 Mae'n fawr i'n fwyaf. Welcome back to our conference today. I am very impressed at all, very punctual and on time. Thank you very much. We will start today with our first session, which is chaired by Andrew Heuser, the Executive Director of Markets at the Bank of England. Thank you very much. As you say, this panel is on non-banks and the lender of last resort. As a veteran of the LDI crisis last year, mae'n rhan o'r rhan o'r rhan o'r rhan o'r ffordd. We've got two great papers this morning. The first one is from Quentin van der Meyer from the University of Chicago Booth, who's going to be talking about the Central Bank's balance sheet and Treasury Market Disruption. Before the event, I was asking Quentin if he'd been up all night adding a cyberattack scenario to his paper. But like a good academic he said, what cyberattack? So I'm glad to see he's focused on the paper and not that. But Quentin, if you'd like to come up 25 minutes on your paper, thanks. All right, so thanks for having me. It's a great pleasure to be part of this amazing conference, an amazing programme. So indeed I'm going to present this paper called Central Bank Bank Sheet and Treasury Market Disruption. This is joint work with Adrian Daverna at Stockholm School of Economics and Damon Patterson with a PhD student at MIT now. Okay, so the great thing about doing this conference on day two is that I probably don't need much motivation at this stage. So basically Wanchin did all the job for me yesterday. So I'm still going to go through this slide, but quickly. So we've seen a lot of drastic evolution in Treasury and Ripple markets recently. We had a lot of market disruption. So Wanchin mentioned the quarter end disruptions between 2014 and 2020. We also talked a lot about the spike in September 2019. We've seen the Treasury markets being disrupted in March 2020 and then in the UK as been just said now in 2022. At the same time, I don't know what's going on with the slides, I still remember what's in it. So at the same time we had non-banks getting more and more involved into Treasury markets in particular. This has been the case when the Treasury cash future basis started opening. At that time we saw a lot of hedge funds getting in and trying to arbitrage this spread. And a little bit later we've also seen central banks getting more and more active in Treasury markets and Ripple markets by acting as lender of last resort and sometimes buyer of last resort. So it's kind of like the general background for the paper. So of course these are topics that have been studied before. So what is the thing that this particular paper wants to do? Well we want to provide a theory that can jointly explain all those facts together. So it's a little bit ambitious. Another way of putting that together is basically saying we want to find like a minimal set of assumptions that can actually join the experience of those facts. And so how are we going to do that? We are going to build a model, an asset pricing model, with basically four frictions. And I'm going to mention these four frictions in the next slides, which hopefully is going to be there at some point. But at this stage I want to mention that the way we are going to validate the model is by looking at different types of shocks. So when she mentioned the quarter end shocks for instance, these are typically types of shocks that happen when you have a foreign European, I guess foreign from a US perspective, intermediaries contracting the balance sheets because of regulatory arbitrage or more like a window dressing reasons. Okay, slides is there now. So that was short. Okay, so okay, is there? Okay, so I mean basically the second bullet points. I'm talking about the intermediation shocks. So these intermediation shocks are these quarter end shocks where the balance sheet of foreign dealers is basically contracting so that US dealers have to absorb some of the slack and because the balance sheet is costly, that creates some disruption in the market. There's a second type of shock that we want to look at, which is the net repro supply shock, which is typically the type shocks that will happen around tax deadlines. As you remember 2019 was actually around the tax deadline. The third type of shock is going to be some net treasury supply shock that will happen when you have treasury issuance, but also when the central bank does QTE or even when central banks are going to do some FX rebalancing. So let me preview the results already. The first result is that central bank balance sheet is really going to be the key state variable of this model and both sides of the balance sheet are going to matter for independent reasons. So on the asset side it's going to matter because it's going to alleviate the balance sheet cost constraints. The central bank is going to absorb some of the treasuries, whereas on the liability side it's going to matter because it will alleviate the reserve, the intraday liquidity requirements that some of these banks may have. The second result is that there's going to be a policy trade off between the shock frequency and the intensity. If you have a policy tool that makes the shock less frequent, you're also going to increase the intensity of the shock and the reason is that agents are going to anticipate that the shocks are less frequent and are going to leverage more. The shock duration is going to be very important in determining if it's the treasury market or the repo market that gets most affected and that's really related to the fact that there's going to be some transaction costs on trading treasuries. And then the last result is that if you want to have a repo facility, its efficiency is going to depend on the interaction between the design you have and the type of shock that you're facing and I'll be more precise. Okay, so it's a short presentation so I won't have time to go through all of the models so instead I want to show you the balance sheets. So this is what the model looks like. So we have central banks, we have treasuries, it's working. We have traditional banks, shadow banks, households, dealer and also foreign dealer. So like many agents in the model and there are really like these four frictions that I mentioned before. First, I'm going to assume that repo and deposits are imperfect substitutes for households. I'm going to capture that with some preferences but you can think about it as being basically anything that makes them different. Then we are going to have this intraday liquidity requirement which we are taking from an all year paper. So this is basically a constraint that is going to determine how much repo the banks can do for a certain quantity of reserves. Then we are going to have a balance sheet cost coming motivated by some form of leverage ratio creating some debt overhying type of problem as Anderson de Finsong has shown in the journal finance paper and then we are going to have that trading treasuries until some transaction cost and we are going to incorporate all these shocks that I mentioned. Okay, so that's for the full framework. In order to get some intuition about the paper at first, I'm going to show a simplified framework. A simplified framework is much easier and much simpler. We only have traditional banks, shadow banks and households and I'm going to replace the balance sheet cost friction with another friction that's even starker which is that traditional banks cannot borrow in repo and at the same time also going to remove the multiple shocks and focus on one single shock which is going to be a preference shock. So households will suddenly want to hold more deposits and less repo. So let me go through that model quickly in just one slide. So we have our three agents. They maximise their lifetime utility from consumption. The treasury bonds, they incur transaction costs new. Households, they get this utility from holding money in the two forms in the form of repo, in the form of deposits. They think about it as being intermediated by a money market fund. So they don't hold, in reality they don't hold repo but they hold repo through a money market fund. And this is going to be the only shock. So you have this preference parameter in this cobb do glass aggregator. So they want a mix of the two, a convex mix of the two. And the parameter in this cobb do glass aggregator is going to be shocked. Basically it's going to move from a lower level in the steady state to a higher level that is uniformly distributed with intensity or probability lambda. And then once you are in this crisis shock state there is a probability lambda prime from moving back to the steady state. Traditional bankers, they solve a portfolio problem where they can hold treasuries, they can hold reserves and repo and they can issue deposits to households and they are subject to this intraday liquidity stress test that tells them that the quantity of repo they can do is going to be limited by the number of reserves they hold. And the idea is that there is some settlement constraint in the back here. And so this is the additional assumption that I mentioned. They cannot borrow in repo and this is going to be relaxed later, as I mentioned. We have the shadow banks as well and the shadow banks are going to solve a portfolio problem of holding treasury bonds and borrowing in repo. And the main difference with shadow banks is that they cannot issue deposits and they cannot hold reserves. They are not banks basically. Okay, so let me start with the starker types of friction that I can have. I have this perfectly inflexible benchmark where basically all the bansheets are completely fixed. So the only thing that traditional banks can do is completely transform the reserves into deposits whereas the shadow banks they transform treasury bonds into repo and that's the only thing they can do. And then we have households consuming these two assets. And then what happens now if you have a preference shock to households? Well, bansheets cannot adjust. So this means that prices needs to adjust and that's what we see over here. So in the middle panel I have the repo spread, basically the spread between the repo rates and the interest on reserves and we can see it's a positive function of the level of the shock. And then we also have the liquidity services. We see that what's going on is that the households are going to move away from their optimal holdings of liquidity which is an interior solution of deposits and repo. Okay, so what's going on now if I'm allowing the traditional banks to land in repo? If the traditional banks can land in repo, then this is not a problem anymore because if there is a preference shock, the preference shock is basically met by traditional banks getting some of the repo from the shadow banks and then transforming this repo into deposits and then the model solves. So that's what we see here. That's going to put a bond to the spread exactly where the liquidity services are optimal and the repo spread is going to be zero all the time. Okay, what's going on now if there is an additional shock but the banks are subject to this liquidity stress test regulation? Well now there are constraints so basically back into square one. So we have this green region where the banks are extending repo elastically so we don't see the repo spread shooting so banks are acting as a lender of last resort in this case and then when the constraint is binding you enter into the red