 So, good morning everyone, my name is Thomas Blasopoulos, I'm one of the two deputy directors general in our market operations area, and it's a pleasure to also welcome you from my side here at the ECB, it's a pleasure to have this conference again in person. So this morning's session focuses, first session focuses on collateral, and obviously this is an issue that we cannot overstate in terms of its importance for financial markets in the current juncture. We already heard during the Q&A session with Mr. Panetta that there is interest in the question of collateral scarcity and central bank policies that might help to alleviate that. And the first paper in our session looks exactly at this question, it looks at central bank facilities to lend securities back to the market to alleviate scarcity and assesses quantitatively the effectiveness of these policies. So, Stefan Grepmeier and together with his co-author, Stefan Yang from the Bundesbank, have prepared this paper that Stefan Grepmeier will present here today. And the second paper on our session on collateral looks at the role of collateral in another setting still within the context of central bank policies, namely the role of collateral in banking, in backing central bank lending operations, a more traditional role, if you like, for collateral in central banking. And on this topic, Pia Hütl and her co-author, Matthias Kaldorf examined the impact that an important change in collateral eligibility rules, which in fact harmonized the collateral eligibility across euro system national central banks, the impact that this changed had on bank lending behavior. And importantly, this paper traces out also the real effects of this policy change all the way through to the investment and employment decisions of the ultimate borrowers. So, these are the two papers for this morning sessions. We start, as I said, with the paper on securities lending. And Stefan, the floor is yours. The way we will structure the discussions in this session is we will have around 25 minutes for the presentation of the paper. Then Jean David has kindly agreed to discuss the paper for around 10 minutes. And then we'll have 10 to 15 minutes. And then we'll have around 10 minutes left for Q&A from the floor and from the remote participants. So, Stefan, the floor is yours. Thank you. Just a pointer. Yes. Okay. So, yeah, thank you for the kind introduction. And the need summary of the paper. I hope I can add some more on this. Thanks to the organizers for including our paper on the program. That doesn't work. It seems like the battery is empty. All right. At least. Okay. No. Okay. Let's hope that it survives until the end of the presentation. All right. So, this is a paper called Securities Lender of Last Resort on the causal effect of central banks securities lending facilities. It's a joint work with Stefan Janck and Colin from the Deutsche Bundesbank. And since we're both from the Bundesbank, the usual disclaimer applies. These are our views and not necessarily the ones from the Bundesbank or the euro system. So, this paper would have probably not been written without quantitative easing. Since quantitative easing was responsible for making central banks one of, if not the largest, single owner of government bonds. And while the large-scale asset purchases that we've seen surely help to compress yields on long-term bonds. The fact that all those assets then sit on the central bank's balance sheet means that they're missing in other places. For example, they're missing in the repo market where those high-quality assets are often used as collateral. So, quantitative easing in becoming one of the main drivers of collateral scarcity had negative side effects on the functioning of repo markets and also second-order effects on the functioning of secondary bond markets because market makers in those secondary bond markets rely on well-functioning repo markets in order to source collateral or funding when they want to intermediate in the secondary bond market. So, in response to this and in order to alleviate those negative side effects, major central banks in the developed economies, so the Fed, the Bank of England, the Bank of Japan Euro system, all established securities lending facilities as kind of a backstop. So, it's mainly primary dealers that can access those facilities when they're looking for a scarce bond or a scarce asset. They cannot locate in the private market and then they can borrow the bond from the central bank and use it in order to support market functioning. So, in a nutshell what happened is major central banks in establishing those facilities became something like the securities lender of last resort and what we're going to do in our papers we're going to look at the Euro system as one of those securities lenders of last resort and we're going to look at a policy change in the securities lending program in our case that's going to be a pricing change, a change in the pricing conditions of the facilities and we're going to use this change in order to understand how the change is transmitted to repo and secondary bond markets and whether the program works in line with the stated goal and the stated goal of those programs and this is a quote that's directly coming from the ECB's home page. So, the aim of securities lending is to support bond and repo market liquidity without unduly curtailing normal repo market activity and this quote already gives us kind of like a nice neat guideline which dimensions we need to consider in order to answer whether the programs work effectively, whether the transmission works or whether not. So, what we're going to look at are three dimensions, we're going to look at whether the policy change had any effect on the utilization of