 So, let's move on to the second paper for this session and this second paper deals with the issue of trust and the paper basically has a great introduction saying that financial markets and banks, well, they run on trust and they stumble in the absence of trust. And we have really seen crystal clear evidence of this in the last few months in the US, for example. The paper makes a convincing case that the take up of sovereign investment in foreign countries, well, it has a significant and also economically important link to the amount of trust placed in the residence relatively speaking in this country itself. And this holds at bank level also here that the paper argues that banks with branch networks which are geographically well diversified and also whose management teams are well diversified, well, they are less likely to suffer from trust bias in these financing decisions here. And the paper argues that the effect of stereotypes is persistent over times and stronger for less diversified banks. So the paper will be presented by Orkun Saka at City University of London together with Barry Eisengreen from Berkeley and both of them are dialing in remotely for this session. As to discussion, this will be Eleonora Alabrese from University of Warwick. So let me hand over to the presenters. Thanks. Great. Thank you very much, Thomas, for the nice introduction. And I wish I could be there, but I couldn't be this time because of the family commitments. And thanks for the organizers for allowing me to present the paper in online version. So I guess I have 20 minutes. I'm going to start quickly. So this is a paper on cultural stereotypes and it's a co-authored work with Barry Eisengreen next slide. So we know that cultural stereotypes are quite prevalent. And here on the right side of the slide, what you can see is a recent cover from the magazine Economist where the Prime Minister of the UK at the time was pictured as an Italian soldier, especially referring to the kind of mess that she may have created at the time because of the new budget that she had introduced. But of course, you can see all kind of symbols of the stereotyping in this very picture. It shows you that the cultural stereotypes, even in the age of the political correctness, they're well alive in the minds of people even at the sophisticated magazines like the Economist. So we know that they're historically determined and they change very slowly. But at the same time, one could say, are all stereotypes wrong? Is there any grain of truth in it? So that's another separate discussion. There might be certain fundamentals that may create, for instance, a certain perception of a certain nation by anyone else. So in this paper, I'm going to try to tell you how we are trying to separate biases from the fundamentals that will come later. But our main focus in terms of capturing the cultural stereotypes is going to be focusing on how nations perceive each other in terms of trustworthiness. So how one nation's citizens or residents perceive another nation's citizens or residents in terms of how trustable they are. Next slide. So we know that cultural closeness matters in financial markets. There's quite a bit of evidence by now, both within country evidence and also international evidence showing that, for instance, investors are underweight in culturally distant markets when they're investing and also within the same country, for instance, they are investing into such companies with whom they are sharing more cultural heritage. But of course, the tricky part is that is it really the higher trust that these people have? Or is it just that with these distant type of investment opportunities, are they just lacking information? So that's discouraging them from investing further. Next slide. So in this paper, we are going to try to dig into this cultural trust phenomenon by looking at the financial institutions. So for this purpose, we are going to be focusing on the Europe. And we think that this is an ideal laboratory. I think I don't have to say too much in this audience that we have the supranational supervision of banks in Europe. We have homogeneous regulatory treatment of the government bonds, which is going to be the focus of our paper in terms of investment opportunities. And what we also know from previous literature is that there's quite a bit of divergence in terms of how nations see each other, in terms of how trustworthy the residents are. So this has been shown by GUSA 12, for instance, by using Eurobrometer surveys in the past and many other following papers that I'll mention later too. So what we are doing is we are merging this unique EBA data set that I will detail later with this trust phenomenon both by using the Eurobrometer, but also I'll tell you later that we are also collecting our own data set to update these measures and also expand them compared to the original survey. So we are going to construct not only at country-level trust measures, but we are also going to dig into the bank-level trust measures by using the branch networks of the banks in Europe. And I'm going to show you what is the micro-foundation there because we're going to look at several channels or kind of hypothesize about several channels in which the banks can actually might be affected in terms of their culture because of their branch networks in different parts of Europe. So in our identification with the bank-level evidence, we are going to be focusing on banks that are headquartered in the same home country, in the same country, at the same point in time and we are going to