and c can agree to program the asset so such that it will be impossible for c to hold on to it the assets will automatically return to be at at the end of period two and this will occur regardless of the incentives of c to fail so this is great right it allows c to commit and it should eliminate it and it does eliminate the commitment problem but the issue is that it's going to worsen the holdup problem as i'll describe next it also creates an issue with trade in a synchronicity which is somewhat less interesting but i'll i'll briefly touch upon as well so why is the holdup problem worsen well now when b meets c b must reveal whether or not she has ownership of the assets right in the system in the token system for to be able to program the assets you must have the ownership right from that asset so if b doesn't have the asset there is no potential for even negotiating a trade because you can't deliver on that trade at the moment of trading you can only deliver on that trade you can only enter a productive negotiation if you own the asset and that means that c knows that behold the asset and that increases the incentive to make a low ball offer remember in the legacy system b benefited from the fact that c couldn't be sure that she had the asset now c knows with certainty at the moment trading whether b holds the asset and that means that the c's optimal strategy is to is going to be to offer an extremely low price and this can kill b's incentive to to intermediate entirely not get back to that the other issue is trade in synchronicity so in this environment we've hardwired the fact that the order of trade is random was probably one half that means that in some cases a is going to meet with b during the second half of the trading period so when b and c meet there's simply no asset to be traded because the asset hasn't been obtained yet and that means that no trade is possible was probably one half again this is a feature we could endogenize we could have a system was through some efforts agents are able to modify the order of trade or they can search at some cost but but but that would not eliminate our results it would continue to be the fact that it is much the order of trade is much more essential in a token system compared to the legacy system and that would have a cost that needs to be born somewhere in the economy so we think we're fine by simplifying the model and just just going with with hardwiring of the timing okay so the result is that and one you have three more minutes to to reset conclusion thank you yes perfect thank you um so i'm practically there so so if the hold up problem's fine then c will never get the asset because if the value of b is sufficiently low if b doesn't value the asset for itself enough then then b will refuse to to intermediate the only is a only agrees to intermediate assets that she values enough for herself otherwise c will obtain ownership of the asset but only was probably one half when when the order of trade is is exactly the right one so what do we see well we we have an environment where depending on the parameters either system could be dominant the token system that promises commitment but with string attach is going to work well if the commitment problem of c is extremely high and if b values the asset enough right so basically token systems are valuable in environments where the intermediaries care about the assets to intermediate for themselves but it's going to it's not going to work well in environments where the intermediaries do do pure intermediation intermediating assets that they don't want to hold themselves um and and so that eliminates uh arguably a very important role for intermediation which is that you can you can hold assets on behalf of other agents that you don't really care to hold for yourself and so if lambda c and lambda b are sufficiently low then the legacy system system actually will do better from a welfare perspective than uh then the token system all right um so to conclude uh tokenization has clear advantage it does allow agents to completely eliminate uh a limited commitment problem but as i've tried to argue it comes with strings attached collapsing the trading settlement comes at a cost um because it reveals information and and we know uh that in finance too much information can kill risk sharing opportunities or beneficial trades and this is exactly what happens here and so some features of tokenization that just not amenable to certain market structure if in a market where um intermediation is valuable um tokenization can be problematic and lead to lower welfare and and so one of the things one of the interesting things that that our model highlights in addition to to to the key result is that efficient settlement uh protocols are intricate intricately tied to uh to their trading mechanisms uh and in in OTC markets in markets where you rely on intermediation the trading mechanism that kind of trading only works uh with certain types of settlement or works better i should say with certain certain settlement arrangements and with that i'm all done thank you thank you very much Antoine uh for this clear presentation and spot on in terms of time let's now move on to our discussion today our discussion is to you hey and i hope i pronounce it right um he's a professor of finance at the University of Chicago in the booth booth school of business uh choose a main research interests include not various topics in financial markets and macroeconomics with a special focus on contract theory and banking but recently has also been writing articles in the area of cryptocurrency and blockchain too um you have 10 minutes