the securities lending facilities themselves, then we're going to move one step further and we're going to see whether the change in the securities lending facilities utilizations also had an effect on market participants normal repo market activity and here we're going to focus on like negative consequences so did it hurt normal repo market activity or not and then in the last step we're going to examine whether the securities lending facility helps to support bond and repo market liquidity. Now there are already some studies out there on securities lending facilities, there are not too many because those securities lending facilities are a rather novel policy tool and what these studies usually find is that that higher usage of those facilities is associated with lower scarcity in the repo market so there's one study by Fleming and co-authors for the US case some time already ago and there are two more recent studies, one is from colleagues from the Bundesbank the other one is from colleagues from the Dutch national bank who look at the securities lending facilities in the euro area. Apart from that there's another study by Pellitzoning co-authors that showed that due to the lower scarcity that we have thanks to the securities lending facilities that also helps to improve treasury market quality by lowering limits to arbitrage and by allowing arbitrage shares to have better funding conditions or collateral conditions. Now what's the challenge that's common to all studies that look at securities lending facilities is that the use of those facilities is usually endogenous so while the scarcity of an asset can determine whether I use the facilitator or not and to what degree I use it the use of the facility itself can then also affect the scarcity of the asset so there's kind of this reverse causality problem we have here and we need to solve that and our approach in this paper of how to circumvent this problem is that we're going to exploit a pricing change to those securities lending facilities of the euro system we're going to use it as a national experiment to really get to the causal effects of how those facilities affect repo and cash markets and the second thing that's distinguishing our paper from the literatures that we're going to use detailed transaction level data from the repo market activity of major banks so we can directly see how much a bank borrows securities from the euro system and then we can also observe what happens in terms of securities borrowing and securities lending behavior of those banks thereafter so we're really trying to to track the whole process we're you know trying to track the transmission of the policy chain policy change throughout the repo throughout the repo market and the cash market so before I go into details here let me give you some some quick institution institutional background on the securities lending facilities of the euro system so those are implemented in a decentralized fashion which means that each national central bank sets its own rules related to counterparty eligibility collateral eligibility haircut schedule this is mainly done to reflect differences in domestic market practices since repo markets are are to a large part over the counter what's common to all operations is that securities lending from the euro system of the euro system can either take place against securities collateral so one security is exchanged for another or it can take place against cash collateral these are the two options and what's also common to all the activities that they're subject to individual counterparty and the global limit in order to to contain the risks um the most important point here is however um although the securities lending facilities are organized in a decentralized fashion um there's an overarching pricing framework that should ensure the backstop character of those facilities and the pricing framework um is is exactly the one policy element that we're going to look at um because that one changed um in november on november 2nd 2020 um so so we highlighted the change in red here these are again the two options that we have so borrowing against securities collateral prior to november uh second meant that um you can you can do that against the fixed minimum fee of 10 basis points after or at on november 2nd this was reduced to five basis points so 50 reduction um the cash collateral option was also affected so prior to november 2nd um you could borrow from the euro system at a rate of the deposit facility minus 30 basis points and the spread of 30 basis points was then reduced to 20 basis points after november 2nd it's also a sizable reduction so in a in a nutshell the the pricing the change in the pricing uh schedule we see here made the use of the securities lending facilities of the euro system cheaper uh it moved it moved it closer to market conditions so what we're kind of expecting is that after the pricing change there should be there should be more utilization of those facilities and we can immediately or we can directly check that with with data that's available um publicly on the website um so what you can see here is the the daily amount of securities that are lent from or borrowed from the euro system um in the years around the pricing change so the red line don't know if uh in the middle of the graph that's the pricing change um the black line is the aggregate amount of securities that are on loan um the blue and the green line are securities collateral and cash collateral respectively and you can see that prior to november 2020 um the the daily balance um hoovered around uh 30 to 35 billion after the pricing change volumes immediately went up they sharply increased um and one year after the pricing change um securities borrowing