zone and the repo spread starts shooting up here even more because the reserves constraint is actually binding so you have an additional wedge coming from that. Okay, then the last thing that can happen I can also allow the traditional banks to buy some of the treasury bonds from the shadow banks and if that happens then even if the banks are at their constraints we are still going to see that the model solves completely because the shadow banks are basically just going to sell the treasury bonds so that the traditional banks can just extend the deposits and there is no problem anymore. But that's only going to happen whenever the profit of doing so is beyond the transaction cost. So that's what we see now we have this additional green zone that happens that also bounds the repo spreads beyond a certain threshold and when this is going to happen it's basically the size of this transaction cost. Okay, so that's the dynamics of the model here so effectively what we have is a model that moves between a steady state and in the steady state it's efficient for the model to have shadow banks holding the treasury because only shadow banks can create repo in this setting and households want repo and then the model is going to move into shock states where we see some the repo rates spiking up and potentially some fire sales of treasuries so this is going to create some liquidity risk for the shadow banks and the model is going to live between these two states and the shadow banks are going to trade off the marginal benefits of arbitraging if you want the cash future basis they have this comparative advantage of creating repo with the marginal cost of getting additional exposure to this liquidity risk. Okay, so let's think about this model dynamically and now let's ask the question what's going on if you change the frequency or the probability of moving into the stock shape the crisis states so in that case the first result is some form of volatility paradox what's going to happen is that when the probability of getting into the crisis states gets lower then the shadow banks are actually going to take that into account and they are going to get even more treasuries into the steady state regime which means that now when you get into the crisis states you're going to see a larger expected repo spread conditional on the shock heating which is what we see here and also a larger probability of entering into this fire sale of treasuries zone so we have like some form of like volatility paradox where there is some substitution between the intensity of the shock and the frequency of the shock so if you make the shock less frequent for instance if you have policy tools to do that you're also going to make the shock harsher so more market intervention in this case can create some fragility okay so let's move to the second dynamic result second dynamic result has to do with the probability of moving outside of the crisis states and the dissipation from the agents okay so here what happens is that when the probability of moving outside the crisis state is very large means that you're on the expectation you are going to stay in the crisis state for a very short period of time then in that case it's actually going to be better for the shadow banks to not to fire sale the treasuries but rather pay a very high potentially a very high repo rate because the transaction cost is a fixed cost whereas if you think the duration of the shock is going to be very long then in that case it's better off they are better off just paying the fixed cost once and then being done with it so they are trading these two things and potentially we think this can explain why we've seen in September the repo market spiking whereas in March 2020 it was more the treasury spiking probably the argument here is that in March 2020 it was more about there was a lot of uncertainty about the COVID shock so we didn't know when the shock would actually end okay so let's now move to the full model so I'm going to reintroduce the treasury, the central bank the foreign dealer and the dealer effectively the dealer and the foreign dealer they are just a pass through they are just here to pass on the repo just to match some institutional setting in the United States that most repo transactions are actually intermediated and here there is going to be a balance sheet cost on the domestic dealer because the domestic dealer is going to be a subsidiary of the traditional bank we are also going to have the treasury treasury is effectively issuing the treasury bonds against future tax liabilities of the households and it's also going to have a reserves account called a treasury general account with the central bank the central bank is going to do QE's going to buy the treasury bonds and hold reserves on the liability and also the reserves to the treasury general account so we have now I'm removing the assumption that the traditional banks cannot borrow in repo and I'm replacing that with a milder assumption serving exactly the same role which is that traditional banks subject to these balance sheet costs so the dealer as I mentioned they are just doing this match book intermediation the fact that the dealer are doing this intermediation means now there's going to be two repo markets the first repo market is going to be between the households and the dealer or between the money market funds and the dealer and I'm going to call it the tri-party repo market and then there is going to be another repo market which is going to be between the dealer and the traditional banks sorry and the shadow banks and this is going to be called the bilateral repo markets and potentially there might be a spread between these two rates so I'm going to call that spread the intermediation spread and I'm going to introduce three additional shocks not four I didn't get the time to think about that this night so the first one is going to be the foreign dealer capacity shock the second one is going to be a shock to the treasury balance sheet so this TGA accounts and then the third one is going to be about the better central bank balance sheet okay last thing you want to mention about the complete models that now I'm also going to potentially allow for the central banks so here's a representation equation for the balance sheet identity of the central bank I'm also going to allow potentially for a repo facility and a reverse repo facility to see how that modifies the model okay so the first type of shock I want to look at is this intermediation shock this quarter end shock and so we have some regression in the in the paper but basically these are regressions that have been done many times including by people in this people in this room so here I just say a question as if summary of what happens during these different shocks so on quarter end we see the repo rates like the bilateral repo rate shooting up we see the intermediation spread shooting up as well and there are p-volume shooting up too so we are going to try to capture that so let's see what it looks like in in this balance sheet representation so we have the foreign dealer that contracts on quarter end which means that the domestic dealer has to absorb some of the slack at some point if the reverse repo is active then the money market funds or here the households are going to prefer to lend to the central bank at the given policy rate at the RAP facility and so that's going to expense the balance sheet that's going to get into the balance sheet of the central bank and it's going to remove some of the reserves on the balance sheet of the traditional banks and because now there is some additional space on the balance sheet of the traditional banks traditional banks are going to be able to extend some repo so that's at the end of the day all of the everything clears all the balance sheet clears in this market but prices will need to adjust according to the balance sheet cost so let's look at that in terms of graphs as well so here I have a first case where there is no reverse repo facility so if there is no reverse repo facility the only thing that can really move is this intermediation spread so this intermediation spread is so the the three parts of the balance sheet cost so I should have said before so this graph is interpreted the shock is basically moving the balance sheet of the foreign dealer contracting the balance sheet of the foreign dealer meaning moving on the left in the graph so what we see is that the balance sheet when you move on the left to the graph the balance sheet cost increases which means that the three-party repo rate has to decrease in order for the intermediation spread to extend and that's really the only thing that happens here now I'm going to introduce the reverse repo facility if you have the reverse repo facility that's put a bound to how low the three-party repo rate can go and so what happens here is exactly what I described with the balance sheets effectively when you reach this reverse repo thresholds then the traditional banks are basically going to get into the market and start lending repo to the shadow banks and that's what we see here their lending goes up and that means that the spread is actually still remains constant as long as the bank can lend until they reach their LST constraint their reserves constraint at that time the banks cannot lend anymore and because they cannot lend anymore then we see the bilateral repo spreads the bilateral repo rate starts shooting up in this case okay let's think about what happens if we also have a repo facility if we also have a repo facility that puts a bound to how far the repo the bilateral repo rate can go because at some points we'll have the shadow banks getting to the repo facility and so that's what's going on here and basically the central bank absorbs all of the slacks but let's think about exactly what's going on what's going on here is that the central bank here is really going to act as if you have the repo also on the left hand side as an asset then you have the central bank as act as an intermediary and the reason why the market can clear better is because the central bank is effectively an intermediary that doesn't have a balance sheet cost so that's why the central bank can act but this is not going to work if the repo facility is only available to traditional banks and not to shadow banks so you need actually the repo to be available to the shadow banks as well okay so I can also think about what are the necessary conditions to generate this spike in this model where we actually need four conditions and if any of these four is not there then we will never see a spike in the model so we need a balance sheet cost we need a liquidity stress test regulation we need a transaction cost and also maybe somewhat surprisingly we need a reverse repo facility because otherwise all of the adjustments will be in the three-party repo rate here we won't really only see the spikes because the three-party repo rate is bounded so that all of the adjustment needs to happen on the upside instead of on the downside okay let's move to the tax deadline on this net supply repo supply shock on the tax deadline actually the empirical pattern is quite different what we see is that the bilateral repo rate is increasing but the intermediation spread is not and we see actually a decrease in reverse repo volume rather than an increase so let's try to understand what's going