please go ahead John yeah yeah sorry although that did it many many many many times still have the issues uh this is a very interesting paper i'm glad that i am being asked to discuss uh so it's may fundamentally change the settlement in financial markets uh that's a something uh you know not always being the center of financial research uh because in a way that the settlement itself we in it we long time ago we thought it was too detailed to study uh but during crisis you know Antoine presented some other things that are showing that this sometimes will you're nagging and nagging the system and a lot of efforts research focus to get move on to this if this is this area blockchain obviously you know promote this idea called automatic execution when conditions met so page four in this paper says the following program program ability program ability enables traders in the tokenized market to enter trades that are defect or insulated from credit and a counterparty risk that is a really fascinating point and i will just explain one application in the second slide now the the paper is just asking a simple question is this always good um you know i without reading the paper i will say it's you know in general it's good real depends on certain setting but like economists are very good at figuring out that the work could go wrong right so so so what they do is basically arguing that materially otters the information environment for economic agents who then distorted their strategies in callable i really like this way of thinking a lot a lot of times is that you know indeed that these kind of distortion could have a very very big macro impact and often the time that we do not foresee when we implement these changes okay so i will in in some sense it's general perspective shared by my paper with will arguing that the the i o setting that that could change but i will just let me explain that there are stuff first rather than telling you the connection this fundamental information issue is typically ignored in computer science community because they have a different way of thinking about information i think that that's the all right so let me just spend one minute to tell you about the flash along to me to me that's a fascinating idea and also it's it's actually is is working there right so it's interesting development toward this direction it's basically smart contracts allow for the programmatic the ability of enforcing the so-called automatic execution and to mention that this is a loan which is basically saying that you know i borrow at a land only being executed when both the conditions are met so just immediate you know if you you would not say oh i'm letting to you worry about paying back why this is the important you find you know if you don't you're not adding to the finance side you probably say okay whatever it is it is really just open the open the new gate of this like a riskless arbitrage and we do you know as a finance guy that we think arbitrage is such an important force to making sure market is efficient uh often the time we will say you know so-called limits to arbitrage because of the timing issues right you enter the trade and you don't owe the price moves out of surprise or sometimes that just people just negate negate on the trend all these things that with the better infrastructure we can making sure that the price of a one the law of a one price that's the fundamental cornerstone of the financial market uh become more relevant okay okay so let me just tell you the model i saw a question in the in the in the chat room saying just it wasn't clear i you know i don't i'm not sure that my explanation will make it more clear but just different way to explain it maybe that works uh so there are three people abc you can see that these h uh h are the other other other other other guys that will really like this you know getting the benefit of holding so as a result the optimal path is this way there are two trades to to implement that that path one is this between a and b saying that okay you give it to me and then later later on at the time too i'm gonna return back to you so that that's basically the c ab13 that i used in the in the paper if you want to read it okay it's not easy to read i have to warn you and then the second the trade is basically b to c so that's circulated here that's called c bc 23 uh 23 means the two is from b to c and the three is c to b okay so the question when i read it first is that do you really need these report contracts probably yes probably no i you know it's just i i i really applaud the authors to come up with the with this setting to exactly implement what they want to say uh this is a limited commitment as well as a whole problem but i really hope that you know to me the issue is so fundamental that maybe you could do a simpler version of it but anyway that uh this kind of a repo effect is just when you do the contract you promise future doing something else seems to be important for the one set one one friction which is a limited commitment so this is a limited commitment which is the basic friction that drive anything or almost everything except that the first level all that are going to explain uh so this limited commitment is that the c guy that later on he might find this this asset very very uh valuable to him to code so therefore that even though he promised the early on to trade but then he will say you know whatever it is i'm i'm not enter the finger put the finger on the on the on the trade is then you will keep it so i would you know just a small thing that put the h2 in contingently at minus epsilon because then it will make sure that the