amounted to um an amount of roughly 75 to 80 billion and this amount keeps on increasing so the the use of those facilities is is still a very common tool nowadays now in order to to tease out the causal effects of the pricing change um what we're going to do is we're going to have a difference in difference approach um using the pricing change as a as a shock as a supply shock a positive one and then we're going to um have a treatment variable and here the idea is that um we're going to argue that securities that are borrowed um are heterogeneously affected by the central bank induced collateral supply shock so what we're going to do is we're so what we're going to do is we're going to um split the universe of government bonds into bonds with elastic repo market supply and inelastic repo market supply and the way we're going to do is is we're going to look at the investor base of each bond and we're going to um argue that some investors are more likely to lend their bonds whereas other investors are less likely to make their bonds available for lending in particular they're less likely to condition their lending decision on on prevailing market conditions and the investors that are that are less likely to do that are for example households non-financial corporations um governments uh or insurance companies and pension funds so if there's if there's a lot of um holdings of the bond going if if a lot of uh one bond is held by those investors then this bond is likely going to have a low supply and private repo market which means that market participants in need of those scarce securities are more likely to borrow them from the euro system per se and even more likely to borrow them after the pricing change when when the use of the facility has become cheaper so we're going to construct a continuous treatment variable which is the share of inelastic investor in each bond and we're going to do that based on on detailed ownership data and we can also we can also back this claim um in the data um if we plot securities lending volume by different counter parties then we see that the the the investors we term as inelastic are are indeed the ones that have very low securities lending volumes in the repo market all right um in terms of hypotheses that we're going to test um as I said um our starting point is the policy change so um what we have are cheaper borrowing conditions at the euro system securities lending facility then in the first step we're going to um examine how the use of those um facilities change around the pricing change and the expectation is that um when conditions become cheaper then there's a higher use um or a higher securities borrowing in particular for securities that have that have an otherwise inelastic supply to the repo market we then move on to overall repo market activity and here we have two competing views um the first one is the substitution hypothesis which means that um if if banks borrow more from the euro system then they might borrow less from from private market participants so there's some kind of crowding out taking place which means that if those two effects cancel out then then we should see no effect on overall collateral availability unless borrowing from the market um under the second hypothesis which is the collateral multiply hypothesis um things look a bit different here the higher use of the securities lending facility translates into a higher overall collateral availability because market participants are going to use the collateral they borrow from the central bank in further repo transaction they're going to reuse the collateral which means that overall collateral supply should go up in the last step then we're going to look at effects on the repo and the bond market and here the two hypotheses um are again at play so under the first one that's related to the substitution hypothesis if there's no effect on overall collateral availability then we should not see an effect on overall repo market scarcity and bond market liquidity um if um the collateral multiplier hypothesis is at work so an increase in overall collateral availability then we should see less repo market scarcity and and a higher bond market liquidity so improve conditions in both markets um the data we use um the main two data sets are uh two data sets from the euro system so the first one is the the money market statistical reporting data set which contains uh transactional level data of uh secured money market transactions of the 47 largest euro area banks um we observe both the amount of securities those banks borrow and the amount of securities those banks lend from the euro system from other market participants so from other banks that report to the data set but but also from from further counter parties that uh that don't need to report to the data set and the second um data we're using is the securities holding statistics um and we're going to use it to construct our main treatment variable um this data gives us um a sectoral decomposition of of the investor base of each bonds so on a sectoral level um at the at the quarterly basis we observe how much of the bond is held by banks how much is held by insurance companies and so on and so forth and as I uh argued before we're gonna split the investor universe into into two types of investors elastic investors that are likely to supply uh bonds to the repo market those are monetary financial institutions and investment funds and inelastic investors that are less likely to supply their holdings to the repo market households insurance companies pension funds governments and non-financial corporations and we kind of borrow this definition from the literature now if you if we then start very simple and uh we simply look at how the how the