 So we go to the second paper, which will be presented by Dmitry Cebotarev from INSIAG and it's the paradox of conservative haircuts. Thank you very much to the organizers for putting my paper on the program. I'm going to talk about the paradox of conservative haircuts and it also happens to be my job market paper and therefore all comments, questions and especially critiques are more than welcome. So this will be a paper about how the collateral requirements of central counterparties affect the selection of participants in the centrally cleared market. And since there hasn't been too big a discussion about central counterparties today, let me quickly introduce what that is. So CCPs are financial intermediaries, special financial intermediaries that exist to address the counterparty risk and financial markets. So imagine that there is a bilateral deal if one counterparty defaults, the other is of course affected. By contrast, if there is a centrally cleared transaction, the CCP arises and separates the initial transaction into two back-to-back deals, each done with the CCP itself. Then, as previously, one counterparty defaults, the other is not affected. Of course, provided that the CCP itself is resilient. Now, if we imagine several deals going through the CCP, then it will become apparent that the CCP concentrates risks and therefore becomes systemically important. And this is especially a crucial issue now when centralized clearing is spreading across contract types and geographically. So if we ask the question empirically, are CCPs financially stable, then we will see that the empirical evidence at least is a bit mixed. On the one hand, we have papers that show the examples when the CCP was acting as a shock absorber in a stressed market or helped to avoid fire sales. But we also have cases when the probability of default of the CCP was priced sufficiently high to be noticeable in the repo rates, or we have documented cases of default of CCPs. So it is pivotal to understand how well CCPs are protected by a very risk management system given their central position to many markets. And if we look at the risk management system of CCPs, it will turn out that they are fairly complicated. They have multi-layer structure. First of all, the individual collateral, that is, my collateral, which can be used only to cover losses inflicted by my own actions. By contrast, the guarantee fund, the second pillar of this risk management system, works as a risk mutualization device. My contribution to the guarantee fund can be used to cover the losses triggered by actions of other traders. And finally, there is the dedicated capital of the CCP, also known as skin in the game. Since the story will be chiefly about the first component, the individual collateral, it is important to know what is the common understanding of its role. Typically, it is assumed that the more collateral the CCP requires, the more incentive it gives to the market participants not to default and the higher the coverage is. That is, if the market participant defaults, the CCP liquidates the collateral and covers the loss. Therefore, the more collateral, the higher the CCP stability. The question that I'm going to ask here is whether this effect is unconditional, which means whether it is always the case that high collateral requirement benefit the CCP stability. And I will try to answer this question negatively. I'll try to show that high collateral requirements can in fact push the best traders out of the CCP market, which can affect the guarantee fund, which works as an insurance. Imagine if we push the best agents out of the insurance, the quality of the insurance deteriorates. But how can this happen? So imagine that there are two markets, the centrally cleared and the over the counter. And there is a set of agents which can choose where to trade. Also, suppose that there is some initial allocation, everyone to the right of the red line go to the OTC and all to the left go to the CCP. Now, one of the biggest differences between these two markets is that in the OTC market, everything is negotiable bilaterally. We can negotiate the collateral constraint, we can negotiate the collateral, how big the collateral requirements are. The price, the term, everything. In the CCP market, usually the collateral requirement is set on a uniform level for everyone. It is not negotiable. Imagine that the CCP exogenously decides to increase the collateral requirement. What will happen is that those who trade in the CCP market will be affected. For them, the trading costs will increase. Then some of them will probably want to switch to the OTC market and probably some of them actually will. So the question of the effect of this change in collateral requirement on the CCP market in my case will boil down to how different these green switches are from the blue who stayed in the market. In this paper, I'm going to argue that those who switch are the participants who have the highest credit quality. Therefore, after the change, the quality of the pool of traders in the CCP market deteriorates. To show this, I will use the data set from Moscow Exchange, which will be the data set on repo deals, both centrally cleared and over the counter. Using this data, I will establish three lines of results. First of all, I'll make a statement about how the credit quality of the borrowers in the repo market affect the position of this red line. More precisely, the position of the borrowers with respect to this red