on here effectively we have a tax deadline showing is going to be that the households are going to pay their taxes so they balance sheet contracts and the TGA account is increasing the TGA account increasing means that it's also increasing on the central bank balance sheet which means that now we have less reserves on the traditional banks balance sheets and again like the banks can actually use the space on the balance sheet in order to lend in repo so that's what's going on here we can look at it from a graph from looking at these graphs as well in the model here the shock is interpreted as moving to the right so A little A is the volume at the TGA account so what we see is that at the beginning until you reach the LST constraints actually there is no movement in the spreads and the reason is basically the algebra or this accounting identity here that the banks can basically lend and so the spread that maintains the spread constant until the banks are at some point the banks are going to reach their intraday liquidity constraint and then the bilateral repo spread is going to shoot up but here we don't see the balance sheet cost actually does not increase the reason is that there is actually less action on for the dealer actually if anything decreases so let's think about introducing repo facility in this case so the reverse repo facility is never going to be binding in this case if you don't start from position where it is we can introduce a repo facility when you have repo facility the repo facility is going to bounce again how far the bilateral repo spreads can go and but there is something very different here which is that it also works if the repo facility is only available to treasury to traditional banks so you don't need to open it to shadow banks why is that well because the constraint here is not going to be the the balance sheet cost the constraint is really going to be uniquely the liquidity stress test thing so in that case the traditional banks are going to be able to intermediate the liquidity they get from the repo to the to the shadow banks okay so in the paper we look at additional shocks but I'm running out of time so I just want to mention this additional treasury issuance shock which is kind of like a hybrid between these two we have that the bilateral repo rate is shooting up and the intermediation spread is also shooting up and it also results in increase in TGA volume in the in the data and that's because at the time of treasury issuance the treasury gets the proceeds of the treasury issuance and put it into the TGA account which is going to again like drain some of the reserves available to the banks so we're going to see the similar types of arachmatics but as I'm running out of time I'm leaving that to the to the paper so this brings me to my conclusion slides so we have developed a generic room framework to understand the treasury and repo markets together and so here I want to stress out that I think in thinking about these questions it's very important to maintain some of accounting identity or like generic rubrium because and try to think about exactly what are the frictions that are going to prevent the market from allocating the liquidity in the place where it is needed and so that's what we try to do in this paper and we find that we can build a reasonably simple model that can rationalize all of the recent market disruption so at least many of them um we stress that the the facility access design is going to matter for some shocks but not others so that's something that's worth thinking about and also that this volatility paradox which is that if you have some discretionary intervention you need to be a little bit careful because it might be that the market get accustomed to you intervening and then just leveraging more and in that some point you have some issues intervening you can end up with like a much larger disruption than you would have otherwise thank you thank you Crentyn let me invite Lloriana Pellezon from the Leibniz Institute for Financial Research as a discussant Lloriana thank you so thank you very much for inviting me to discuss this paper that as you can figure out is a complicated paper and it's very ambitious so let me just you know try to figure out from my perspective what are the research question that has been addressed in this paper so we know there is a a lot of evidence that treasury are exposed to funding shocks originating in the repo market and this paper is exactly trying to figure out how on one side this shock have an impact and the characteristic of this shock have an impact on the treasury and on the other side how different shocks can create different type of disruption in the repo market and the paper is really pointing out on what is the role of the traditional banks in and all the constraints that there are for the banks to trade into the repo market and also what the and this is the part I think more new what is the role of the central banks balance sheet in on one side you know amplifying or in some case generating some shock in the repo market and on the other side also on providing repo facilities and reverse repo facilities that surprise surprise on one side they're creating a boundary on the spike on one side and the other of the repo rates but in some case can also help to generate in some sense or increase the level of the spikes clearly the objective is to propose a dynamic model of the treasury and repo markets I'm adding a nest because it's in the paper you know it's not just proposing the repo market but it's looking also to the three party versus the bilateral repo market so it's a very complex things and is in some sense trying to have a model that is able by you know looking to what type of shock you have to match perfectly the empirical evidence that we observe in reality and clearly is going to show on one side as I say the role of the central bank balance sheet on one side the treasury and on the other side the reserves in the liability side and the role that the different facilities repo and reverse repo are having in the treasury market disruption I'm stressing treasury market disruption and repo because you know in the model there is really an attempt to link the two but then I will mention also something that there is something miss in the story is motivating the analysis by you know all the event that we know very well and it's showing that indeed you can have a spike in the repo rate but not all the same in terms of intermediation spread in the repo market or reverse repo volatility or something like this so different type of shocks even if they generate the same spike in the repo rate do have different effect in the other part of the market this is why the model is very rich and is able to explain all these different things as you can see this is the model he presented very well and you know clearly he's moving from one part to the other very easily but he's not a simple model clearly it is the minimum that I say he needs in order to explain all the shocks but clearly it is a complex model and he's considering different shocks on one side the fact that changing preferences and this is the key part in the theoretical model by householder between reserve and repo a sort of demand for liquidity in the classical LM model you know individual do have a preference to old cash but if I'm giving to them a good remuneration they will move also to treasury this is what we know from the old literature while here is the same in order to move from deposits to let's say to the repo market to land in the repo market they need to have a remuneration this is simply what he's doing and then you know there are all these different shocks so clearly the topic no doubt is very interesting and the paper also is quite new and interesting very preliminary also the first thing that I'm asking is he claimed that you know we need all of them all these different parts in the model in order you know to justify this type of shocks that we have but I have I'm going back and I'm asking to all of you why do we have such a complex model that can create all these type of shocks you know so by looking to this graph one thing that we can think is that well maybe we can think to have a structure that can be simplified I don't know household that maybe land directly to to the treasure to the shadow banks or something like this rather than going by the repo or via the traditional bank balance sheet this is a general question that I think it will be also interesting to investigate and then it is clear maybe in the presentation it was also mentioning but really in the paper it is very difficult to understand what are these shadow banks are the shadow banks money market funds or the money market funds in this type of framework are in some other part because in his presentation for example he was mentioning today that the money market funds play a little bit the role of the dealer but it is in the paper it is very difficult and given the role that the money market funds are having now in the market and we know that they were playing also some role in propagating the shocks in the repo market it will be good if he is able really to characterize better in the economy of the model what is the role of the money market funds and then you know why do I I understand that he needs that households that are investing in repo in the market but in reality households that are not investing in repo they invest in money market funds or maybe they invest in treasury so why households are not investing in treasury why do we have all this other part that is so complicated so just to go to the fundamentals it will be very good and if households include firms that are investing in money market funds then I think it will be good if in the model you are showing to me where are the money market funds and and you know even if they are investing directly in money market funds you know how this is is a sense reflected in your model and in this way we can really figure out better how these shocks are going for example you have also and I know that clearly you know all the time you are considering the different type of player in based on what you need but for example here you get that directly house in another paper that you have directly householder are investing into the shadow banks thinking that the shadow banks is the money market funds so you know it is easy that I'm taking out maybe one of the ingredient of your model and then that and is more close to the reality and then the result are not any more there so you know how much do you need all these type of assumptions and why has to be explained a little bit more I think in the paper that all the in the paper at least it is not clear who are having access to the three party repo and who are having access to the bilateral repo if the household that access to the three party repo only for lending and not the shadow banks because only traditional banks can borrow into the three party repo this is a friction that I understand you you are in some sense imposing because is in line with the market I'm not an expert of the repo market in the US but you need to explain to me why you're how this who are allowed and who are not and if you open up the three party repo market to all the trader in the market are we solving some of the spikes that you are observing or not because you know so far we are just looking to the case