the you know it's clear that in the social perspective that always the spot is the optimal one but really does not matter it's just the minus epsilon okay so that's the that's the model and another question that the two systems so how do we do it so this is a legacy system legacy system has this there are two key issues one is trading stage trading stage is called ac you're going to meet with b sequentially so initially when i read it i wasn't paying too much attention turns out to be a very important part of the tradeoff uh there's no recall this means that you know i bet you once then we just we didn't leave the phone number to to come back to you okay i do believe that those things are very is is a very relevant to in certain circumstances where you just trade it like a super fast away but when you think about the vc context just not not true okay so and it's also important is that in order to prevent the bargaining with the private information it's a and c make it a ticket of a leave it offered to be a and c doesn't have a private information b has because b knows when did he meet a and c in water water okay uh but but you know bargaining game without the information it's much easier to solve so ac make a difference the whole department is reflected in some low price to be offered by c which entrant did a great job in presentation but i'm not sure i was working so hard to read their paper uh was you guys you know whatever you talked about in the in the in the in the in the presentation move there with some examples will help a quite a bit okay so essentially that it is that in the holder problem is basically if i know that it be already had it then c gonna say that i were gonna just give you a whatever your uh your holding value reservation value because b at that time doesn't have any uh bargaining power the settlement the stage uh then then c may renege that's a little bit okay so legacy two two please two uh just to interrupt a little bit you have three more minutes good good okay thank you so uh legacy system is the comments is quite clean on the limited commitment issue the delivery of a hold up the delivery means like the model to deliver that idea not that the delivery asset can be significantly i already talked about that example will be very important you mean yourself sorry you in the end you basically focus on certain corner solution anyway so here is my understanding okay so for c ideally he's facing b who's stuck with the asset offering a price of em to that what's em to em to em is the reservation price it's a value you can choose the zero in some sense then okay so em is is busy holding back if c's i'm sure that the price will be higher that's the equilibrium efficient the price you're gonna make uh if you want to make implement that that that path and and when em is a low that's high basically so b gains bargaining power by opacity that's you know i believe that's a very gentle result in a colubran actually be randomized saying that even if i meet a i'm gonna say you know i'm not a trader with you to make sure um uh unsure to make sure i'm sure that kind of idea okay maybe as a very benefit of answer so so so that's the t-logic and i think that you know it works just that you perhaps you can do it a more cleaner way okay tokenized system very clear that in tokenized system there are two inefficiencies as as entrant entrant already told you one is a random sequential meeting so only trader with half probability which kind of like very very stark i think entrant defended a little bit when it was a presentation so may could be empirical relevant depending on application i already said worsening hold up problem i hope it's already clear now is that now if you enter into prep proper proper and blah blah blah the trade then i know that you has asked that therefore i will give you a roll call that's it that's my concluding remarks i think it's a very fresh and innovative uh uh perspectives uh uh information going to affect the equipment pricing and the intermediation and just just to link back to my paper with with the con written several years ago but we were imagining this like on more like uh uh not the financial side of whether that trader finance context it's basically blockchain technology orders the nature of a monitor among firms because every firm behave as a node to to let's say you know not uh this is like a you know a product called a verification node and then then it's the monitoring problem that is important to just immediately change it kind of share the same feature and i hope i do think that this is a very important concept that often just ignored by at this point it's a mainly technical advancement so last point is that it will be nice to discuss a certain application thank you thank you thank you too for a very provoking discussion before we open the floor for the q and a's and perhaps uh on to one do you want to sort of make a quick reply to let's say brief reply to what you was was was sort of saying yeah happy to so thank you for other great comments i think i agree with everything you've said so uh while i'll take away is we we need to improve the presentation of the paper i i want to again mention uh two of the things that that you pointed out so um the model for it to be as simple as possible we we make stock assumptions first in terms of the fact that b has to be an intermediary this is uh this is hardwired into the model and in the fact that the the timing of the trades um it is uh it is also um something that that agents cannot share change now i i do believe that both of these things could be endogenized at at the cost of considerable um uh