um amount of securities borrowed from the euro system changes according to whether the bond is held by a lot of inelastic investors or uh a lot of elastic investors so we're splitting the sample in in two parts um we again look at the time around the pricing change the black line is the securities borrowing volume of bonds that have an inelastic investor base the blue line has an elastic investor base and we can see that the increase we observe in the public figures is entirely driven um by securities borrowing of bonds with an inelastic investor base so those bonds where private repo market supplies low those bonds after the pricing change are borrowed to a much higher extent from the euro system we then do a formal test um of the graph so we we set up a difference a difference model um with the dependent variable being the amount of securities borrowed from the euro system scaled by the amount outstanding of each bond to make things comparable um we first look at one year prior and one year after the change and we can see that um if we focus on the first column here um in general that's the post coefficient um the securities borrowing from the euro system increased after the pricing change and in particular it did so for bonds with an inelastic supply so that's the that's the interaction coefficient post times inelastic supply and where we have a highly significant positive coefficient we then saturate the model with time and bond fixed effects um the coefficient decreases a little bit but but stays highly significant and then we also we also shorten our event window a little bit um to eight weeks around the pricing change in order to have have very few confounding factors that might influence our results and here we also have a highly significant positive effect so there's more borrowing from the euro system um in terms of economic significance what do the coefficients mean for a one standard deviation increase in inelastic supply um we're going to have a 68 percent higher SLF utilization relative to the period prior to the pricing change um if we then go into the details of the um of the securities lending facility and look at the two different options of how to borrow securities against cash and securities collateral we see that the effect is roughly equally strong across those two options if we um split up borrowing in terms of the repo tenor market participants use we can see that borrowing mainly takes place at at term repos of one week or longer which is um something that we don't observe that frequently in the market itself um all right how much time we have left five minutes all right um if we then go to the to the overall repo market activity so here we have the same model um we just replace the amount borrowed from the euro system with the amount borrowed from the market um if we look at the overall amount borrowed from the market the first column we don't have a significant coefficient but it's it's it's positive um and if we dig a bit deeper um we can see some nice effects um the total amount borrowed increases which means that here this is evidence that is rather consistent with the collateral multiplier hypothesis market participants seem to use the collateral they borrow from the euro system in further repo market transaction such a total amount borrowed increases and we can even um quantify the collateral multiplier here since we since we scale all our uh variables by the amount outstanding of each bond and we can directly compare um the coefficient of what was borrowed from the euro system with the coefficient that's related to the total amount borrowed and the ratio kind of gives us the implied collateral multiplier um so for one additional unit of collateral that's borrowed from the euro system after the pricing change there are three additional units of collateral overall so the collateral multiplier here amounts to three if we then again split up repo market activity a bit more um into the tenors um we can see that in contrast to the borrowing from the euro system which mainly took place at at term repos of one week or longer in the market um borrowing mainly increases overnight so there seems to be some kind of maturity transformation along the repo chain market participants borrow term but then in subsequent transaction we see an increase in overnight borrowing um if we further then then check some more facts on the collateral multiplier hypothesis we can see that um that's the that's the first um three columns that uh banks do not only borrow more securities they also lend more securities which is kind of like a condition that needs to be met in order for the collateral multiplier to make sense and we can see if we consider both the reuse amount so that's collateral reuse divided by amount outstanding and the reuse intensity that's collateral reuse divided by the amount of securities borrowed um those two um variables also increase um in particular for for transactions that are centrally clear so what we have so far is that um central market participants borrow more from the central bank after the pricing change this translates into more collateral availability overall thanks to the collateral multiplier effect if we then look at effects on repo market scarcity which we measure by the by the specialist premium which is the repo rate minus the gc pooling rate um we expect to scar scarcity to decrease so specialist to decrease since we have more collateral in the system after the pricing change so here we have the same model different dependent variables that's the specialist spread um we look at it for different tenures overnight tomnext and spotnext for different data sets in order to check robustness and again we find effects that are consistent with our expectation we