where the central bank is in is a sense helping to bound the spikes by doing all the facilities but maybe we can also to think in order to reduce these spikes how to change the structure of the repo market maybe I don't know if your model changes the structure of the repo market can solve the problem or not and how relevant it is in your framework that you know you are really having a repo market a repo market characterized by the fact that if you borrow you need to give a collateral and in this case the collateral is a treasury that is exactly the two main assets that you are considering in your model treasury and repo but the fact that the treasury are used as collateral in the repo market is not something that you are considering in your paper because you know when the treasury there is a shock in the treasury price this will have also an impact on the repo because the collateral you know the quantity of collateral that you need to provide is changing and again this part that I think can be also relevant in your paper in your model is not considered maybe it is not but then at least tell me something about the role because at the end there is no difference in your framework having a repo or having simple an uncolateralized interbank market the key thing is that you just borrow without looking to the role that the collateral may have on the repo and then on the interest rate then you know clearly you are looking to some shocks but there are tons of shocks and that are also mentioned in the paper here and there you are referring to the march 2020 but the key point that I think it will be also good is that all the time that you are looking to the impact in the repo rate for the different shocks you tell me what's happened to the treasury endogenously and even this part is missing the paper because if this is a paper on repo and treasury you need to do the last mile and tell me when a particular shock in the repo market may be based on the intensity we'll have also an impact on the treasury price and then a feedback effect in all the different you know balash sheets otherwise you know you are telling me a nice story about the volatility paradox but later on you're not telling me anything about the treasury market unless it's producing some shock and well this party you presented it very well and I buy the story you know that clearly the intensity will have a different impact on the on the price or in the yield of the treasury but again there is one part miss that is you know how much this is will have an implication on the collateral value because if the treasury yield is going up then my capacity to borrow in the repo market will be affected and you know in the in the model and in reality shadow banks cannot borrow in the three-party repo rate repo market how relevant is this assumption and how relevant is to have this type of structure if we change the structure are we solving the problem and then clearly you are stressing and the model is also set up because you want to tell us something about the role of central bank repo facilities and I like this because you know this is really a new territory and also for the central banks this is something that they are implementing it but we are not having all the picture clear on what are then the potential effect but you are pushing in your let's say model to suggest that the central bank are becoming some sense the intermediary of the repo market because in this way they can control this rate and impose a bound both on the shock when the repo is increasing or when it's going down but you know we have a repo market we want that the repo market is also telling us something and you know muting all the capacities having in providing information to the market may be it is not what we want and so how can we really design central bank repo facilities in a way that we still allow the market to provide the information so can you tell us something on when the central bank should set up this type of repo facility all the time they should be there or with a very large markup or should this repo facility be activated all in extreme cases and it is better to penalize on quantity or on price so on the rates given that you have this model maybe you can tell us something more about also how to set up how to organize how to you know implement this type of facilities and then as I said you know you are looking to all the different shocks some of the shocks just create a repo spike some other create also an impact on the treasury some other create an impact on both I want to know about each of these type of shocks when is the case that you have a spillover effect in the treasury yield and and then you know also how this central bank repo facility not only mute eventually the repo spikes but also how they have an impact on the treasury yield because remember this is the main part that maybe we are interested because we want to have a safe risk free asset that is not too much volatile this is our main objective at the end so if you have a model that is telling us how to organize central bank repo facilities to reduce spikes tell us also how to this will have an impact on the on reducing the volatility in the yield but in any case very interesting paper I learn a lot I have a lot of other question but I will ask to you later on and I'm suggesting to all of you to read the paper thank you very much thank you Loreanna right I came prepared with some questions in case there weren't any but I think you have about a million there Quentin to answer but let me before you do if I may let me see if there are any questions in the room or online to add to that list because you've got a few minutes for discussion if there aren't any straight away there is one here if we could get a microphone and then Quentin maybe you come back on some of the discussant comments as well but let's get a question in the room yes thanks this comes back to why do we have this structure what Loreanna already mentioned it's kind of a philosophical thing right rather than saying starting with the primitives like we used to do in corporate finance where we say here's limited commitment and then we derive who writes what contracts from that we start with kind of a demand system right like we do in IO we just say there's here's what person x supplies and here's who person y supplies and we go from that which is valuable but I'm wondering if in finance we have the problem that this stuff isn't really stable because in you know if there's a big shock if there's a crisis or something the system might some of the markets freeze the system suddenly looks very different from your picture with all the balance sheets and we're no longer sure that we can use your predictions and I'm wondering how you think about about that part thank you very much any any other questions it's one at the back here thank you very interesting paper thank you if I'm not mistaken in September 2019 the Fed also engaged in the purchase of T-bills and in a way in this kind of market functioning purchases that Aldefi refers to and I was wondering if there would be a role for this type of market function QE or set purchases in your model and if why you insisted a lot on the importance of the access to the central bank facilities do you see any particular situations in which market function QE would be superior to these facilities and I wanted to have your views on that, thank you thank you and then there's a third question at the front here and then Quentin will come back to you probably only got a couple of minutes for answers just no thanks for the paper very interesting my question is very similar to the previous one in the sense that you consider repo facility by central banks and not as a purchase this is very important in my opinion because central banks repo facility has two limits I would say the first one is if leaders have a banashic constraint this limit also the participation to the repo facility and also the counter party limits by central banks that reduce the lending to each counter party so why you focus on central banks repo facility and not as a purchase which is a much more important instrument thank you contentious claim there but certainly a provocative one Quentin has come back to you yeah so I'll do reverse order so it's easier so we actually consider that but in different ways so the way the model works like we have the model laid down and the facilities are part of the model but we also consider in the paper what happens when the central bank buys the securities and when that's the case that's going to reduce the ban sheet costs that's effective the way the way it works we also create some more reserves so that's what I said at the very beginning didn't have much time to show that but like the key state variable of the model is really the size of the ban sheet and like in both sides are useful in this case so this would help as well in the model okay on your question about the stability of the model so I think we are somewhere in between in the sense that we are taking institutional settings as given that's kind of like the epistemology of the paper we're like trying to take we are taking the institutional details as given and then we try to understand what's going on in that world taking that as given but we also consider anticipation so the whole model is like fully dynamics in the sense in that sense like proof to look at critique so I think it's actually reasonable it's reasonably stable to the extent that we are remaining within these institutional settings which can of course change which is going to be different different jurisdictions and things like that but okay so maybe it's a good time for me to answer some points to Lauren's discussion so thanks a lot that was very helpful I'm looking forward to chat more so the first question you are asking yeah it's kind of like almost a philosophical question how do we end up in the system as we have it now I don't think that's the question that this paper is trying to answer except for maybe one element which is that which is about these intermittent disruptions that's kind of from the genus in the model here it's like it's really generated by the balance sheet cost so the balance sheet cost gives a comparative advantage to the shadow banks in holding treasuries in steady states and they are willing to do that despite taking some liquidity risk but it's like that's because of regulatory arbitrage so that would be there so what are the shadow banks and you write there's like a lack of consistency across my two papers so I should apologise for that but in this paper it's very clear like the shadow banks are going to be institutions like hedge funds or like a secret dealers or foreign investors that are effectively holding treasuries finance in the report market so there are not many market funds the money market funds are here effectively within the household sector so I was just thinking about the the money market fund as being a past truth that's why we don't have them in the model and they are just like from repo to money market funds to the dealers and so we just forget about the money market funds so okay so that I addressed so on the using okay taking seriously the fact that the repo market is collateralised there might be effects on the haircuts some amplification that the treasuries are going to reprise that's going to matter all of these things are going to matter