complexity but but i don't i don't think it would modify the key results and so i think to your point it is that it points to the fact that these concerns are particularly relevant in markets where intermediation is is important and in markets where uh the timing in order of trades uh it is um is reasonably fast so yes in case of venture capital or things like that i think that that's not a big issue but but in in financial markets where people trade at even if there is recall there's a cost of having of talking to the wrong person first and so if if you have to reach out to many different counterparties before you can go back to the the the counterparties you really want to get to uh those those would add to the cost so i think it it suggests that our that our model would continue to um to deliver similar results even in in in an environment where where we endogenize these aspects and and i'll leave it at that. Thank you Antoine it's now time to open the floor and the q and a's and let's do it like this that the panelists they are able to unmute themselves and ask questions themselves and perhaps via raising a hand i already see a hand and the other attendees can write questions in the q and a chat area i already see a few there uh perhaps let's first move um jonathan was the first to raise a hand jonathan please yeah go ahead first of all thank you thank you for the very interesting talk i think asking very fundamental question and very important understanding you know the limitation of smart contract so i want to follow up with uh on c equals one of the point and example which is the flesh loan example and at the end he also mentioned this contingent execution it seems to suggest that maybe smart contract can also expand the contract space right say maybe when you sign the two contracts these two contracts execution can be conditional on each other but your terms can be conditioned on each other so i was wondering by doing this would it you know mitigate the whole problem let's say when you try to offer me bad terms right maybe by doing that maybe it will you know cancel the whole the other execution of the athlete or maybe determine yeah so this is a great question i'll take a shot and if michael wants to add anything i'll let and get there um i think so our activity in this paper is to show that there are real limitations to have commitment with strict detach and so one argument that that you've made and other people have made is like oh wait perhaps we could redesign our financial system to get around that problem and i don't think we're making any uh claims that it's impossible to redesign the financial system around those kinds of constraint but at a minimum we're saying like that's a pretty serious redesign so we can't just say hey let's implement this technology and not worry about unintended consequences but but fundamentally our model doesn't speak to uh to the point that that you're making which is perhaps there is uh an interesting way to get around this constraint but but one way or the other you have to deal with with the constraint so i think it's a great point and um i guess i'm sidestepping that a little bit michael do you want to add anything yeah i i think that's a great question and it goes to the question about uh our tolerance about settlement risk uh now our stance is we're showing that uh unequivocally eradicating settlement uncertainty may not should not be the objective when we're trying to redesign the system so if we have a tolerance for settlement risk we can arrange for contingent execution that can sometimes fail but that can still alleviate some of the issues that we're showing for example with the ordering of trades so jibbo as jibbo and anton both kind of talked about the order of trades matters whenever you have to be able to credibly commit at the time that you meet so that's a huge issue if there's a huge randomness to when you meet someone and when you can actually bargain with someone uh now uh contingent execution means that oh well i can agree to trade conditional on me finding a different arrangement in the future and that that i think is the direction that we need to think about um and this is kind of a big tech takeaway as we're thinking about certain applications in very institutional markets like repo markets and other money markets there has been a push to to completely eradicate this kind of time between trade and settlement in repo markets and and we also kind of discuss a little bit about the practical applications that are going on right now and i think there the the takeaway is before you think that this collapses desirable and necessary we need to think about the other complications that might occur given that a lot of these markets heavily depend on intermediaries i saw another hand serial your hands down again um is your question answered or do you want to raise your question serial yeah hi everybody um so this one is a similar question as janet yeah can you hear me yeah prima yes so it was it was a similar question as janet and it was basically on the fact that the two parties could set up an escrow account where uh you know when both sides deliver their assets in the escrow account then the escrow account frees the asset to the other party and so essentially you can solve the information problem in this way so i was wondering what otwan was thinking about this this idea but this you know the fact that the technology allows for this type of accounts to be set up so you know all you know i would be much more optimistic than the authors regarding what the technology allows us to do in the future yeah so i mean i think i'm gonna have i'm