line. I will show that risky borrowers are more likely to borrow in the centrally cleared market and further that it is the lender's risk preferences that affect the borrowers allocation. The idea is that if you are a risk averse lender and you are faced by good and risky and safe borrowers and you observe their credit quality, you are more likely to trade bilaterally in the OTC market, that is to take the risk bilaterally with the good on the high quality borrowers. And you will give more incentive to the risky borrowers to go and buy an insurance to decrease the reducing credit risk, for example through trading through the CCP. Then the next result is that the higher CCP collateral requirement induce safer borrowers to trade over the counter. I will show that the quality of the blue plus green pool of traders is higher than the quality of just the blue in terms of the credit risk. And finally, a result that I will not devote time today, unfortunately, due to the time restrictions, is that there is a shade. There are different shades of this effect, which depends on how collateral constrained the borrowers are. Imagine that I'm a bank that have some stock of collateral and yesterday I already pledged it to the lender. Then if today the lender increases the collateral requirement, which in terms of this paper I will proxy by the haircut. If the haircut increases, it mechanically decreases the amount of money that I can borrow against the same fixed amount of collateral. By contrast, another bank that has some unencumbered collateral and reasonably low cost of pledging additional collateral to the lender will be much less affected by this change in collateral requirement. Therefore, the degree by which this uniform increase in collateral requirements by the CCP will affect the borrowers depends on the degree of collateral constraint, which of course will interact with this main effect. All right, the literature that I relate to is first of all of course literature and centralized clearing. Of course, this is the empirical literature on the stability of CCPs, which is quite thin due to the low availability of the data. And also to the theoretical literature, especially the latest ones on the design of the risk management system of CCPs. Since the paper is a selection paper, I'm talking about the selection of the borrowers between the CCP and the TC market. Of course, I relate to the literature on the endogenous market selection between the centralized market and the over the counter market. In contrast to the main part of literature, my papers first of all empirical and secondly has even for empirical papers has quite a peculiar data set in Russia in the in the ripple market. The CCP is not very picky, at least in the interbank segment with admitting traders to to the centralized clearing. That is, it thinks that if it is a bank, it is supervised by the central bank of Russia, therefore it is a safe member. By contrast in the Europe, at least in terms of repo, it is quite hard to get access to to the centralized clearing. This will allow me to sort of look at this selection effect without thinking of what is the role of being or not admitted due to high entrance barriers in the CCP market. The starting point for the paper is the regulator and practical literature. For example, the principles for financial market infrastructures, which discuss which has a discussion about the procyclicality and suggests that collateral requirements should be high throughout the cycle, not to increase them in the downturn of the market. And there is very little discussion typical in the regulator literature about what is the cost of high collateral requirements. So my paper is trying to suggest some some cost to this. And of course, since it is a paper on repo, I am related to the discussion on the stability of different segments of the ripple market. The data that I'm using comes mostly from the Moscow exchange. It is individual deal level data from January 2013 to July 2016. I see who is trading with whom, what is the haircut rate date, what collateral is what the loan amount. So basically everything I can match this with the data on the balance sheet of banks, which I get from the central bank of Russia. And I get the historical credit rating ratings by a web scrapping a practitioner's web page. In general, the data set is comprised of the OTC market and the centrally cleared market centrally cleared market is not uniform. It has two blocks by the bilateral CCP deals are very similar to the OTC. As an OTC market, first lender and the borrower communicate either on the phone or through the terminal, but then they decide to tick the box that the deal should be centrally cleared. Then it is the CCP who chooses the effective size of collateral requirement. By contrast, the exchange traded repo in the centrally cleared market is implemented as a limit order book where the collateral requirement is preset. And the interest rate is determined by just by supply and demand. The most of the data is coming is in the OTC market, which is due to to the timing of the sample. My sample starts when the CCP market is almost non-existent and then it takes it bites pieces from the OTC market throughout the sample. So if we compare unconditionally the OTC and CCP market, we will see that that the haircuts in the CCP market unconditionally. So it's over all borrower's lenders and collateral haircuts in the CCP markets are higher and the reparate is lower. The lenders and the borrower's