compare them only with regards to their exposures towards the same target country sovereign. So this is going to be kind of tight in the sense that, for instance, any kind of shocks that will operate, for instance, in terms of some countries getting into a crisis, they are sovereign risk-rising. This is not going to be affecting our identification strategy. So what we are going to be taking is basically taking the previous literature from country-level to bank-level evidence and I'm going to show you why we can kind of reach to the bank-level evidence by showing a kind of micro-foundation for why culture is important with regards to bank branches. Next slide. So there's quite a bit of literature. I'm not going to delve into many of them, especially the first part is quite important. As I told you, there has been some examples in the recent literature showing about looking at the effect of trust in different dimensions. This has included, for instance, foreign direct investments, portfolio investments, even the decentralization decisions of multinational banks are influenced by the trust across countries, venture capital investments, and also like more lately we have seen that even the stock recommendations of ecutanals were affected by the trust of their home countries towards the country of the firm that they are analyzing. But all of this evidence is very much at the country level. So we're going to discuss a little bit later about why country-level evidence is kind of hard to trust on, no pun intended, but we are going to kind of try to kind of dig more and bring this bank-level evidence to for better identification. And then there are two more literatures that we contribute to, but I'm not going to discuss in detail those, especially the last one is quite, I think, familiar to most of you looking at the sovereign determines of the sovereign exposures in Europe. So next slide. So this is what identification strategy looks at the country level. When you have this, you know, home countries and target countries, of course, you are looking into country pair relationships. You can use in this setting home country fix effects and target country fix effects, and the trust is going to operate only within certain country pair, right? So you can include home country and target country fix effects, but you will also have concern for many omitted variables that may be operating between different countries, right? So in the bank level, in terms of, you know, taking this to the bank level, we have started thinking with my co-author about, you know, how we can kind of operationalize a bank level measure of trust. And then we started thinking about the branch networks of the European banks. Next slide. So in that sense, we think that especially when it comes to the sovereign debt market, the branch networks would be important, at least for three different mechanisms. So we thought that, you know, one mechanism would be that some of the sovereign debt investment decisions could be delegated to the branches, and the local branch will make the purchase, in which case the local branch culture is going to be important in terms of determining the overall portfolio of the bank. However, this seems to be a little bit of a, at least anecdotal evidence suggests that this may not be the main channel, also because of the fact that we are focusing on the exit entry decisions of banks into sovereign debt markets of a country, which means that these decisions are really important decisions and they're more likely to be centrally decided by the headquarters. The other two channels, one channel could be that, you know, the headquarters, when they are making their decisions of sovereign investments, they may be affected by the biased, culturally biased information that may be coming from various branches that they have in Europe. This might be one channel, but we cannot observe this, as we don't have this access to confidential reports of the, internal reports of the banking system. However, a third channel, which is, I think, a bit more intuitive and also we can observe the data for is the fact that branches also send employees to the headquarters, right? When you have, when you are a UK bank and you have branches in Spain, it's likely that you are going to start employing more Spanish employees, even in your headquarters in the UK, and if the decisions are made in the headquarters, these Spanish employees are going to impact the way that you decide on those investments. And we are, I'm going to show you evidence today for this channel as well. So next slide. So when we take the branches seriously, this is what the little, what the identification strategy is going to look like, just to give you an example for a single country, like UK, we are comparing, let's say, three banks, HSBC, RBS and Lloyds here. What we have is, for instance, HSBC, we know that it has branches, it has quite a bit of branch presence in France, as opposed to RBS, which is headquartered again in UK, but has branches not in France, but in Ireland, right? Five to 10% of its branches, I think, are in Ireland. So what happens in our setting is that the HSBC is going to have a slightly different culture than the RBS, because of the fact that its culture is going to be a combination of the cultures of France and UK. On the other hand, RBS is going to have a combined culture of UK and Ireland. So how these two banks will perceive Austria, the same target country when they are investing, is going to depend on this combined culture. So we are going to be comparing these two