gonna answer in a similar way to to the way i answered that johnson's question i think it's really interesting to think about these potential ways of redesigning the financial system around around new arrangement but i don't think that's costless um in uh so so it is again somewhat outside of the scope of our paper but um all of these attempts require uh significant changes and uh if we undertook that say in a different paper it would be interesting to see if if if that creates other potential problem that that we have enforcing uh let me just add uh i think the escrow point is really helpful to understand also a broader kind of benefit that we're actually highlighting in our paper which is in a token system you can eradicate settlement on trades for assets that you don't currently possess in your account so we're not arguing that you need to own the asset literally at the time of trade you need to own the rights to an asset in the future so uh there conceptually this already is very innovative relative to the legacy system where ownership is uh is a point in time thing you need to have the asset in your account that's when you're the owner if you don't then you're not the owner the second aspect is that you know in principle what we do want to achieve is to avoid any level of collateralization that's unnecessary so uh one way to easily kind of resolve a lot of these issues is to incorporate collateral which we're already kind of doing in a lot of arrangements or penalties the hope is that we can redesign it in a way that we can minimize the liquidity costs associated with other practices that can kind of prevent these types of issues thank you thank you Michael for this additional information i'm looking at the q&a box now and i saw i'm sort of browsing a bit i saw a question of jamie but i have the feeling i cannot see that completely that this question on flash loans was already answered in the in the box jamie asked if if let's say the if you do that if you have these flash loans using the blockchain and the repayment must be included in the same block does that mean that time elapses between a provisional loan and an actual loan and repayment and he expended actually on that um but that question is gone i believe so jio did you already i saw your writing did you already um send an answer there yes i think i did and to me from an economist's perspective if things are sure and then within three seconds four seconds even within one day that why would those things matter like the elapses times elapses that that's yeah so that's my answer maybe you have some other perspectives that could matter but i so one thing that's important jamie just raises hands so maybe he'll he'll want to clarify but the idea of provisional loan is uh is potentially problematic here because there is no loan unless there is the repayment so so uh unless all of the ask all of the the legs of the trades are part of the same block none of them actually exist so there's no in that sense there's no time allows but jamie do you want to add something yes uh thank you very much this is a great presentation great discussion uh yeah my question on flash loans is is that similar to the difference between trading and settlement because i can go out and get a loan do a bunch of arbitrage and if it works out but then i can submit all of that to the block uh and it it all settles if it doesn't work out of course it reverts back to the original case uh so isn't that similar to uh kind of a time um between i mean it's a different thing than trading and settlement but uh you can try out all these arbitrages can't you without uh without penalty uh before they become a part of one block and are traded and sold maybe i will just you know i kind of like your question getting to to me the right direction and then some of the parts of the comment like leading me to think that i'm not a totally understanding your point let me just just you know i think that this is a shared with entrant is that the idea of a riskless arbitrage that's like foundation of you know when we do pht is like first first time that we learned this finance is saying that there's a two price same exact asset you ideally you want to trade the same time so that you profit you make the profit but that never being the real world that you're always getting to the issues so here let's say you do you want to do arbitrage and then now this infrastructure innovation basically provided a possibility of doing that you can put even a little more complicated the execution strategy says that if that price is above that if that the other side of prices below that and then put it into the block and then making sure that you at least make that kind of profit this is what i meant and i do think that in this type of a situation whatever the time elapsed i do not think it really mattered so yeah and just to add jimmy i think i think you're right that flash loans allow you to collect a bunch of different trades at different prices and then if and then you can submit them all together if you think that that's beneficial or you or you can submit none of them and basically there will be no trade unless all of the all of the likes together can be part of the same block thank you so reading also in in the q&a room area there is a there is a question by francesca carapela i think francesca you can also unmute yourself perhaps you want to raise this question hi yes thank you thank you anton that was very nice um i just had a quick question about whether there is an implicit assumption that uh only the asset can be programmed but the parties to the contract cannot write a smart contract so a piece