booth in the OTC market are bigger. And most importantly for us, unconditionally again, the credit risk in the OTC market is lower than in the CCP. But what do I mean by saying credit risk? By that I mean the latest credit rating by Moody's S&P or Fitch converted to the Moody's Rating Scale to which I assign points from one being the safest to 13 or 14 being the most risky. Okay, so let's go to the results and first of all discuss how the collateral, how the credit quality of the borrower's effect where the borrower is going to trade in the CCP or OTC market. But if we think about it, how the effect, how if we think about how the credit risk affects the location of borrowers, it is not very clear exactly what we would expect. One way to look at it before we see the result is to think about it in terms of the degree of information asymmetry. If the lenders have little information about the quality of the borrower's, we can think about it in terms of Bester 85 paper where CCP is a costly signal. Indochinously, collateral is more costly for risky borrowers and therefore a separating equilibrium may exist in which safe borrowers separate themselves by offering more collateral. Since the CCP intuitively has some advantage at managing high amounts of collateral, there can exist an equilibrium in which CCP asks more collateral. And the best borrower separate themselves by taking this contract and therefore getting lower, lower report rate. By contrast, if there is not too much information asymmetry, for example, if ratings are sufficiently informative, one can think about the CCP playing the role of an insurance device or insurance company. If lenders are risk averse, then one would expect the insurance, which is the CCP, to be more useful for riskier borrowers. So the lenders will trade with the safe borrowers over the counter and give the incentive to the risky borrowers to go and insure themselves through the CCP. Then it is the risky borrowers that will turn out to borrow through the CCP. If we look at their empirical results, we will see that they are in fact consistent with the letter hypothesis, which means that the CCP plays the role of an insurance company. The higher the credit risk of the borrower, the more likely this borrower is to trade in the centrally cleared market. This is in line with the letter hypothesis, but also I show some evidence in the paper that this effect has something to do with the lenders risk preferences. First of all, this effect is strongest for the non-anonymous, for the bilateral market in which the lender have more information about the quality of the borrower. As practitioners explained to me, when the borrower and the lender contact, for example, on the phone, it is not clear yet where they are going to trade. And after discussing all the details of the deal, it may be that the lender will say, OK, we will do it, but please let's do it in the centrally cleared market. So there is this amplification of the selection effect. We will expect the selection between the OTC and bilateral CCP to be stronger than between OTC and exchange traded CCP. And this is exactly what we find. Another evidence suggesting that there is something to do with the lender's risk aversion is that the effect of credit rating, the effect of credit risk, it decays with the age of the credit rating. This is in line with understanding that the credit rating is a noisy signal about the borrower's credit quality. Then if the credit rating was released yesterday, it's much more informative about the state of my balance sheet than if it was released a year ago. This is in line with the lender looking at the credit rating and making a decision about whether to trade or not with this borrower over the counter or what rate and haircut, how to negotiate the rate and haircut with this particular borrower. To wrap up this part, I show that risky borrowers are more likely to borrow in the CCP repo market and that this in general, the story is in line with the CCP playing the role of an insurance company and the selection being influenced by the lender's risk preferences, by their risk aversion. Now, going to the main result, just to remind you, here I'm going to argue that the collateral requirements of the CCP affect the selection of borrowers. That is high collateral requirement of the CCP will push the best borrowers out of the centrally cleared market. How am I going to do this? For identification, I'm going to use the difference in haircuts between the centrally cleared and over the counter repo markets. The idea is very simple. CCP haircuts are collateral specific. How do I know this? Well, the CCP of the Moscow exchange has a methodology that is published on its webpage. And according to this methodology, it does not take into account the identity of the borrowers or lenders or anyone else. By contrast, the OTC market does not even have a unified methodology. So the average haircut in the OTC market, it is sort of a decentralized view of the market on what the collateral requirement at this point in time for this security should be. And of course, both haircuts in the CCP and OTC market are sensitive to the security specific events, but they react in different ways. CCP reacts in a predefined way, and the OTC doesn't. The OTC can react to different events or can react to the same events in a different way. Therefore, I'm using the difference between these haircuts as a measure of overreaction of the CCP to security specific news. I start