empirically but as you mentioned the model is already quite complicated so that's the reason why initially we had all these things in the model and we just decided take it out because it makes the model too complicated maybe you can have it in an appendix or something just to this would all these things would just create some amplification of the dynamics we have already here and then probably just one last thing if I remember correctly oh yes you ask should we have the repo facility at all times or not and that's not really in the paper the only thing that's in the paper is like there's a little bit of a caution against discretionary intervention and that's coming from this variety paradox something I mentioned already is that if you want to intervene if you think you can intervene 99% of the time then the 1% of the time where you don't intervene things are going to get very bad that's what the paper is saying so in that sense it would be a justification for actually doing something like a repo facility which is like an automatic intervention you're sure you're intervening 100% of the time maybe maybe not maybe there's something else that the institutional details is not really predicted all right well thank you Quentin thanks for a much excellent discussion let me now introduce Jeremy Stein from Harvard we're known to many of you who's going to be talking on a slightly different theme central banks as dollar lenders of last resort Jeremy okay thanks let's see if the thing comes up ah okay so this is a joint paper with Gita Gopinath two of her colleagues at the IMF Metali Das and Taehoon Kim and one of our graduate students Haleen Hall and so this paper takes up Gita and I had done some earlier work trying to think about the role of the dollar and why the dollar is sort of a dominant a dominant global currency and that has many facets so the dollar is very important in trade invoicing it's very important in international banking many many companies outside the U.S. borrow in dollars and then finally central banks outside the U.S. hold large volumes of dollars so this paper is going to be not so much about the trade invoicing but about the other the other aspects of of this phenomenon so just to give you a sense of magnitudes yeah here we go as of this year non-U.S. central banks hold something close to 7 trillion dollars of reserves about 4 trillion of that is treasuries the rest of that think of that as being like agencies and things like that roughly 60% of all foreign currency reserves are in dollars also if you'll recall the treasury market disruption in March of 2020 foreign central banks were big sellers so this is by way of saying you know these foreign central banks are probably big players both in terms of the level of the rate as well as some of the volatility that we see in markets and so you know one question is why do they hold so much they're a variety of motives we're going to focus on one particular motive but I don't mean to sort of you know downplay the others but the one we're going to focus on is the idea that many non-financial firms in especially in emerging markets tend to borrow in dollars they do that I guess because it seems cheaper to them at the time but when the dollar appreciates that can cause trouble and one reason for the central bank to hold dollars is if we get into a bad state of the world and they have to bail out either the non-financial firms or the banks they'll be better positioned to do that if they're holding dollars so the paper basically has two parts one part is essentially a bit of positive economics is thinking about what are the motives of an individual think of it as an emerging market central bank what are their motives and why do they kind of hold what they hold and then the second part is more normative in nature it's making the observation that if you let each individual central bank do what they think is optimal in terms of accumulating reserves you get an externality because no individual central bank internalizes the fact that when they hold reserves doing that tends to push down dollar interest rates that tends to make it more attractive for non-financial firms in other countries to borrow in dollars and in some sense you're chasing your tail right central bank a is holding this because their companies are borrowing in dollars but in doing so they're making the problem worse for central bank b because they're sort of exacerbating the mismatch so I want to sort of talk about that externality okay so just to rehearse the argument because we'll see how we do with the actual model there's going to be a model of an individual small country central bank there's some risk of a banking crisis and the central bank is going to bail out the banking system if that happens that crisis is going to be more problematic to the extent that some of the firms or the banks essentially have dollar denominated borrowing okay dollar reserves are a nice thing to have from a risk management perspective because your profits on your reserves will occur in exactly those states of the world where the bailout is most expensive that is to say when the dollar is appreciated so as a risk management thing it's attractive for you to hold reserves it lowers your having to instead do the bailout by having expensive x post taxation okay so it's attractive to hold reserves now there are other things you can do as well you can be more cautious in terms of like bank capital requirements okay and an individual central bank is going to strike some balance in doing that right they'll they'll rely to some extent on capital requirements some extent on reserves what they don't internalize is that when they again when they accumulate reserves that's going to tend to push down dollar interest rates and induce more mismatch in in in other countries okay so if you were imagine a global planner the glow if rather than individual central banks are setting this all up the global planner would say let's do more financial regulation and let's do less reserve holding okay because of this externality through the interest rate but I think an important observation I want to make is this is a second best argument in the sense that the only reason you care about the interest rate is because you can't control the dollar mismatch directly okay I'm going to allow basically regulators to set bank capital requirements but what's probably less realistic is the idea that they can go into these that they can go to their non-financial firms and tell them in which currency to borrow if they could if they could this is sort of a hypothetical if they could you control the mismatch directly there would be no need to use interest rates basically to do this so this is a familiar argument in other settings where you know you can use this in a monetary policy context if you have perfect financial regulation you don't need to worry about monetary policy affecting financial stability if financial regulation is imperfect then the interest rate has a financial stability aspect okay so this is this is something that's in that genre all right there are some empirics in this in this paper they're pretty sketchy and it's a theory paper so I'm just going to like tell you sort of a thing and it's it's sketchy so I'm not I don't I don't want to claim too much but if you plot this is just to show you that it appears that at least in some cases mismatch is one of the motives just one but one of the motives why central banks hold dollar reserves so this is showing you on the horizontal axis the dollar denominated liabilities of the non-financial corporate sector scaled to GDP and dollar reserves again scaled to GDP on the vertical axis so there seems to be some correlation countries where the non-financial firms do more dollar borrowing are also countries where the central bank holds more in the way of dollar reserves if you cut into this you can't see this it's it's pretty robust among emerging markets and you know you'd have to squint to think that there's something much more than that in advanced economies okay so again the picture I want to leave you I think this is a thing or at least it's it's a thing relative to sort of the other variables that people look in this literature there's something about this in EM and I wouldn't want to claim much more all right so let's see if I can do a quick run through of the model so so pretty simple model there's two dates there's households households can save in one of three assets in equity which is risky or in safe local currency deposits or in safe dollar denominated deposits okay and I've just made some assumptions here pretty brute force there's a there's a preference just a money in the utility function kind of thing there's a preference for local currency deposits and there's an even greater preference for dollar deposits and so their interest rates are going to be you know there's going to be a higher rate of return Q here is one over one plus the the interest rate so there's a higher rate of return on equity there's the next is is local currency deposits the lowest rate whoops right I did this at the IMF once I fell off the stage all together this feels a little too familiar so anyway the first two of these guys are exogenous and then the dollar interest rate will depend on the supply and demand of dollar claims okay so that's the only that's the only endogenous rate here there's an exchange rate it's exogenous it can be higher or higher or lower there's what I'm going to call banks but what you should think of is basically I've smooshed together the banks and the and the non-financial corporate sector here okay so the bank is raising money for projects and it can finance itself with one of three things it can finance itself by issuing deposits in either currency or by issuing equity and this is just anyway that thing in the middle there is just its balance sheet constraint okay and then there's going to be some probability of a banking crisis and if there is a banking crisis some of the banks are going to be basically just disappear the revenues are going to go to zero and then the government is going to step in and bail out the depositors of those banks okay and just for the moment and just to make this a little simple assume that the banking crisis occurs independently of the realization of the exchange rate more realistically it's going to tend to occur when the local currency is depreciated and I can put that in in a minute so first of all what are the banks so if you just leave the banks to themselves again this is the banks plus the non-financial corporates what are they going to do in terms of dollar borrowing it's going to be a very simple trade-off between on the one hand dollar borrowing is cheaper and s here is the spread that's basically how much cheaper dollar borrowing is and they're going to trade that off against the fact that even if they're not bust if the dollar appreciates and you're basically borrowing in dollars that's going to make you more liquidity constraint so I've got this liquidity constraint parameter gamma I don't know why this doesn't so the banks are basically going to borrow more in dollars if the spread is bigger and less in dollars if it causes a liquidity constraint mismatch problem what is the central bank going to do well the central bank can hold dollars but there is a carry cost to them of doing that because if they hold the dollars those