find a decrease in the specialness of bonds in particular bonds that have an inelastic investor base in terms of economic significance we find that after the pricing changes the one basis point reduction in the overnight specialist premium that's where most of the market borrowing takes place so that's where most of the effect on rates takes place and this one basis point decreases equivalent to a 13 decline relative to the period prior to the pricing change and last but not least we look at a bond market liquidity so given that market makers can can borrow more bonds and they can borrow the bonds more cheaply in the repo market to the extent that they pass on those lower costs to their customers we should observe a decrease in the bid-ask spread following the pricing change so we regressed the bid-ask spread on again the post dummy and the inelastic supply variable and we can find that there's a reduction in the bid-ask spread following the pricing change the reduction amounts to a point six basis points which is a five percent relative decline relative to the period prior to the pricing change so also kind of sizable all right so let me conclude with again looking at the quote that gave us the guideline the aim of securities lending is to support bond and repo market liquidity without curtailing normal repo market activity what are our insights from the pricing change and how the how the tool works or and whether it works effectively we can see that utilization of securities lending facilities sharply increased after the pricing change in particular for bonds that are otherwise hard to locate in the private repo market we do not find strong evidence or no evidence for substitution effect rather total securities borrowing and lending increases via the collateral multiplier effect which then leads to improved price improved conditions in the repo market lower scarcity and improved conditions in the treasury market through higher bond market liquidity thank you very much thank you very much all right i would like to thank the organizers for letting me discuss this very nice paper so i'm going to start with a very short summary of the paper and i'll talk about the big picture question yeah so what's the research question of this paper what is the impact of central banks programs of securities lending and what is the empirical strategy it's very nice they exploit heterogeneous exposures to an increase in central banks securities lending what do we find first they find that there is a positive supply shock meaning that we have a lower price lower spaciousness and higher volume of securities lending second main finding of the paper is that the change in policy decrease bid aspirates that is it increase liquidity in the underlying markets they find a lower bid aspirate and i'm going to focus the discussion on this second finding because i think this is the the future for this this paper what is the big picture question here and in the next slide i'll motivate why it's important to me is the following does securities lending have an impact or not on market liquidity and there are two ways to answer this one way is to think about a direct impact and the direct impact goes as follows market makers decrease bid aspirate when they can borrow easily the securities that's straightforward to test this is nice and second it's not trivial and i'm going to give you examples where you can even go the other way the fact that it's straightforward to test and not trivial is a promising avenue for the paper to me the second way to to think about this question is an indirect impact as follows market makers decrease bid aspirate when short sellers are able to incorporate negative information by borrowing easily the security it's a bit complicated to test and i think it's clearly true based on the literature so i would not focus on this okay so why should the paper address this question well it's clearly a key question monetary policy has large impacts on money markets in turn money markets impact securities lending if securities lending impact securities this is yet another mechanism where monetary policy can impact market liquidity so it's clearly an a key question second the authors have a better technology to answer this question than the literature so what the literature has used is quantitative easing as an impact to specialness so basically the central bank buys a lot of securities it makes it harder to to borrow those securities and specialness increase the problem is that quantitative easing also has as a direct impact on the outstanding amount of tradable bonds so it has an impact on how many bonds are available to trade in the market so quantitative easing has two impacts on specialness and on the outstanding volumes so with quantitative easing one cannot attribute the change in liquidity to specialness maybe the change in liquidity comes from the lower a tradable volume but the authors here have an exogenous shock or seemingly exogenous shock on securities lending only not on tradable volumes and that's a much better identification than the literature and finally this paper is encouraging first results of bid aspirate so why not go go for it further so i probably see you that show you that it's not trivial that specialness impacts liquidity and that makes it an interesting question imagine the following situation at t0 you have a mid price of 100 euros the bid price is 25 cents less and the ask price is 25 cents more and to facilitate the interpretation suppose the dealer's inventory at t0 is new so basically when a dealer has to sell a security what it's what she's going to do is to borrow the securities on the market and to pay the specialness so specialness is a cost on the ask side when the dealers buy