by aggregating the credit risk and the haircut at the collateral month level in each market. And then I run a regression first starting the aggression of difference in credit risk between the two market on the difference in haircut between the two market, of course, with security and month effects. And first I find that the coefficient is positive and significant, which is in line so far with the idea that a high collateral requirement in the CCP market leading to a high credit risk in the CCP market. But we don't know yet whether this effect comes from the CCP haircut or OTC haircut. To notice, let us decompose the exogenous variable. By doing so, we see that the effect is indeed coming from the centrally cleared market. Further, we can support this by univariate regressions. Yet, the indogenous variable is still the difference in credit risk. We don't know whether it is coming, whether the effect is going through the credit risk in the CCP market or the OTC. This can be shown by decomposing the indogenous variable as well. And here we see that indeed when the haircut in the CCP market goes up, the credit risk in the CCP market goes up as well. By contrast, in the OTC market, the credit risk goes down in response to an increase in the CCP haircut, which is in line with less risky borrowers going from the CCP into the OTC market. Which decomposes the column 2 in the previous table. And of course, we can obtain the same results with the previous indogenous variable, the haircut difference. Here, if we think about the economic significance, it will turn out that it is quite modest. This coefficient, for example, says that it takes a difference in haircuts between the two markets of 20 percentage points to move the credit risk difference by one point. This is kind of a lot. But remember that in these regressions, this difference is composed of two parts, the CCP and the OTC. First of all, all the noise in the OTC haircuts that come from liberation were not very much interested in. It goes into this measurement and it can affect the estimate as the measurement error in the exogenous variable leading to an attenuation bias. So ideally, we would like to look at an experiment when the CCP moves the haircut exogenously and the OTC doesn't react to this. So that with the haircut difference moves only due to the action of the CCP. Not having this as an experiment, let's try to get to it as close as possible. So imagine that the OTC haircut is composed of two parts. One is common with the CCP haircut and the other is orthogonal. Then, of course, the haircut difference will also have the same structure. By regressing the haircut difference on the CCP haircut, we will identify the correlation, which turns out to be roughly 40%. Now, if we regress the credit risk in each of the markets and the credit risk difference as well, on the estimated haircut difference, that is, on this component, this will be, first of all, we will throw out the noise coming from the OTC market. For example, when the OTC haircut changed, but the CCP haircut did not move. And also, we will sort of get close to this idea that the CCP changed the haircut, but the OTC did not change the haircut. We're kind of undoing the reaction of the OTC market to this component present in the CCP haircut. By doing so, we will find that now the credit risk difference is more sensitive to the haircut difference. It takes roughly a 5 percentage point difference in haircut to move the credit risk difference by one point. Now, let me summarize these findings. First of all, the main takeaways are that there is the selection effect, that collateral requirements of the CCP induce the selection effect, and that it has something to do with collateral constraint. It is amplified by the collateral constraint. But the direction of this effect depends on the institutional characteristics, not only of the centrally cleared market, but also of the market into which the selection is going. For example, if in the OTC market the lenders were not that sensitive to the credit risk, then this effect would not be correlated with the credit risk. Then the safest borrowers would not be more flexible in switching the market. Okay, so every time we're thinking about the selection, we need to identify what is the alternative contract and to compare the institutional details of the tool. Can this effect be first order rather than second order? Because of course, that's what we started with. When the CCP increases the haircut, it gets more collateral. Therefore, it should be more stable. I would like so far at this point to leave it as an open question because this requires a model. And in fact, I have a model in my paper. It's in the appendix. It still has some things to be changed. This is a model of market selection where the contract in the OTC market is endogenous and in the CCP market is exogenous. In this model, I show an example of the situation when the CCP increases the collateral requirement. And at some point, the effect of selection becomes stronger than the first order effect of the direct effect of just having more collateral and therefore becoming more resilient. But it still has to be calibrated. And this of course has implications for the regulatory policy. So first of all, it provides some justification, some additional reasons for the mandatory centralized clearing of the standardized derivatives. And some motivation for the regulation of collateral requirements in the OTC market. But we need to know that