are lower yielding than other stuff they might hold for example just you know general assets of other sorts okay so on the one hand they're going to not want to hold it on the other hand there's going to be some some ability to hedge against the fact that if they don't have reserves and they have to do a bail out they're going to have to raise taxes and I'm going to assume there's some deadweight convex deadweight cost of taxation okay so you get a very simple trade-off here it's it's you know it's kind of what you'd expect if there is no carry cost I don't know why I can't make this thing show up on here anyway at the top line if there's no carry cost if the SK is zero they'll basically hold enough reserves to cover the bail out entirely okay on the other hand as the carry cost goes up they're going to do less with reserves and more with expo taxation okay it's a very very sort of simple trade-off for the for the central bank the one thing I want to emphasize here people often say well can't you do the same thing with central bank swap lines as reserves no they're very different at least in this model what the reserves are doing are they're playing a risk management role that is to say when you hold dollar reserves you're transferring wealth to states where the dollar has appreciated all right you have this this pile of dollar reserves that'll be worth more if the dollar appreciates that's exactly the state of the world where your bail out is going to be more expensive because you've got to bail out these dollar denominated liabilities is very analogous to corporate risk management you basically got costly external financing here it's taxation so you want to move wealth to states of the world where it's most needed okay swap lines don't allow you to do that swap lines are basically state by state you can borrow against the wealth you have in that state but they're not allowing you to move money from one state to the other okay so not to say of course that swap lines aren't valuable but they're just a different tool they're a liquidity tool this is in some sense an insurance tool where you're moving you're moving wealth across states of the world okay and then finally you can think about capital requirements to do that you've got to write down the planner's problem okay i'll spare you i'll spare you the details but this is sort of what the planner cares about and you know what they're sort of thinking here is on the one hand capital requirement is costly because the banks like and you know the households ultimately own the banks they like borrowing in dollars or they like borrowing just cheaper than equity and that's in the creative value basically for the households because the households get utility for money so you don't want to do too much of that that's the capital requirement on the other hand if you don't have much of a capital requirement you're going to be stuck with more in the way of kind of crisis costs which you're going to have to manage either with reserves or with taxation okay so there's going to be an interior optimum here and not surprisingly it's going to depend one thing I should say is a capital requirement just tells you how much equity the banks have to have it doesn't tell you the mix of dollar funding versus local currency funding right so it can only pin down the sum of those two the banks will then choose their optimum which is as before okay so in effect all you're doing with the capital requirement is controlling local currency borrowing deposits right you can't control them you can't control the sum okay and so you'll do that basically you'll do less capital requirement when there's a lot of value created by local currency deposits okay as a hypothetical you can also ask what would happen if we if we allowed the central bank to directly control both capital or send you know essentially control the entire capital structure okay and they would like to do that okay because they really don't like mismatch let me I'm not going to show you all the all the math let me just give you a little example here so in a world in the first column when there's no regulation okay no capital regulation the banks won't hold equity capital voluntarily okay they tend in this example to do a lot of the borrowing in dollars that's the 67 and as a result the central bank feels like it needs to hold a lot of reserves if you let them only do capital requirements they'll push the capital requirement roughly up to 10% in this example what they'd like to do but I'm not really going to kind of allow them to do is if you let them do both capital and funding they'd say not only do I want capital but I want to change the funding mix I want my guys to be borrowing much less in dollars and much more local currency because of this mismatch problem okay so that's the last that's the last row over here that's the dollar borrowing goes down from 67 to 14 okay now that's the thing to keep in mind when we go to the global thing is that if I shut down if I have imperfect regulation where I only allow them to do capital but I don't allow them to directly control funding their second best way of doing that is going to be by caring about interest rates okay and that's where the externality is going to come in all right so the global economy is going to be basically the same as before there's just going to be lots and lots of these countries there's going to also be the US which I'm going to have to kind of keep track of because the dollar interest rate if I want to do the welfare right I've got to take account of the fact that it like it affects US taxpayers so the US is just very very small role it's just basically borrowing in the bond market and it benefits so US benefits from lower lower interest rates all right yeah don't look at all of this stuff other than to say I'm assuming that there's a downward sloping demand for dollar assets so the more dollar assets there are the the lower will be the interest rate okay and reserves basically if central banks hoard a lot of reserves there's fewer dollar assets available for everybody else so that'll tend to drive down interest rates okay and so the only thing that's sort of different between the global planner and the local planner is they understand that these reserve accumulation decisions affect interest rates all right okay so here's the only thing to look at and just only to look at a little piece which is this whole planner thing is a godawful mess if you do everything and you sort of net out all the transfer terms you can write it like this you can this is saying what is the how is how does the planner's welfare depend on dollar reserves right so this is what's the planner's interest going to be in essentially moving around dollar reserves it has two pieces sooner or later I'm going to either fall off the stage or I'm going to get this thing right okay so part of it is they're just aligned with the local planner and then there is a wedge between the two of them and the only important thing to look at is this thing phi which says there is only a difference there's only an externality there's only a difference between the global planner's perspective and the local planner's perspective to the extent that this phi thing is non-zero what is phi it's d dollar borrowing d reserves so it's only if the reserves the reserve holding decisions influence mismatch by the corporate sector okay that in turn is only non-zero if reserve holding decisions affect interest rates so okay so in this imperfect regulation setting the only reason there's a divergence between the global planner and the local planner is the global planner thinks ah if I let's say get them to hold less in the way of reserves that's going to affect the interest rate and that in turn is going to lower mismatch okay that's the only otherwise we're perfectly aligned okay so doing that I'm going to skip this one here's one way of thinking we have a result it's not it's not the case that it's always true that the global planner is going to want lower reserves but there's a very intuitive sort of a scenario where they do which is if there is mismatch in the following sense if I hypothetically allowed you to control dollar borrowing directly if I said here I'm going to give you an extra let's pretend I give you an extra tool and I let you control dollar borrowing directly if you did would you lower it below the the free market choice okay the answer to that is yes I'm going to say there's the the market basically generates socially excessive mismatch okay if the market generates socially excessive mismatch in that sense then you'll always want to control reserves why because that's the only way of getting mismatch down okay so I'm sort of taking the what we're doing is taking as given that basically the private sector the incentives of these private sector firms is to do too much dollar borrowing why because they don't internalize you know what this does in a crisis state of the world so if I say to you boy wouldn't you really like to control that directly as in my previous numerical example but I take that away from you well what else can you do is try to operate on interest rates right it's the same thing again as in many other contexts if I say wouldn't you like to have really good financial regulation so we have no financial stability problems well if I take that away from you make it imperfect maybe you have to bring interest rates into the game okay same story same story here so that's that's what we've got here a sort of almost another way of saying the same thing is if I do give you perfect regulatory power and I could allow you to basically not only regulate banks but regulate all the non-financial firms in the economy and just tell them guys no mismatch then there's no more externality through the interest rate right because the only thing was because you're doing it because as a second best way of controlling mismatch but if you can control it directly you're done okay so that's that's the story that's the story there let's see how am I doing yeah so I have some another new couple of numerical examples I'll skip these and just say a few words about a possible extension which is so far to just focus on this externality I didn't say this but what I had been assuming when I did the global case is that everybody's banking crisis happens at the same time okay I did that because I was setting aside what's another natural motive for economizing on reserves is just risk sharing right I mean this is why you know this is why people have credit lines from banks right each firm could accumulate enough liquidity to kind of get through a bad state of the world but boy they'd be holding a lot of liquidity if their needs for liquidity aren't perfectly correlated better to have it sit in a bank and the bank give out credit lines right there's an identical logic here if the financial if the crises hit if the crises hit different economies in a non perfectly correlated way another motive for essentially economizing on reserves is if you have some way of redistributing right of you know basically you know take the extreme case where the things are perfectly negatively correlated we don't both need to be holding enough to get us through only one of us does and we've got to just somehow have a mechanism for reallocating it's interesting you would think you would think that this is sort of a natural role for the IMF