banks, basically, within the same country, with service, the same target country. And of course, you might be concerned that, you know, we are also comparing these banks with regards to their own countries, that also is something that we show is not really important. We drop, for instance, UK here, we drop all the exposures to France and Ireland for these two banks, and we only compare them to the countries where none of these banks have direct branch presence. So that kind of gives us an ideal situation to make a comparison, to kind of cut off all the informational links that you might be thinking of. Also, like another concern could be the indirect relationships. Maybe HSBC has better information because France is closer informationally to Austria or Ireland is closer to Austria, so RBS has better information. We also try to control for that by looking at these indirect relationships. Next slide. So this is the Eurobrometer data. I'm not going to look too much of it, or definition is going to be at the country level. It's the percentage of people in the home country expressing a lot of trust towards the people in the target country, and then I'll tell you how we kind of turn this into a bank level measure. Next slide. So about the EBA data sets, I think I don't have to talk too much. So this has been, this is, I think most of you are familiar with these. It has been almost 11 years of data collection, data disclosures from the EBA and also its predecessor, and we have collected all this data and it shows us in some disclosures sovereign breakdowns of the portfolio in terms of country by country breakdowns up to 200 countries. We only focus on the 30 countries that we can consistently find in each of these disclosures, and this gives us 199 banks in total located in 27 European countries in 11 years. This is kind of a b-annual data, so like we see almost in every six months. Next slide. So this is the just summary statistics, nothing to mention here, but we have also quite a lot of controls coming from different sources. Next slide. So this is going to be the empirical setting for the country level evidence. So it's going to be basically like a gravity regression. Country level trust bias is going to be turned into a country level trust that I have just shown you the matrix for is going to be turned into a bias by taking into account the effect of home country and exposure country or target country fix effects. So we just look at the residual from that regression first of all, and then we plug that residual into the regression that you see in the below where the main dependent variable is the sovereign exposure of a bank B headquartered in country H with regards to target country C at the point time T. And this is going to be a dummy variable as I said, we are mainly focusing on the entry exit decisions, but I'm going to show you the results with the continuous measure as well. And the country level bias is only going to change between home country and target country. Okay, we have bank time fix effects here. We have target country time fix effects, but of course we cannot include country pair fix effects because that's the main variable of interest here. So let's move to the next slide. So here what I'm doing is basically I'm looking at the standalone effect of the trust bias, which is positive as you can see in the first column. And then I'm comparing it to various structural informational measures or historical measures that are also operating at country pairs. And from the from the columns two to nine, what you're seeing is basically pairwise comparison of how they are doing with the trust bias variable. And you see that some of them are actually correlated with the trust bias variable. So that's why they are eating up the coefficient, but in all the settings, the coefficient seems to be none of these variables standalone is able to eat away the coefficient from the trust bias. And in the final column, what you're seeing is basically like a kitchen sink type of regression where I'm putting all the various controls or variables that might affect the contra pair relationships. And the trust still comes out quite significantly. But this kind of relationships or this kind of regressions that has been done in the previous literature is really tricky to interpret because most of these variables that I'm including, such as geographical distance, like sharing a colonial relationship, etc. These may affect trust, first of all, and also some of these, such as bank mergers or media coverage, they may be impacted by the trust across country. So there is all this, all types of relationships that are going on between the independent variables here, that makes it very hard to interpret the coefficient in this kind of relationship. So let's move on to the next slide. This is where we are going to kind of focus on the mechanism and why we are going to, I'm going to tell you why we are taking this branch presence or branch network more seriously. So here we have the headquarter employee list of a subset of banks in our sample and we can actually observe their nationality at various levels. So we know the positions of the people. So we are looking at the highest positions here and we are, this is just a cross-section, there is no time variation. We are trying to see if the branch presence in a country for a bank is able to predict the nationality composition of its employees in the headquarters of the bank. And if you can move to the next slide. This seems very much to be the case at least for the subset of banks that we have in the sample. It