this is unlikely to solve the issue completely because there is a very nice paper by Stefano Ungaro who analyzes the introduction of the CCP in France in the beginning of the 20th century. And shows that when the CCP collateral requirements were high and centralized clearing of repo with certain collateral was mandatory, it gave actually rise to another market which was legally different from repo but economically similar. Sorry Dmitri, may I ask you to wrap up because I don't want to dance in the discussant time, sorry. Okay, so basically that's it. Thank you very much for giving me the opportunity to present here and I'm looking forward to hearing the discussion. Thank you. Thank you very much. So the discussant is 1.1 from the BAS. Thanks again for having me discussing this interesting paper. So, my name is from the BIS. So the usual disclaimer applies. The views expressed here are all mine and not necessary of the BIS. So, yeah, so first of all, I'd like to congratulate Dmitri for this nice paper. I think you have a lot of interesting results and just and I think you did a great job in presenting the results as well. But just to give a high level summary, so the paper studies the impact of the uniform collateral haircuts of CCP in the on the on the clearing incentive. Dmitri has already showed that he has a very novel trade level data on interbank removal from Moscow exchange. And the paper shows that when CCP raised a haircut, it actually can push away the lowest followers. And this effect is particularly strong for those one with binding on the constraints. So in other words, the CCP was ever selected. And then, as Dmitri mentioned in the appendix, the author shows a theory model that such feedback could potentially impair CCP stability. So my comments will start from the literature. I think Dmitri has done a great job in sending a comprehensive literature. For me, I think the closest papers to this one is the ones on clearing incentives. So the selection between O2C or CCP when it comes to clearing. So on that one, I think Lorena and her co-authors has a very interesting paper on the clearing incentives in sovereign CVS market. They investigate the key factors which are first the liquidity and riskiness of the reference entity that would that would have impact on the capital requirement and the marginal requirement. And then second, the credit risk of the counterparty. So that is similar to what Dmitri is sending here. So whether the more risky counterparty would choose to clear the CCP while the safe ones would go for OTC. And then the third key factor is the clearing members portfolio net exposure with the CCP. So that is about the netting efficiency. So if the clearing number has higher netting efficiency when joining the CCP, that would push them into the CCP instead of the OTC. So related to that, as an owner and his co-author from CFTC also has a related paper on the impact of clear margin rules on clearing incentives in non-delicable forward. So basically the Basel committee requires that for eligible entities, if they are cleared bilaterally, then they have to impose collateral requirements as high as CCPs. So that is a way to incentivize central clearing. And the authors from that paper finds that these rules actually increase clearing rates for MDFs, but only for those ones who are already CCP members. So that shows that there are some frictions in becoming CCP members, at least for the MDF market. And then they also show that the netting benefits are the main driver for clearing DSCCPs. And then also Dimitri mentioned this paper by Angela Romado and his co-authors, where it shows that actually traders with lower counterparty status are more likely to clear DSCCPs. So I think it's very much an open and critical question about what kind of factors are behind the clearing decisions of the traders. And I think Dimitri's paper provides some interesting evidence on the ruble market in Russia. So then going back to the specific institutional background of this ruble report clearing, I think compared to the majority of the CCPs nowadays, there are two fundamental differences in this ruble report clearing. First is that the membership requirement of the CCP is rather loose. So everyone basically who trade in the ruble market can become a CCP member. So in that sense, the friction of becoming a member of CCP is rather low. So that is rather different from what the other CCPs in the major market nowadays have. And I will talk about that later. And then the second thing is that on clear margin rule, apparently in the data sample of Dimitri's paper, it's not implemented in ruble report clearing. That's why you will see that in this paper, the CCP haircut are higher than the bilateral ones. While generally speaking with the UMR implemented, the CCP haircut are generally lower than the bilateral ones. So yeah, so I think in Dimitri's paper, the CCP really is an insurance company offering a standardized contract. So in that scenario, indeed, only the risky traders would like to pay for the or would need to pay for the insurance premium. So only the risky ones would be required to go to the CCP and pay for this, the higher care cuts. But I think generalizing this result to other major CCPs nowadays could lead to misleading policy implications. So I would be happy here. First of all, yeah, as I have mentioned, the membership requirements nowadays is rather selective. And the collateral requirement actually can become a party dependent. So it's