right for the IMF to be the bank in my example you know and the countries to be like the the firms the IMF does have these credit line facilities that look a lot like what we have in mind it's basically a form of insurance across states okay they have in that there's a couple of these things ones called the flexible credit line the others called the precautionary and liquidity line as they're currently set up they have relatively strict eligibility requirements so there's actually very few countries that use or have used them but the model at least suggests that something like this might have this other virtue of essentially helping with this with this externality through the interest rate is interesting because the IMF's head Crystallina Georgieva just recently gave a speech or had a paper which wasn't this specific obviously as our model but was saying you know there's a lot of self insurance with with countries holding reserves but jeez you know it might be a good thing to strengthen the role of the IMF in providing in providing this insurance so you can think about this this model is at least sort of trying to relate to to that to that way of thinking okay so I think I okay thanks thanks very much thank you Jeremy and now I'd like to welcome Yvgenia Posari from the University of Paris to find as a discussant so hello everyone a big thank you to to the organizers for including me in the program I'm very excited to be here and to discuss this paper and I certainly learn a lot in the process so I'm going to start again in a similar way with with the starting point of the paper the main idea is that central banks tend to hold large volumes of dollar denominated reserves and they do that for a number of reasons right so so I don't want to repeat the presenter there is obviously the role big role of invoicing and and some of the authors have earlier contributions so a lot of the countries are trading directly with the united states it's a good thing to hold dollars and then there is this idea of risk sharing right because the US is an extremely developed country financially so effectively there's a counter cyclical safety premium into holding dollars and there is the natural incentive for central banks to to hold dollars as they want to intervene in the FX market to control the level and the volatility of exchange rates and then there is a lot of papers now talking about the role of the dollar as a global reserve and global anchor so there's an appeal again of holding these currencies and I'm going to to basically focus and not just me like the paper is about has this view of the world that central banks they have a precautionary motive of stockpiling dollar reserves because effectively they're concerned about a potential bailout essential event so the way this works there's a currency mismatch that that's a fact that among others is actually that has documented that for the emerging market space along with co-authors so there is a currency mismatch in the private sector liability composition so then again effectively it makes sense for the central bank to have the dollars for an event of a potential bailout now there are two main questions that this paper is asking so the first one the authors are thinking about the potential implications because as a result of this you know stockpiling that happens across economies globally and the second one whether there should be a friction when it comes to the incentives of individual central banks as they're thinking about as they're planning their policies and that with a global planner that that for the paper is going to be the US so basically that is going to see certain effects on the dollar and on the interest rate as a result so the paper starts with an empirical exercise that gets again motivated by the findings of Wenxi and Jesse so but Wenxi finds in this paper is that firms like corporate firms more importantly different than the government keep effectively running these currency mismatches in their capital structure so this paper is about emerging market firms and there's a correlation between this finding and sovereign defaults so starting from that the authors try to basically draw this meaningful correlation between the central banks reserves in US dollars and given that we cannot directly measure this mismatch that's extremely hard to get in terms of data and these are confidential data there is a proxy that has to do with cross border dollar denominated bank boring of the corporate sector so there's a crucial hypothesis that goes when they're assessing this correlation and that is that effectively this dollar denominated bank boring approximates well the corporate sector's currency mismatch I'm going to come back to this point in a second and then the logic is that effectively when you have mismatch firms when it comes to their balance sheet then you have a higher probability of a banking crisis and that is going to fuel the motive of the central bank to act as a dollar lender of last reserve and hence accumulate reserves so I'm going to very quickly go through the model so it's a two-period model of optimal reserve accumulation for a small open economy central bank that basically is trying to mitigate this bailout risk so this model borrows a lot of elements from some of the authors earlier work and you have utility maximising households that save in home currency denominated safe assets then safe assets denominated in dollars and also some home currency equity and then consume in home goods so you have then the banks that really are not the banks conceptually but these are the firms this is the non-financial firms that seek to minimise the sum of the expected funding and the mismatch costs and then you have the central bank right who tries to avoid a bailout event and in that effort they have free tools right so they can stockpile dollars in terms of reserves so they can just step in and act and bailout in a better way and then effectively they can set effectively they can request from the banks or from the first if you want to be better capitalised all right and the third one is to directly effectively intervene on the mix on the deposit mix of the banking sector which is not very credible because there's not an easy way to do that right so that was already explained so there are two natural trade-offs arising from these incentives so on one hand when you stockpile dollars what you have is reserve carrying costs and deadweight cost of taxation but on the other hand if you impose higher capital requirements what happens is that you're harming profits right so the more you retain the less you can invest you're harming profits you're harming social welfare so what is going to happen in the model where we're going to have two outcomes depending on basically whether the equilibrium is going to be locally chosen or we're going to have an equilibrium with a central planner so if it's locally chosen then the central banks are going to choose a mix that is going to allow for a higher stockpiling in dollars and lower requirements because effectively they don't want to harm welfare but what they don't internalize in this process that when they collectively engage in this behavior what effectively they do is that they're compressing down the dollar interest rate and that's an issue because then what happens is that their firms are going to engage they're going to binge into more dollar borrowing and that is going to exaggerate the problem right the risk of the bailout so when a global planner is allowed to take this decision they're of course going to think about that so they're going to choose a mix that is going to have lower dollar reserves and higher capital requirements so I think that's the story there are extensions there is a correlated risk between banking crisis and dollar depreciation and so on and so forth now the paper makes several contributions so if we think about the small open economy version of the model the authors are able to match some of the empirical findings and then I think obviously the main contribution of this paper is theoretical and it's conceptual even is this equilibrium when you have a global planner right so you have these externalities and then it's very interesting to sort of think about this the model's normative properties now the authors contribute to the precautionary view of reserve stockpiling so there are two literatures there that I stockpile because of trade so it's a mercantilist view or I stockpile because I want to basically mitigate some risk in the financial system and this is where this paper falls and there are very interesting normative essentially conclusions for regulators so it's very thought provoking to think about a centralized institution such as the IMF who's going to coordinate and engage in this allocation of reserves every time a country has this need so actually this is going to be you can think of it as a separate mechanism over Basel right so you have Basel for capital requirements and then you have this centralized institution that is going to give you instructions about how the dollar reserves are going to be allocated so I'm going to basically have few comments so given that I'm in a process by training I'm going to start with the data and then I'm going to try to think about the missing elements of the model so the authors have this first empirical exercise we have already seen some of the plots however there is an alternative explanation to those scatter plots especially if we focus on the develop markets panel so central banks can accumulate dollars as we both started the presentations to stabilize their currency or to prevent it for further strengthening because of trade or different reasons right so case in point and that's going to be a cheap shot on my part so I apologize I'm going to take the Swiss experience right so it's part of your develop market sample so your data is starting 13 but you do have inside this period between 11 and 15 where the franc is pegging to the euro so the franc kept going up in value because at that point in time it was perceived as the last remaining safe heaven right so Europe was battling with the sovereign crisis the yen was artificially held low and then you have the economic policies in the US that caused a lot of uncertainty so effectively that became an issue for the swiss companies right who were considering even moving operation out of the country so the incentives for the reserve accumulation in dollars were different so on the left panel of this slide I have your your scatter plot in the advanced economies and on the right hand side I have a graph from Chinyto on McCollie 2021 that is to say that throughout this period the swiss were not mainly stockpiling dollars they were stockpiling euros and but that's a good site for your story because you see those two peaks right in the euro accumulation so they actually coincide with the great crisis the first one and the second one coincides is the period be be just before the statement by Mario Draghi of whatever it takes so you see that in crucial points the swiss stockpile euros potentially to intervene with a motive that lines better with your story now that does not change the fact right that throughout this period there is a statement by the bank that you know they they want to intervene essentially and they're prepared without most determination to purchase foreign chains in unlimited