seems that when you have a branch in a certain country in Europe, also your headquarter employees are coming more likely to be from that country. Of course we are not arguing for causality here. It's just a simple correlation in a sense. It can be also reverse causality that's going on here but we are just showing this to show that this is consistent with our argument that one of the channels could be the employee channel. So let's move on to the next slide. And this is going to be our bank level analysis where we are basically using the weight, the amount of branches that each bank has for the multinational banks operating in more than one country. Their branch presence in a different country is going to be weight of that country and we are going to kind of take a weighted average of the perceptions of the host country and then create like a bank level variable for the trust bias. And the regression looks very similar on the dependent variable while on the right hand side now we have actually, we are able to include a kind of more flexible set of fixed effects both at the country pair levels but also we can saturate it with country pair time fixed effects. Let's move on to the next slide. And here you are seeing the main results for the bank level trust. It seems to be a similar coefficient to the previous variable at the country level and then you're also controlling for whether or not these banks have direct branch presence and how many branches they have in the target countries that we are comparing them in and that also doesn't seem to be very important in taking away the effect of the trust bias. And in terms of economic magnitude here one standard deviation increase in the trust bias increases the probability of investing in the target country by 14 percent and this is quite large why because the unconditional mean of our dependent variable the sovereign exposure is 58 percent in our sample. So this is quite large compared to that that sample mean. Let's move to the next slide. So this is basically how it looks like in case some of you are worried that our effects are coming mainly from the eurozone crisis which is in the early part of the of our sample it doesn't seem to be the case. What you are seeing in the middle is a gray region or green shaded region where our coefficients are smaller and that is exactly the episode where the disclosures EBA disclosures started depending on the FINRAP reports and FINRAP reports actually demand less granularity in terms of country breakdowns because it does a threshold. If you are exposed to a certain country below that threshold you don't have to disclose that country you can just put it in the other category and this is a proof that for better identification we need proper granularity in this data set and then as soon as the FINRAP disclosures have ended and then the previous setting kicked in again we see the the coefficients are again quite stable and high even in the in the last part of our sample period. Let's move on. This is the same picture but using the dependent variable as the continuous measure instead of the entry exit decisions. Let's move on. So then we are doing several cross sectional tests for instance if we drop all the domestic exposures this is what the result looks like so it doesn't look like the home bias is the the usual home bias in sovereign exposures is the reason why we are observing this effect. Let's move on. So when we have the when we when we drop all the home country exposures but also as I mentioned to you in the ratification strategy when we drop all those observations all those target countries where none of the banks that we are comparing has any branches right so this is a very tight way of identifying the effect we believe and if anything the coefficients are actually going up much larger than than the main results. So let's move on to the next slide and here as I told you we are also controlling the indirect linkages that the host countries might give to the to the banks such as for instance France giving an advantage to to to HSBC or Ireland giving an advantage to to RBS we service their exposure to Austria so we basically calculate these relationship variables exactly in the same way that we calculated the trust bias by using the the branches as as weights and then we look at for instance merger media political relationships and branch relationships etc and none of these seem to be really important in this setting trust bias seems to be the the driver even in the regressions when we put a lot of those. Let's move on to the next slide. You have one more minute and I'll ask you to wrap up thanks. Sure so here is an IV results where we kind of like thought okay is this really culture and how can we really make sure that you know there is there is an exogenous variation in that so we look at this six different measures of cultural differences across countries these are you know arguably historically determined some of them are genetically determined so they are they are good candidates for instruments and the instruments the the trust bias by using the six different dimensions and then estimate the effect on the sovereign exposure and see as you can see here the F stats on the first stage are quite high and then we see a much higher larger coefficient on the trust bias when we look at the second stage. Let's move on when we look at the eurozone banks only this this also you know doesn't change anything let's move on when we look at the weak countries in the in the Europe such as you know Greece Ireland Italy Portugal Spain this