not necessary to be uniform distributed across parties. For instance, I think out both LCH and CME, they have credit add-ons for collateral requirements, meaning that, you know, if you are actually with a lower credit worthness, then it's possible that your collateral requirement is different from the others, especially in the OTC market. Also, I think you cited John and Lindsay's paper. Yeah, so basically, they also argue that, you know, CCPs are centralized monitors in a sense that they are really like monitoring the credit worthness of their membership. So in that sense, whether the uniform collateral requirement can be generalized, that is a big question. And second of all, yeah, as I mentioned, as in honor of his co-authors, I've shown that, you know, this unclear margin rule really helps to center clearing because it analyzes bilateral peer trades with higher collateral requirements. So I would say that I think you have very interesting findings. It's more about the exposition of the paper and, you know, about the potential policy implications from the paper. That probably needs some careful wording. And then, yeah, and because the clearing incentives in major markets nowadays could be very different from the results of the removal clearing. So then my second comment is about the refreshments. I think as Dimitri has presented, he has a lot of nice controls in the regressions. However, one thing that I find missing is the impact of netting. As I just mentioned, in this literature, we find that netting is the main driving factor for central clearing. And also, I mean, Duffy and Kaohsiung has a very nice paper showing that, you know, the tradeoff between multilateral netting, which is the key benefit of central clearing, and bilateral clearing, which is the key benefit of, and the multi-asset netting, which is the key benefit of bilateral clearing. So I think your data provides that opportunity to understand better that whether the increasing popularity of central clearing partly can be due to the benefit of multilateral netting. And also, you know, more importantly for your story, does netting really affect traders' clearing decisions in the ruble market? So controlling that probably would make your results more robust and stronger. And then my final comment is about the missing paradox. I think you have a very catchy title. But I think you need to show more results to convince the readers that the conservative tariff policy could threaten the stability of the CPP. I mean, that's what you have said in the introduction, but the current results are solely about clearing incentives, so the decisions are whether clear in the CPP or choose for bilateral clearing. So I think you have some theoretical arguments in the intro on the tradeoff between collateral and default funds. I think they sound appealing, but I have to be honest, I do not go into the appendix and check all the theoretical results there. But I have one key point or two points. I mean, one you have already been, you have already touched upon, that is that higher collateral itself has first-order effect in mitigating the counter-party credit risk of the traders that clear the CPP. So that is the first thing. The second thing is that, you know, from the industry practitioners, we hear a lot of arguments about they prefer higher collateral requirements, but a lower default fund requirement, because exactly because default fund is fungible. So if you are a safe trader, you prefer to go to a CPP where the CPP has high collateral requirement and low default fund requirement, so that the risk of your default fund contribution will be used by the others is lower. So in that sense, actually higher collateral requirements should encourage more safe traders into the CPP. So that layer of incentives, I think it's not articulated well in the intro. I think it's also not considered in the empirical results because in the empirical results, we are comparing just CPP versus the bilateral clear, right? So we are not comparing the CPP with higher collateral requirements, with CPP with lower collateral requirements, but higher default funds. And I think for that, yeah, that is a key thing to think about in terms of the theory. So, yeah, and for me in general, I think it's not really a part of that the risky traders would need to pay for a higher insurance premium. I see the value of the paper mainly on the clearing incentives and more on the clearing incentives. I also, yeah, while listening to your presentation, I was also wondering like, can you rule out the reverse costality from your data set and from your empirical identifications, because you are showing the difference of haircuts between the CPP and the bilateral clear OCC market, right? But then is it possible that the CPP charged the highest haircut exactly because of the migration of the risky followers? Just the example you just mentioned that, you know, like while you ring the bell of your counterparty, but they actually do not want to take your view with a bilateral clearance. So they said, okay, let's go to the CPP, right? So in your data or in your empirical identifications, can you clearly identify an extraordinary shock of increase of the CPP haircut that leads to the migration of the risky followers? And I think I, as far as I recall, I did not see that kind of evidence and that might be worthwhile to think about. And then I have some minor comments that I can. Sorry, may I ask you to skip it? I can communicate offline. So