of code that can be tailored to whatever needs they have and they want and that is triggered once the asset let's say it's key is inputed in it's put as an input in the contract so it sounded like if this was the case if you're allowed for let's say a smart contract with all of the contingencies that the party's desire and that smart contract is triggered triggered when the input is provided then some of the results in the tokenized economy uh might might disappear would would that be a fair conjecture so i don't think that works um and again i'll take a shot and michael will clarify everything that i say that's wrong uh but we have a companion paper um also with rob where we where we dig deeper into the these ideas of of programmability and we look at at some of these options where trader could have multiple account and do like fake um you know uh transfer the assets in in specific accounts what if he or she wants to fail and so basically in order to really guarantee the absence of failure to settle you need to have very very specific condition and i think that the kind of uh contingent contract that you imagine would not be robust to that because they would allow traders to sort of circumvent um so basically you're setting these conditional payments and then the trader could choose to make sure that the particular contingency doesn't occur so that the settlement that was promised doesn't occur when they want that the when they want to keep the asset um michael do you want anything to that yeah i this is a pretty important point and i i think uh so i'll i'll give you a very simple counter example for why there's a settlement risk issue here uh so in the context of our model b obtains the asset and anticipating that maybe he can sell it to c now you could imagine that b can not share that information and instead say look if i have the asset i'll sell it to you now what that means is that b always retains the option to also renew on his contract with c because now as long as you're able to control the asset in any set of accounts if after you agree on that kind of trade if d comes over and says oh we'd like to have i'd like to buy the asset from you and offers a higher price well then b can now fail on the settlement agreement with c and importantly without any repercussions in the token system where the contract technically has not been broken so i think this goes kind of a in our paper we don't deal with this explicitly because this kind of thinks more deeply about what the conditions are for avoiding settlement risk but because of this issue you know simply providing more flexibility with the contracting will not necessarily resolve this issue and and that's because we just need one friction if b has different set of preferences maybe he has outside opportunities he can still there's a new settlement risk that involves the rises now i think the broader takeaway though is not that the smart contract environment is can be powerful i think it can be the thing is that practitioners and a lot of policymakers are focusing on completely getting rid of settlement risk and our kind of takeaway is maybe we should be a little bit more open to that and think about unlocking the potential to facilitate trades even when there is some settlement risk in the way that we're describing right now can i add one more so so i really like michael's point and i think that if you keep keep pushing this keep pushing this because then you will say you know let's write the d on the on the contract right but once you're pushing this pushing it in the end you need some very very very valid like those like a caught a caught of allicles that you need to trust and at end they're going to make the final nail saying okay this is the right thing to the right contingency to to the right right right on i think that's in some sense that it defeated the purpose of the smart contract because in the end we have to ask what are we trust in the end that's always that i have i do think it is a lot of limitation of smart contract just to be honest thank you sorry can i follow up can i sorry it will go i just want to go up my point and thank you for that and that's that's insightful thank you very much the follow up on that though the thought that i have is but in the model because what you described is so it reasonable but outside the model in the model that you guys presented if you allow for a smart contract of the let's say basic type to that described then you would not have that trade off for the tokenized economy that you have so you would solve the problem of you would go back to some opacity let's put it that way in the model in europe in this paper so there is there's got to be an implicit assumption i i'm guessing that this type of basic smart contract rather than programming the asset so i presume turning the asset into some sort of token that like a liquidity pool tokens i guess uh and uh program that token with the features that you described exactly as you described it now but in this paper if if you allow for a simple bare-bones smart contract you would get the opacity back wouldn't you and i think about it more uh a bit more before i can give you a definitive answer the way that i think about it if we allow them to have contingent trades where they can actually fail on settlement then yes i think it's Pareto improving the caveat here is that once we think about a more general payoff structure for b as well there's still inefficiencies and it still will not obviously dominate the legacy system so i think there's a trade off still that emerges but you're absolutely correct and that's part of the point that we are