doesn't change the result let's move on. So when we also look at the set of banks that are only regulated by the single supervisor mechanism so just in case you know what we are capturing is not the effect of of the bank but also the maybe it's the effect of the local regulator that's kind of demanding the sovereign exposures of the bank to be in a certain way this doesn't you know this doesn't seem to be the story really here because now in this sample all the banks are being regulated by the same regulator so that again you know doesn't change the result if anything the coefficient is larger and then we can move on to the heterogeneity we can skip this and then maybe more importantly when when there is a crisis when we look at the you know eurozone crisis episode and maybe when we interacted the trust variable with the with the occurrence of a crisis in the target country this seems to be really important in the baseline you can see that for instance when me include when we calculate the crisis with bond spreads or CDS spreads in the baseline trust bias is still important but when it's interacted with the with the eurozone crisis it seems that the importance of trust is magnified with the occurrence of a crisis yeah let's move on and then in the working progress what we are doing is basically to collect our own data set as I said one of the good things about collecting this data set across 30 countries this is going to help us to include more banks and more target countries in our sample because the eurobarometer is on the consist of 15 by 15 metrics and then this is kind of the preliminary results what you are seeing here is the persistency of the cultural stereotypes so there are almost 26 years between the two surveys the one that we made and the eurobarometer but you can see this very high very positive strong correlation between the perceptions of trust across European nations when you compare this these two surveys and this is very promising I think this kind of shows that there is very strong persistency in the trust perceptions and then when we actually apply this enlarged survey to our to our banking sample we actually find that our results go through so there's also kind of points to the external volatile for results let's move on so to conclude we are kind of in this paper aiming to extend the economics finance literature on cultural stereotypes specifically on trust by proposing a tighter identification strategy and we have kind of show also why we can do that by looking at the mechanisms where the branches of the banks can affect the trust at the headquarters and of course in our setting the implication is that this is likely to be inefficient because there is no reason why a specific bank in a country compared to another bank in the same country should perceive the sovereign as more or less risky based on their you know trust perceptions so if this is the case especially in a setting where you know the contracts are not relational when a sovereign defaults they default on everyone in this kind of a setting is really kind of implying an inefficiency and one I think important implication is the diversity so in our setting if you have boards or if you have branch networks that are more diversified then actually the positive and negative biases in in different regions will actually cancel out each other so diversity is key if cultural stereotypes are driving investment decisions thank you very much yeah thanks Otto so let me hand over to Ildo Noa that will lead the discussion role thanks cool okay um thanks a lot for quite an interesting read um okay so let me be brief um okay so to summarize a little bit the paper so these are the objectives of the paper so the the paper wants to test whether the trust of the residents of a bank in target countries affects the sovereign the sovereign cross border exposure of multinational banks and it does so by looking and exploiting the the network infrastructure of of these banks and whether that affects the culture at bank quarters and by by doing this it uses this bank specific measure of culture that differ at the level of the headquarters and may in fact impact this exposure now in terms of key key results this this trust indeed impacts positively significantly and importantly the diversification of the investment of these multilateral banks multinational banks and when these um when the analysis is refined at the bank specific measure of bank expo of culture which again crucially differ at the bank headquarters so this confirms the results now in terms of magnitude as you mentioned um and this is mostly focused on entry and exit the magnitudes are quite large and in fact one standard division increase in the in this trust margin increases the chances that a bank invests in a target country by 14 percent which corresponds to a third of the diversification gap and i also find quite interesting the heterogeneity in the results um so um one big point that is indeed that the most diversified banks in terms of the countries in which they invest indeed rely these banks rely less on on trust and the flip side of this is that actually trust is particularly important for countries that are less often targeted and in this case a classic counter example would be Germany and and finally it is quite interesting for me to see that indeed trust the trust element is particularly influential for countries and timing of crisis and this is done by looking at this initial period between 2010 and 2015 now this paper i would say that um that the the magnitude per se are not particularly