the securities she puts it available for lending and earn the specialness so specialness is a revenue on the bid side suppose that at t is equal to one the specialness decreases by 10 cents what happens this is what is going to happen the bid price is going to decrease by 10 cents because the dealers charges the full decrease in revenue fine but the ask price is also going to decrease by 10 cents because the dealers is going to pass on the full decreasing cost in the end there will be no change in bid as spread only the mid price is going to change and that reflects something that Darrell Duffy in the in the 90s found is that specialness represent a stream of income to the holder of the bond and when this stream of income decreases the mid price the fundamental price of the security decreases so in principle securities lending should not it's not trivial that it impacts liquidity and so you need an asymmetry and sometimes you have the wrong asymmetry and it goes in the opposite direction imagine dealers have market power and one can find that securities lending an increase in securities lending will decrease liquidity with market power what happens on the side the same the dealers is going to charge the full decrease in revenue but on the ask price what's what happens is that the dealers would be the dealer would be reluctant in passing on the full decreasing cost she's going to pass only part of the decrease so the bid price will decrease by 10 cents but the ask price only saved by 5 cents and the bid ask price increases from 50 to 55 cents liquidity decreased so for that you need the right kind of asymmetry I don't recommend this kind of model but for example if you support short sellers instead of market power then you will find this because for short sellers short sellers are going to borrow and pay borrow the security on the on the market and pay the specialness and then sell it to the dealer on the bid side and if short sellers have market power they're going to pass on only the decreasing cost the decreasing specialness to the dealer and then you're going to find a decrease in in in the bid ask spread an increase in liquidity okay so basically I don't recommend the the authors to go for this kind of models they're too simplistic but I would recommend the authors to think about more more about the conceptual frameworks that link that would link specialness okay securities landing to market liquidity I'm going to be short and just talk about one other remark the author should spell out the shock a bit better here's the policy change um here I give an example for securities lending against collateral so before the policy change the lending thing was the maximum between 10 basis points and the market fee okay after the policy change it was the maximum between five basis points and the market fee but it's unclear that the policy change decreased the cost of borrowing imagine the market fee is 15 basis points then it's 15 basis point before the policy change and 15 basis point after the policy change and that leads to a conundrum for the policy to result in a price shock the market fee should be sometimes less than five or 10 basis points but if the market fee is less than five basis points why not borrowing from the market in state a potential solution to this conundrum is market breakdown so these people borrow from the central bank only when the market breaks down okay but is that the case during the policy change or adverse selection the the entities that borrow from the central banks are the entities that that can only borrow at a very high market fee and this should be better explained and exploited if there is adverse selection this is very interesting and you could use it the other remark is more a wish list so what would be nice to have is to test the policy against security for the security against security transactions because you use the mmsr and that would be security against cash transaction and what you could use is data from market is a bit expensive so let's see if you have budget but that would be a nice to have okay so let me conclude it's a nice paper with this identification I would suggest the paper to refocus or to devote a large part of the paper to answer a key question does visionless have an impact a direct impact on liquidity the authors have an advantage to answer it compared to the literature so we should go for it there are some efforts needed to develop the conceptual framework that links spatialness to market liquidity and to explain and exploit better the the conditions that lead an entity to borrow securities from the central bank instead of the market and I look forward to the next version thank you very much Jean David so we're basically right on time which means that we we can also take some some questions from the audience and then perhaps Stefan you can answer also including any reactions you may have to the Jean David's comments so the floor is open also for questions from the audience also from remote participants yes please the the microphone will be brought to you yes thank you very interesting paper I understand you use the mssr the mmsr database can you see there a difference also in the database itself is it all specials and single names or is there also data on on GC trades in there and also in the in the usage of the lending facility can you see the difference between whether or not it's being used only particularly for bonds that were special and had a pricing interest to be used in the lending facility or was there also usage of GC for providing liquidity to the markets in terms of collateral scarcity yeah so so there is no like indicator in the MMSR that tells us what is a special trade and what is a GC trade however given that you know there's