quantities to bring down the franc but what I'm thinking is that by looking at your developed market panel where you have a lot of countries that could care about the euro you could basically exploit this period and show this fact but for the euro better for this period now is that problematic for your story no it's not because it does not change the fact that globally countries are stockpiling dollar right but it will strengthen your motivation especially for this panel because its composition tends to to show more of a motive when it comes to dollar that aligns better with invoising stories or interventions now there is a technical issue that's a bit geeky feel free to ignore so when you're looking at these cutter plots there are certain assumptions right number one these are extremely hard to measure variables of interest and and I appreciate you do your best number one when you're looking at cross-border dollar bank loan data you actually creating different distortions in the measurements I think so for the developed markets panel right you're trying to approximate always the currency mismatch now there's a good reason to think that developed market banks are better hedged right so what could be happening it could you could be overstating the mismatch in that panel and then if you look at the correlation between cross-border lending and total lending it seems that that's weaker on your emerging markets panel because you're missing the dollar banks right lending reserves and there's again this possibility that emerging market firms rely more relatively to local bank funding so the second correlation could be slightly understated so as you said obviously when you're looking at these cutters I agree with you the emerging market panel comprises of countries that really tend to borrow in dollars right to reserve basically sorry to accumulate reserves for precautionary reasons now let me go to the missing ingredients so the model focuses on one channel this precautionary view to to avoid the crisis now obviously there are different incentives especially for your developed markets country as I said before the traders are different that becomes a little bit crucial for your model I think because the most important holders of dollar reserves are not necessarily the emerging market countries right they tend like you can think about okay you have China there but then again the motive of China should be more like a trade story you have Japan that I don't think potentially stockpiles for a bank run so you have different types of countries right so the developed markets country panel includes a number of net credit donations and net exporters now is this a problem for your model again not necessarily for what the global planner is thinking because to the extent these guys are stockpiling dollars for whatever reason then the dollar rate is going to be lower is going to be compressed and that's a problem at least for the emerging market sample because that brings up the possibility of a bailout right so that's fine your story goes through but what one would think is that you could model the individual central bank trade-off in a richer way right so to include a trade-off about inflation or to include a trade-off about invoising for instance now some other minor issues so firms response to policies there is a possibility of regulatory arbitrage right we have this example of Chinese firms so once the government tried to cut credit in firms in a risky sector what happens is those guys like the shadow banking like increased and they engage in intra firm lending and there's there again I think that the last part of the paper which is extremely interesting you're very open about this their moral hazard consideration political frictions and issues of coordination I'm not saying this paper should solve them but then again they call potentially for a richer structure so I think it's an extremely important topic it's a rich model in a sense of you know this externalities and the global planning like I really enjoyed it and I think it's a really significant contribution and in the way we get to think about those issues that are becoming very important because the model gives us a very useful toolkit to basically address those issues now I don't know if you need the empirical section but maybe you can make it work for you once you tweak it a little bit and I think there is room for enriching the model or for a second paper and follow up research so thanks great thank you very much we've got just a couple of minutes to take any questions in the room there are obviously some points there for Jeremy to come back on as well does anyone have any questions that they'd like to raise there's one right over here in the far side so it seemed in your model the demand for dollar deposits is because local depositors get sort of extra money of the utility function from dollars I'm wondering the extent to which you think dollar borrowing is so that international lenders can reach you and if that's the case whether you think capital controls are effective for the externality you have in mind thank you anyone else if not Jeremy, why don't I turn back to you so on that yeah it's a good question this model is not all that well suited in the sense that in our previous model I think we had a better motive for people wanting to hold dollar deposits because they were hedging the fact that they had to purchase goods that were invoiced in dollars so that gave you kind of a micro here it's so ad hoc that I don't really know how to quite think about this very good question I don't have a good answer for you but it's a good question thank you for a terrific discussion the empirical points are all very well taken I mean I'm a little bashful about the empirical section I thought it was useful to kind of have it as motivation but you know both because we ran into just more data limitations than you would think you would run into in this thing in spite of the fact that we tried pretty hard and because again there are lots and lots of motives and I think your swiss example is a good one I you know I that's part and the swiss the swiss data point actually played a pretty big role in the advanced economy results to the extent there was one that's part of the reason I was down playing it precisely because of your your thing I think there's something going on here in EM but then you're of course right that the EM countries by themselves aren't explaining a lot of the mass of dollar especially if you exclude China are not explaining a lot of the mass of dollar holding so that's all very fair you mentioned one thing about the FX mandate of central banks and do they hold reserves for that reason one sort of limitation of the model but I think we could sort of fix this is we take the exchange rate as exogenous but of course if it was a little bit endogenous and if the central bank could manage it there'd be a motive that's very similar which is you know one reason you don't want the exchange rate your local currency to depreciate against the dollar is that makes a banking crisis worse so you'd be intervening you'd want to hold reserves in a sense for very much the same sort of reason and I think the overall logic would would probably go through but let me let me stop there but thanks again for for to thank you I think we have one question here well maybe two at the front since one of them comes from the organisers he'll be well aware that we're already over time and I'm sure it will be a concise question wait for the microphone from the 80s and 90s spare headed by Jean-Marie Melissa Ferretti Philippe Lane our chief economist who spoke yesterday also discussing the overall level of countries foreign indebtedness and this literature ascertains that the level below 60% of GDP was considered sustainable but in some cases 30% was not not sustainable in some countries so these authors looked at the the strength of the current account balance the trade balance and so forth so I wonder if you can also build an anchor to this old literature and differentiate when you have these scatter plots that are all dense all these countries you can add a few more charts along the line suggested by Eugenia but also considering the overall foreign indebtedness private and public from the various countries and you may tease out some more empirical results okay and then I think one question here and then we'll we'll conclude no thanks for the paper no my my question regards the the choice of central bank to avoid in order to avoid increased tax in t in t plus one plus two they decide to accumulate reserve now but maybe it could be some this trade off could be explained a bit better in the modern sense that the cost of community reserves is not only buying dollar now but also demoralized and other aspect that maybe it could be more convenient for the central banks to not accumulate reserve now and wait for t2 in case to increase tax later than stockpiling reserve now so if maybe why you didn't elaborate on this aspect in the model great and if I may if you can introduce a liquidity regulation to address the regulatory witness which kind of regulation you want liquidity cover ratio or mandatory edging or dollar exposure for corporate thank you so much thank you okay so on moral hazard there is moral hazard in the model in the sense that the bailout creates a form of moral hazard and it encourages people to it encourages the individual actors to overborrowing dollars in part to the extent that the bailout comes in a dollar strong state of the world so there is that there is not an extra moral hazard because the bailout is financed with reserves as opposed to taxation though I guess you could you could you could have something along those lines as well on the look the ideal liquidity regulation would be to control the dollar funding mix okay now what's important is that that dollar funding mix in reality even though the model kind of blurs this lives on the balance sheet mostly of non-financial firms in other words the banks themselves in part because of bank regulation themselves are not too outright mismatched but they've passed that on to the non-financial firm so what you would need is the ability to basically see through and regulate those guys and sort of our operating assumption is that that ability is imperfect in part because there's lots of ways to regulatory arbitrage that and given that it's imperfect then there's this role for caring about the interest rate on the other question about sort of these other motives for holding reserves we haven't done a great job here you know we've thrown in a bunch of covariates which are some of the other variables that people have looked at you know in in this literature um and our result is pretty insensitive to those covariates but I don't feel all that great about that because of all these other these other issues I just think the data at least the data that we've been able to put together is so limited and we've you know not even taken at all of a a sort of crack at any kind of identification that I think your suggestion that we take the empirical part out I'm guessing an editor may well say that so you know it's more of a wish to have a little bit of evidence along these lines but but but very much limited yeah right well on that bombshell let's draw this to an end thank you very much let's show our appreciation for everyone who's there