surprising but where the big contribution of this of this paper is actually the identification which i think it's quite neat and relies on these uh bank specific measures of trust and indeed without repeating too much but this uh this uses the level the the country um the geography of the branch networks of this bank and uh and whether this impacts the culture at the quarter level and indeed they show this relationship between the composition of broad banks which maps um the branch network and this allows not only to um to control for country confunders but also and crucially this country paid level of um of potential factors that may confound this relationship now on the comments so in terms of mechanism you rely on the fact that potentially this um the trust of these branch employees is transmitted to the airquarter via this common practice of um of promoting internally and in this case uh by so doing this paper makes a parallel with between cultural stereotypes and gender bias now the literature on gender bias does shows that the diversification in terms of gender does improve the performance of firms and also affects uh the bank lending policies now in this respect would be extremely interesting to see whether this is the case also for this cultural diversity and and furthermore as i said the paper documents that this um trust uh differential affects the portfolio composition of these banks and this generates in some inefficient portfolio allocation but to some extent we know this already as as this is quite similar to the own bias and in general we know that trust uh matters and and i would like perhaps to to read a little bit more why should we expect that in this context so uh this might not be um be the the case and furthermore um a little bit more on the implication of the papers so it would be nice to expand a little bit um on the implication of the paper and uh in this respect this relates to my um my first comment and so given that this might affect firms performances uh why are these banks not not trying to eliminate this cultural bias and should we perhaps impose some regulation in terms of diversification of this sovereign sovereign portfolio okay and finally some additional points but you mentioned it at the end a little bit so at the moment the paper uses time invariant um measure of trust um which is also not super recent and so it is very promising indeed and to think of some time budding and more recent uh measure of trust uh which would allow also to um to look at this time variation however persistent trust maybe and finally i do find extremely interesting this heterogeneity in um with respect of uh exposure to um to crisis and to this respect you focus on this period between 2010 and 2015 so i wonder would what would happen uh when you use the full sample instead and perhaps i would find it interesting to see whether uh in more recent times this relationship and this interaction kind of vanishes um or diminishes and and yes just uh one last point i think that um that although it's it's nice and relevant to look at the extensive margin it would be also nice to have a slightly better understanding of the intensive margin as well um but overall i find this uh this in the identification quite neat and the paper was extremely enjoyable to read so thank you very thanks a lot Eleanor so with this we move on to the q&a session we'll use the same modalities as last time so take all of the questions in a row and then we can hear answers after that uh so when you intervene here please state your name and affiliation and and in the web experts please use the the chat or ratio flag so i think the first question was over here and then Jacob you're next man good morning from my side nicoz grapsas from malta financial services authority one question to the author is uh regards the control variable so i understand that uh you focus on the number of branches but as we see banks moving to a digital uh business line and they try to reduce the number of branches uh in their business model perhaps we should consider changing the control variable from with a variable that would be more on the retail side of business so when we have let's say a big number of branches in a bank we intuitively think that this bank wants to have a retail presence in this country so perhaps digital banks that still need to have retail business in these countries might be explained better in their behavior with this business line analysis rather than the number of branches thank you thanks and then we have Jacob and we have another person there thanks jake from the eba so interesting use of the data i i must admit i never really thought of this as a trust problem but let me let me try and challenge you on the control variables i guess right uh currency i don't know if that's in the soup but some of the banks you you or you have have i think business in a number of different currencies and as the liquidity risk management of the bank would naturally predict that you would want bonds in the same currency that you could repo quickly either with the central bank or the market to get that particular currency in that flavor right so i think that's that's definitely an issue in terms of how you operate right the other the other point i guess is um and i'm in frankfurter i can say this right i think the monetary policy operations and how you interact as a bank with with the central bank also varies across europe typically expected to interact with your local bank right local central bank at least in certain parts so there's also an aspect there right