this we're working in an environment of huge excess liquidity there are estimates out that that say that around 90% of all MMSR transaction are related to special repos which means that we're kind of leaving the GC side apart and assume that the main motive for an foreign market participants also to to get in touch with the euro system is is related to collateral motives and not funding motives so we cannot directly distinguish it however I think it's a reasonable assumption to say that that the overwhelming majority of the trades we we see in either cases related to special repos Andrew please thanks um if I understand your paper correctly it's sort of framed in terms of deltas you know what happens if a change here affects exchange there but um many of our markets at the moment have really high levels of gap between I don't know whether you want to do sort of bond swap spreads or whatever mantra you want to use so how can we use your paper of a tool to understand how to get closer to the optimal level as opposed to the delta which of course is the big policy question we're all facing or does the paper not really help with that thanks so I'm not sure if I if I fully understood the questions but the the so at least the results we're showing um in terms of specialness and so on it's on levels so we see that it's indeed the level that decreases for specialness um going closer to the optimal level would mean that we need a guess on what the optimal level is um we certainly don't don't provide that guess um I think what's what's what was our at the heart of our paper is really to to understand how you know twisting one of the policy parameters of such a novel policy tool um how that transmission then uh works throughout the reap when bond market um whether it helps us to get closer to an optimal level or not um it's not the main part of the of the research question I think it's more about the mechanics of how such facilities work thank you let's take one more question over here please thanks for the interesting presentation and the paper as you distinguish between elastic and inelastic investors I was wondering how you deal with uh foreign investors so outside the euro area given that for boons we know that quite a big share is held by uh foreign official institutions I think and they are quite inflexible as well but those are not included in shs to the extent that I know so yeah that's right so the foreign investors are not included although there are major investor group um in the in the government bond space um so um what what we do is um we kind of first of all follow follow prior literature um which puts foreign investors into the the group of elastic investors um um yeah so we treat foreign investors so in the shs we treat it as the residual between you know the amount we can attribute to each sector and the amount we cannot attribute this is the foreign sector for our definition of the variable the foreign sector goes into the elastic investor part um probably we basically also assuming then a large part of those foreign investors might be again monetary financial institution active in the repo market which would then be the elastic investors yeah so those are elastic elastic yeah there's a pertinent question also knowing that uh some sovereign wealth funds outside the euro area are big holders of of these securities and indeed also active in securities lending albeit episodically I believe there was also another question here more to the front yeah thank you I think it was um very much along the lines of the question that we were just discussing so maybe it fits well so I was wondering who are which bonds do those inelastic bond holders hold and do they have a preference and does that have a role for your result uh we would need to check that whether there's there's cross-sectional differences in which kind of bonds elastic and inelastic investors hold so ex-ante I cannot think of of one um but yeah that yeah might be something we need to check whether there's there's some systematic difference in in the bonds they hold um which might explain our results well taken so I I'm not sure we have any further questions but I'd like to give you also some time uh Stefan to get back to if you like if any reactions you you may have to Jean-David's comments sure um yeah so thanks for the thanks for the discussion uh very insightful I think that's uh the the um the part about the liquidity so we're still thinking about the the really big picture question that we want to address because you know leaving it like this might be a very specific topic um so that might be a good way to to to go for the future um we're still at the moment working on on more dimensions of dimensions of market quality to kind of leave up that part of the paper um we haven't really thought about the the micro linkages the direct and indirect impact of of liquidity so that kind of is a bit it needs it needs more work um so we we will surely think about the the your ideas um yeah apart from that um so I mean we do have a model in mind if when we talk about what links liquidity and specialness um that's uh one by uh in front and who um so that it's related to your thoughts but I think yours go a bit bit further in in bringing in this asymmetric impact so that's that's a really interesting thought um yeah so I think I'm I'm gonna leave it with that uh I definitely need to think much more about about the discussion before giving giving quick answers but yeah very much appreciated welcome okay and with that we're right on time so many thanks to both for being disciplined and for very interesting presentations and many thanks for for the questions and um yeah uh let me invite you all to to thank Stefan and Jean David thank you very much