trying to make as well yeah thank you thank you thank you francesca and michael um it's 605 well in Amsterdam at least perhaps we have time for one last question and there is one in the q&a box um that is pretty high in the list that was by hogar um oh hogar do you do you want to ask your question live i i can do so yeah thank you sorry i'm not sure whether my camera is on but anyway um so my my question basically is uh is this tokenized token economy that you describe here is this actually tantamount to saying that uh if i have to own that security in order to make the smart contract on it that more or less the security is no longer fungible really i really have to have that token and that is being delivered also back at the end of the say of the repo trade um i mean and is there i mean is this actually inevitable can only this be done or could one also make a smart contract on something else with a wallet account whatever but but where where basically i can say okay i make a smart contract where i will deliver this security in say two days and then i still have the possibility to get it in the meantime of course this would mean that the settlement risk would not be addressed in this token in this token economy on the other side it would allow that the intermediary system can still prevail then in the future so that's that's basically yeah one of my questions so i think i agree with that and i think this is what michael was emphasizing earlier which is if you're willing to give up on settlement certainty then then you can get good outcomes and and perhaps take advantage of uh of programmability and so our our point is not to say that programmability is bad in the itself is that it's not a perfect solution to solving settlement risk because it's commitment with strings attached so i think one thing i wanted to add also to example is there's a sense in which it can create issues with fungibility but also uh programmability should give you a lot more flexibility when you're trying to say substitute assets because you get because it becomes easier to keep track of of them that's quite outside of the scope of our paper um but uh but i don't think that particular aspect is problematic in the kind of environment we have in mind um michael anything to add yeah if i think about in the context of repo markets the fungibility for example across like general collateral pools i think that's fine there's no issue with that because you can still commit to a certain type of asset with that belongs to a category of assets and that's fine but here you know the the benefit is actually that once by committing certain types of tokens now in principle as long as the distributed ledger contains information about future ownership you you may not even have to have the token in your account today and that's i think kind of the innovation that will happen even before we go into the kind of more intricate and complex usage of the programmability of assets so that representation of ownership i think is more pivotal to it now whether it the exact orientation of whether it's in an account form or token form is less important the dynamic representation of ownership is the most important future that we're kind of emphasizing here so let me add one last thing which i think is about all of the questions we've been discussing right now so there's a bit of a tension between saying like oh but we can make the the system even more complex than it is today and that will improve things and the fact that the reason we we decided to focus on on this idea of some uncertainty is because that's the first thing that the industry has been focusing on so we didn't invent that problem this is what a number of actual projects are looking at so if you think of this sort of you know world of like solution in in in search of a problem that's what the that's what a lot of the industry has been interested in and so we can always say oh but that's not the right problem we should be focusing on a different problem and that's perfectly fine but i think there's a bit of tension between sort of saying like what can we actually solve versus like oh we could make things a lot more complicated and solve other things I think thank you Antoine for this this sort of last last words uh in that sense uh i think we have to stop here it's um thank you very much perhaps there's a last uh last well not really advice in the q&a box i saw that people would appreciate it if the people if the paper could have a companion non-technical summary on Fox EU so that would be then the next project Antoine and Michael and and and i'm hope thank you very much for everybody for joining and for the questions and for the i think a very interesting discussion so i'll give it back to to Russell thank you everyone i think we have a very lively presentation today thank you will call thank you Antoine thank you jico um so just to remind on on may 20 we have our next session which will be presented by Simon Mayer currently at Chicago and will be discussed by Haseul Ferdinand or Pan and will be moderated by John Frost or BIS but in the meantime so there's a lot of cvdc events so the next next week so Haru Uleg of Chicago will have the new cvdc seminar series where we'll be talking about the cvdc in ukraine a very interesting development and the next week so we have the may 13 will be a conference organized by the cber so their second annual conference so keep an eye on a lot of events in this space i hope to see you soon i won't keep you guys too long so it is large right here and six in the u of people are okay good to see you all thank you thank you much thank you thank you very much