that you need to be conscious of the cost of having a number of different ways to turn your your bonds into cash right so for me some of your findings pretty much match what i would call normal liquidity risk management right in terms of how you would operate as a bank including in particular some of your crisis findings right it shouldn't be a surprise that if you have more volatile pricing of certain jurisdictions government bonds that you would try to avoid those for the core liquidity part of your portfolio whereas you could take them on to make a little bit extra money in times of peace right but i think in times of war i can use that term then you don't want bonds that have a huge variation in price right so i think there's also that point and then there are obstacles in terms of what you're expected to do in a given country if you are running let's say your business in that country typically there will be either direct or indirect pressure from the fiscal authorities to hold your local bonds right so i think there's a little bit of an issue in terms of leaving that out of the narrative right i i still think you're right there is a home bias there's probably a trust issue no doubt but they're also say less less high-minded explanations for what's going on here that i think you should just try to control for to see how much is actually left for trust to explain what we see thanks we had a question down there thanks tostenbeck ui so this might maybe reflect a bit that i didn't pay enough attention or my lack of clarity on this but so you use the the number of branches but i mean as somebody before said actually the gentleman in front of me i mean branches are for retail um and it wouldn't be possible to rather use like assets to kind of reflect the importance of what is probably primarily subsidiaries right because that's still a dominating model in the EU and apart from the Nordic region maybe so wouldn't that be a better measure and again you might have done this anyway maybe i just misunderstood it thanks and we have a question there um um yes so to be the same point as as jack up was was making actually that i i believe the results i believe that there is something about culture but the interpretation in terms of trust and and credit risk seems i know i i've been a bit hard to to believe i was wondering you know these are these are not for most bonds these are not centralized markets so couldn't there be a simpler explanation that you know you are ben pe paribas you do business in italy or in germany because of this you have many customers in those countries you have relationships with banks in those countries so naturally you find it easy to buy or sell bonds of those countries simply because you have the right trading relationships and of course there is a reason why you are in those countries in the first place because you chose to be present there and this might have to do with with cultural things so it's just another way of looking at your results but this sounds a bit more plausible to me ex-anteed and just you know do i trust more the germans or the italians because of cultural stereotypes yeah i think that will be the last question and ask do we have anything on the webex so no okay hi um it's a manuela ben incasa from university of zürich so i have a question um how does your story differentiate with the respect to the taste-based discrimination or the statistical-based discrimination because i assume that you you are assuming that i don't know banks act irrationally and uh does your stereotype which i implies lack of trust how does differentiate with respect to the literature on the taste-based and the statistical discrimination thank you maybe i take advantage also of the last question and it's picking up a bit up on what jacob was saying also that there are a number of dimensions and and and drivers here i think you have you have an area in in the paper where you're sort of controlling for the ssm also and it's true that we are of course are the supervisor of the the headquarters but also the all of the solo subsidiaries the 1200 or something like that in the system right now where uh it's also true that we have done a lot of work on that the difference between the subsidiary the branches all of the drivers for that and then i think it's true that yes we are responsible for the prudential rules in the system itself but there are a number of things uh which we are not responsible for at at european level itself and when we looked into that uh i think you've been looking into this Giovanni also in in great details where we realize that a number of other things which are important it could be conduct rules it can be market making in the local government bonds also so i very much agree with jacob that that it's important that these things are controlled for also difficult to do that but i think this is an important part of the exercise so with this let me hand over to you okin for possible responses to all of these good questions thanks uh thank you very much to everyone uh first of all to to my discussion Eleonora so very good points Eleonora thank you very much so just to respond briefly to um to your suggestions so performance aspect is definitely something that we have in our minds similar to as you said to gender literature where you know the gender diversity is associated with better performance uh here i think the trick is that how are we going to calculate the specific component of that performance that relates to the to the sovereign bonds which is our kind of focus in terms of investments we