yeah, thanks a lot for the paper. I really enjoyed reading it. So thank you very much, Wendy. Thank you. So we're already over time, but of course I want to give Dimitri the possibility to reply quickly to some of the comments of the discussion. And actually then there is one question in the chat. So Dimitri. Thank you very much. It was a great discussion. I'm super grateful for all the comments. And also, yeah, also for the literature that you recommended that two out of three papers that you put, yes indeed, I overlooked and I need to have a look at them to see how much they have covered of what I say. Then let me start from the end because I think the most important comment was the last one about the reverse causality. Indeed, I don't identify one single shock. It's not in that terms. But you're absolutely right. First of all, that in the long run, the haircuts and the composition of borrowers, they are jointly determined. But on the collateral month level, it is very hard for the CCP to readjust it on collateral month level as a response to the changes in the credit risk of the borrowers because it just, first of all, its methodology doesn't give this flexibility. And even if it tried, this will completely antagonize the market participants. They don't like to have these changes in an unpredictable way. So this methodology is actually, it's basically reacting to the volatility. This haircut can go up on a daily basis, can go up today and in two days it can go down. And it cannot, mechanically it cannot react to the quality of participants, but it can in general the CCP can make the methodology more conservative, for example. But of course not on the collateral month basis. In this sense, I'm using the quote-unquote high frequency variation as plausibly exogenous to the actions of the CCP. Now, I totally agree about the comment on the difference in the structure and the rules in the market, in the major markets like in Europe and the US versus the market in Russia. The question is whether it is always going to be the same or whether there will be more voices for opening access to different market participants to the centralized clearing. Because as we have seen in a paper presenting a year ago in the same symposium, it can amplify the past rule of the monetary policy. I think Yimin was presenting this paper a year ago. So then of course we will have to also change probably the way these credit add-ons work because, well, these credit add-ons can actually amplify the pro-cycle locality on the individual level. I had a conversation with the market participants that were actually very much afraid of this introduction because imagine a bank becomes a bit riskier. And now, apart from all the consequences of this event, also now the collateral requirement in the CCP market is higher. So even the last resort the CCP market is punishing for a high credit risk. So actually participants were afraid of this. So it is not certain how this will change in the future. I think it's an interesting example that can tell us what will happen if the policy kind of changes that direction. Then I totally agree that I have to do more work on the missing, what you call the missing paradox, to really show that this is not the secondary effect. But think about it, even if it is the secondary effect it will change, it has to change the sensitivity that we think about. So then probably we need to increase, if we think about increasing a haircut, we know that there is an effect that goes the opposite way, probably we need to increase it a bit more. So even if it were the secondary effect we need to take it into account. But of course I have a lot of plans to do more with the model that I have. I think I'm a bit over time. I will respond to all other comments as well. I have a response. And I would like to ask you to send me the presentation and also probably to have a bit more of this discussion offline. Thank you once again very much for your comments. Thank you. Thank you Dimitri. So in the chat there is a question that you may have partly answered already, but okay, I'm reading it. So can I ask two clarifying questions. How often the CCP changed haircut levels? I guess my guess is that this is a question related to your data because I mean, of course we discuss about this that there may be a big differences. And how quickly does haircut change in response to changes in market conditions? The CCP haircut changes like literally during the day. If there is a spike in volatility it can go up within the day just in the middle of it. And then it will be on this level for some time and then it will go down. So it is quite responsive in the CCP market. Of course in the OTC market it takes time to a bit of time to rewatch it and so on. But also the OTC market is also responsive to this. So with this I would like to close the conference because we're already 10 minutes over time and it's late for some people. Thank you very much. I think it was a very, very interesting session. I learned a lot. Both papers were very interesting and that's a lot for the discussion to do the great job. And of course to all the panelists and to all the participants, the conference will take place also tomorrow afternoon. There is going to be two great sessions. We're going to talk about central bank reserves and payments and there will be also a panel of market participants that also is very interesting. So see you tomorrow. Have a great night. Have a good rest of the day. Okay. Thank you. Bye. Bye.