can look at the general performance of the bank kind of in a kind of at the bank level bank time level uh i'm not sure if that would you know directly be be a proof for the specific investment that we have in the paper uh to be able to calculate the uh the performance on the on the sovereign bonds only then we need to have like exact timings of uh selling and buying and what at what prices the selling and buying happened which is i think a bit trickier than the general performance measures but this is definitely something we are thinking about um is it similar to home bias it is not similar to home bias so i have uh also contributed to that literature in the past and there are some people in the in the audience who also have done the same uh home bias have many reasons home bias is not something that we can just like package in under one umbrella and then say that you know this is just home bias there are various reasons rational irrational and some of the political economy reasons uh behind it uh so this is distinctively different from home bias especially because of the fact that we are focusing even like letting go of the home exposures and just focusing on how these banks are doing in terms of the area you know foreign exposures uh i think it's uh quite distinct in that sense um why banks not eliminate these biases well it's it's a very good question my first pet would be that they're not aware of uh these biases themselves uh i mean of course you we can speculate further on this uh i don't i wouldn't be able to like come up with a definitive answer but i would say like this is uh something just like you know in the literature as it has been shown why are acutal analysts not aware of their biases when they're judging stocks etc i think it's just unawareness at this stage that would be my first best first pet time variation interest would be nice and that's that's the direction that we are taking so we have undertaken a large survey in 2022 we are kind of incorporating that into into our setting now uh that's something we can do or extensive margin on the extensive margin you said uh that's essentially not the case we have results also in the paper with lots of robustness checks on the intensive margin on the continuous measure uh it's it seems to be you know working quite well and actually i have shown the graph over time where the you can see the continuous measure that that works um and then why 2010 2020 sorry not why 2010 15 for crisis we've wanted to focus on the kind of crisis episode uh looking around 2012 which kind of stopped the crisis pressures in eurozone with the Draghi's famous announcement etc so we want to kind of you know look into a period which is uh which is kind of uniquely characterized as a crisis episode but we can you can extend the sample and then of course here one thing to remember is that it's not just we are focusing on that on that on the sub-sample of periods but we are also interacting with the occurrence of a crisis in the target country and there are some countries there as a control group which are not uh facing a crisis right so the interaction that's that's that's what we are focusing and if you do that with the full sample you see the similar results in the coefficient so we just wanted to make sure that we are focused on the crisis episode and then uh from some of the questions if i have time uh branches versus assets i think this came up a few times in the discussion so this is i think a reasonable um uh concern especially with the digital transformation we could maybe apply a asset-based measure of you know uh of uh presence for these banks across europe so that could be one way of going forward with it as a control for branches i mean branches were the most direct intuitive control that we we taught in mind but again we could we could do more on that maybe by adding additional controls but however i think with the with the additional analysis now we are adding onto the paper especially with the iv results that they're looking at this historical uh cultural distance across countries kind of predicting uh the the trust and then trust predicting this over and uh that decisions i think that's kind of you know in a way mitigating these concerns that you know there might be omitted variables here and then in terms of i think there there have been some concerns in terms of uh identification here like some people raised uh fiscal uh different fiscal policy measures of different countries etc i think uh maybe i haven't done a good enough job in this short presentation to explain to you that you know we are really focusing on the same countries banks uh with a with you know facing another same target country uh in which case you know this country level differences should not really play a role but if you really think that you know there are still differences within the same country pair that will play a role that can be the case but yeah i i think it's probably because of the you know me not being able to uh explain identification quite well um and taste-based versus statistical discrimination uh i think here we are arguing for a taste-based one we don't uh imply a kind of learning and taste-based coming directly from the from the you know cultural upbringing of the of the employees of these banks who are making these decisions um that kind of directly affects the the decision making so i think that's it from my notes if i missed anything happy to respond okay thanks a lot of walking thanks Leonore for very good presentations and discussions um so that that closes the session three